Merge pull request 'Qwen3.5-2B 参数相似度结果与免训练功能复用方案' (#1) from codex/2b-parameter-pilot into main
Reviewed-on: #1
This commit was merged in pull request #1.
This commit is contained in:
@@ -0,0 +1,5 @@
|
||||
# Scientific artifacts must retain the bytes recorded in checksums.json.
|
||||
results/** -text
|
||||
# Generated CSV uses RFC 4180 line endings; Matplotlib SVG paths contain spaces.
|
||||
results/**/*.csv -whitespace
|
||||
results/**/*.svg -whitespace
|
||||
@@ -0,0 +1,5 @@
|
||||
__pycache__/
|
||||
.venv/
|
||||
weights/
|
||||
*.partial
|
||||
*.log
|
||||
@@ -0,0 +1,26 @@
|
||||
# Model parameter similarity experiments
|
||||
|
||||
Bounded, inference-free analysis of official unquantized model weights. The first experiment
|
||||
is **Qwen3.5-2B**. Model weights are downloaded and analyzed on the authorized remote host,
|
||||
then deleted after archived results pass validation. No model weights are committed here.
|
||||
|
||||
- [Training-free functional reuse plan (proposed; not executed)](docs/TRAINING_FREE_FUNCTIONAL_PLAN.md)
|
||||
- [Primary-source literature and applicability](docs/FUNCTIONAL_REUSE_LITERATURE.md)
|
||||
- [Procedure and limits](docs/SOP.md)
|
||||
- [2B pilot findings and limitations](docs/PILOT_FINDINGS.md)
|
||||
- [Raw result report, data and figures](results/qwen35-2b/REPORT.md)
|
||||
- `pilot/download.py`: revision-pinned official checkpoint download and verification.
|
||||
- `pilot/tensors.py`: memory-mapped safetensors/BF16 reader and metrics.
|
||||
- `pilot/analyze.py`: bounded tensor/block search, exact verification, representative groups.
|
||||
- `pilot/plot.py`: scientific PNG/PDF/SVG figures and report from saved data.
|
||||
- `pilot/validate.py`: result/group validation, checksums and guarded weight cleanup.
|
||||
- `pilot/audit.py`: finite-metric checks, test evidence and archive provenance.
|
||||
- `pilot/run.py`: one-model supervisor with bounded download wait and analysis timeouts.
|
||||
- `tests/test_pilot.py`: synthetic numerical and 3D correctness tests.
|
||||
|
||||
The primary result is **verified net byte savings versus direct relative weight error**.
|
||||
This does not demonstrate preserved task capability or faster inference. Different block
|
||||
scales are alternative partitions and their savings must not be summed.
|
||||
|
||||
Dependencies are pinned in `requirements.txt`; install only prebuilt wheels on the remote host.
|
||||
Full resolved versions and run provenance accompany each result. See the SOP for commands.
|
||||
@@ -0,0 +1,23 @@
|
||||
# 功能复用文献核对
|
||||
|
||||
核对日期：2026-11-09。以下为原始论文摘要及相关章节核对，不是复现实验。
|
||||
本项目约束：不训练任何参数；已有静态试验没有执行模型前向。下面的方法不自动获得执行授权。
|
||||
|
||||
| 原始来源 | 研究的共享对象 / 证据 | 训练条件与适用限制 | 对本项目的用途 |
|
||||
|---|---|---|---|
|
||||
| [LISA, 2408.01890v2](https://arxiv.org/html/2408.01890v2), §3–5 | 跨层 token 注意力分布；考察头对齐、浅层敏感性；基准能力和吞吐验证 | 学习头对齐网络及低秩差异修正；不能归类为免训练替换。论文的 attention weights 在此主要指注意力分布，不是 W_Q/W_K 权重 | 借鉴逐头比较、首 token 对照和替换敏感性验证；不采用训练模块 |
|
||||
| [CLA, 2405.12981](https://arxiv.org/abs/2405.12981) | 相邻层共享 K/V heads，降低 KV cache；语言模型能力验证 | 摘要明确实验为从头训练 1B/3B 模型；不是现成 Qwen 的直接合并证据 | 区分 KV 缓存收益和参数/计算收益；作为架构可行性依据 |
|
||||
| [Systematic Study, NAACL 2025](https://aclanthology.org/2025.naacl-short.34/) | 系统比较跨层 KV 共享配置；语言建模、下游任务及生成吞吐 | 摘要报告激进配置涉及额外训练及 prefill 代价；不能由摘要推断零训练可迁移性 | 复用方向、共享拓扑及 prefill/decode 成本须分别记录 |
|
||||
| [MiniLM, 2002.10957v2](https://arxiv.org/abs/2002.10957) | 注意力分布及 value-value 关系蒸馏 | 明确训练学生模型；本项目不进行蒸馏 | 提醒检索分布和 value 内容需要分开考察 |
|
||||
| [CKA, ICML 2019](https://proceedings.mlr.press/v97/kornblith19a.html) | 表征之间的关系相似性，不是模块替换实验 | 论文讨论表征比较及可逆变换不变性的局限；高 CKA 不是可替换保证 | 可作为候选诊断，不能取代输出误差、干预和成本评测 |
|
||||
| [ALBERT, 1909.11942v6](https://arxiv.org/html/1909.11942v6), §3.1/4.5 | 跨层参数共享及训练后能力验证 | 共享架构参与训练；不能直接外推为已训练 checkpoint 可无损合并 | 参数存储减少与前向运算减少分别记账 |
|
||||
|
||||
## 对当前试验的推论
|
||||
|
||||
以上文献给出研究动机，没有证明 Qwen3.5-2B 已存在免训练的有效复用组。
|
||||
静态权重试验也没有否定该可能性。应使用同输入功能探测筛选，再对实际计算图实施单处干预。
|
||||
Qwen 的 full attention 与线性注意力需分别定义探测量；不能把 softmax attention 的结论照搬到状态递推模块。
|
||||
|
||||
数学工具参考：[SciPy orthogonal_procrustes](https://docs.scipy.org/doc/scipy/reference/generated/scipy.linalg.orthogonal_procrustes.html)。
|
||||
纯正交最小二乘有直接的矩阵分解解法，无须枚举旋转矩阵；但求解一个对齐并不自动带来可省略的计算。
|
||||
本项目下一阶段默认不求 R，直接测目标函数和干预效果。
|
||||
@@ -0,0 +1,132 @@
|
||||
# Qwen3.5-2B 有界试验结果
|
||||
|
||||
本轮已完成约定时间预算内的试验。整 tensor 精确比较 2,000 对；子块在内部截止时
|
||||
完成 14,778 / 20,000 对（73.89%），其余 5,222 对未计算。不能把本轮写成全量搜索。
|
||||
|
||||
结论：在本次候选、坐标对齐及误差阈值范围内，没有获得显著的直接权重共享收益。
|
||||
这不等于证明该模型不能通过其他方法压缩，也不代表 MoE 模型具有同样结果。
|
||||
|
||||
## 已确认的试验范围
|
||||
|
||||
- 模型：官方 `Qwen/Qwen3.5-2B`，revision `15852e8c16360a2fea060d615a32b45270f8a8fc`。
|
||||
- 原始 safetensors 文件为 4,548,221,488 字节，已核对官方 LFS SHA256。
|
||||
- 共 632 个存储 tensor；权重数据本身为 4,548,144,832 字节。
|
||||
- 实际存储参数 2,274,069,824 个，包含官方 checkpoint 中的视觉部分。
|
||||
- 维数分布：319 个一维、294 个二维、18 个三维、1 个五维。
|
||||
- 三维权重是 `[6144, 1, 4]` 的卷积核，包含长度为 1 的轴；不能把它解释为一般三维体积。
|
||||
- 五维视觉权重纳入完整清单和全模型字节分母，本轮不扩展五维子块搜索。
|
||||
- 不执行前向计算、能力评测、训练或蒸馏。
|
||||
|
||||
## 整 tensor 的结果
|
||||
|
||||
已完成 2,000 对同角色、同形状候选的精确比较。它们是经过近邻筛选的候选，
|
||||
不是所有可能 tensor 对的穷举。
|
||||
|
||||
| 最大直接替换相对误差 | 非单成员共享组 | 最大组大小 | 净节省字节 | 占模型权重 |
|
||||
|---|---:|---:|---:|---:|
|
||||
| 1% | 0 | 1 | 0 | 0% |
|
||||
| 2% | 0 | 1 | 0 | 0% |
|
||||
| 5% | 1 | 2 | 0 | 0% |
|
||||
| 10% | 11 | 6 | 32,704 | 0.000719% |
|
||||
| 20% | 16 | 15 | 119,424 | 0.002626% |
|
||||
|
||||
20% 阈值下找到的共享组均为一维权重。组数增加不代表显著的字节收益。
|
||||
这里的百分比是权重误差，绝不是任务能力损失百分比。
|
||||
|
||||
净节省按本轮声明的理论布局计算：保留组代表，每个分析对象 16 字节映射，
|
||||
每个共享组 64 字节描述；若大于原存储，则保持原表示。
|
||||
|
||||
这部分结果不支持在本次搜索范围内通过直接共享整 tensor 显著压缩该 checkpoint。
|
||||
它不排除其他未比较组合、子块共享、重参数化或带残差的压缩方法。
|
||||
|
||||
## 子块结果
|
||||
|
||||
共筛选 13,328 个抽样子块。20% 直接替换误差阈值下：
|
||||
|
||||
这里的 tensor 指完整命名权重数组，block 指其局部子张量，不是 Transformer block。
|
||||
三维原始权重为 6144×1×4，本轮切片实际是 1×1×4（4 个数），局部块是
|
||||
4×1×4（16 个数）；小块的高余弦不能推广为整份三维权重相似。
|
||||
|
||||
| 表示 | 非单成员共享组 | 最大组 | 映射开销前节省 | 当前布局净节省 |
|
||||
|---|---:|---:|---:|---:|
|
||||
| 64×64 二维块 | 3 | 3 | 40,960 字节 | 0 |
|
||||
| 128×128 二维块 | 1 | 2 | 32,768 字节 | 0 |
|
||||
| 三维 axis-0 切片 | 45 | 12 | 872 字节 | 0 |
|
||||
| 三维局部块 | 2 | 2 | 64 字节 | 0 |
|
||||
|
||||
二维共享候选均来自 `model.visual.pos_embed.weight`，未在已复核候选中形成大规模
|
||||
投影矩阵共享。不同块尺度是替代方案，不能相加。
|
||||
|
||||
净收益为零与声明的映射成本有关，不应解读为完全没有相似子块。更稀疏或更紧凑的
|
||||
索引格式可能降低开销，但本轮找到的二维毛节省也仅为几十 KB，尚不足以支持大幅
|
||||
减容。三维切片组数量较多，但整组三维权重合计只占模型字节的约 0.01945%。
|
||||
|
||||
模型字节占比：一维 0.01972%、二维 99.89167%、三维 0.01945%、五维 0.06917%。
|
||||
因此，应按字节贡献评估价值，不能按 tensor 数量或共享组数量推断收益。
|
||||
|
||||
全部分组是满足代表半径约束的可行方案，不保证达到最少组数或最优压缩率。
|
||||
|
||||
## 运行经验
|
||||
|
||||
1. 实际服务器内存约 1.7 GiB，新增的 60 GB 是磁盘扩容，不是内存。
|
||||
使用分块 BF16 解码和 mmap 页面释放完成了完整权重扫描与整 tensor 比较。
|
||||
2. 只看进程 RSS 不足以解释运行成本。`vmstat` 快照观察到 75%～85% 的 I/O 等待，
|
||||
同时存在交换活动；这些是共享主机的快照，不能全部归因于本进程。
|
||||
3. 与权重读取并行的绘图合成测试触发过 300 秒超时；观察到测试处于磁盘等待。
|
||||
后续验证应在权重读取结束后串行执行，避免把资源争用误判为数值算法失败。
|
||||
4. Matplotlib 图例改为固定位置，避免自动定位引入额外布局开销。
|
||||
5. 公平地选择各角色的候选，与高效地安排读取顺序，是两个不同问题。
|
||||
下一轮应评估“先固定公平候选集合，再按文件/源 tensor 分批执行”的效果，
|
||||
比较是否得到相同指标及更低 I/O 开销；本轮不在中途改变候选集合。
|
||||
6. 从已保存元数据计算，13,328 个抽样子块的原始数据合计 256,387,072 字节
|
||||
（约 244.5 MiB），对应指纹约 100.1 MiB。这是下一轮原始子块缓存的容量估计，
|
||||
不是本轮已经实现的缓存或实测加速。应先用小样本验证数值一致性、峰值内存和
|
||||
I/O 改善，再决定是否作为默认方案；不能仅凭容量估计保证运行效率。
|
||||
|
||||
参数分析及分组汇总耗时 6,661.13 秒（约 111.0 分钟），分析进程峰值 RSS 约
|
||||
201.3 MiB；这不包含系统文件缓存和同机服务内存。内部比较截止后保存覆盖率与
|
||||
分组，整个参数分析未触及 7,200 秒进程硬上限。块比较阶段将本进程 IO 优先级
|
||||
调整为 best-effort 7，未改变服务器全局设置。
|
||||
|
||||
最终修图版 6 项串行合成测试通过（约 77.2 秒），六类图表的 PNG/PDF/SVG 已生成，
|
||||
并对全部六类 PNG 做视觉检查；程序还核验了分组成员与代表之间的精确误差边。
|
||||
|
||||
## 三张诊断图的修订与解读
|
||||
|
||||
- 热力图只展示 `linear_attn.in_proj_a.weight`（16×2048），不是整个层。18 个相关层
|
||||
的 153 种配对中核验了 74 对；已测非对角余弦约 0.011～0.298。
|
||||
对角线 1 是身份比较，不是跨层发现。未测格子改用斜线，避免与接近零的底色混淆。
|
||||
- 原混合层距曲线受维数组成影响：层距 1 含 183 对一维、19 对二维；层距 8
|
||||
含 52 对一维、75 对二维。一维余弦中位数约从 0.941 变为 0.883，
|
||||
不能将混合曲线的陡降全部归因于层距。修订版按维数/块尺度分开，模块和筛选偏差仍存在。
|
||||
- 随机对照分为保留分布的元素打乱、匹配范数的零均值高斯向量、真实同角色随机配对。
|
||||
元素打乱后仍高余弦的情况主要出现在一维归一化、偏置和 A_log，提示均值/分布影响。
|
||||
修订版按维数分面、按密度归一化，并显示样本数，避免将不同样本量的峰高当概率比较。
|
||||
它们是诊断对照，不是显著性检验，也没有完全匹配每个模块的采样构成。
|
||||
三维对照分层仅有约 3～6 个样本，尤其不能据其峰形作总体推断。
|
||||
|
||||
这次只从已有 CSV/JSON 重绘并导出诊断表，模型权重已删除，没有增加参数比较或重新运行模型。
|
||||
|
||||
## 限制及后续决策
|
||||
|
||||
- 只研究一个 checkpoint；近邻筛选、同角色分组和子块抽样均可能漏掉其他共享机会。
|
||||
- 直方图和层距曲线存在候选筛选与模块混合偏差，不能用作总体分布或独立样本推断。
|
||||
- 光谱/子空间分析仅使用有界子体积；分布或子空间相似不等于可直接替换。
|
||||
- 本轮没有测量能力损失、困惑度、运行内存或推理速度，也没有产出可运行的压缩模型。
|
||||
- 大幅减少推理计算量需要进一步的计算图和路由分析，单纯共享存储不能保证加速。
|
||||
|
||||
下一轮先讨论候选选择与执行局部性、原始子块缓存、映射格式和阶段预算，
|
||||
再修订 SOP；不自动下载第二个模型。只有获得额外授权后，才进行校准前向及
|
||||
留出任务测试，以验证替换造成的能力损失。
|
||||
|
||||
数据与标准化报告见 `../results/qwen35-2b/`；具体阈值、分组、覆盖率、来源和校验记录
|
||||
随原始图表数据归档。权重在交付副本校验后按授权删除，重现实验需要按固定 revision
|
||||
重新下载。
|
||||
|
||||
## 功能复用假设仍待验证
|
||||
|
||||
本轮坐标对齐权重结果不能排除 Q/K 的联合功能接近或 MLP 分支可复用。
|
||||
下一阶段采用同一隐藏输入下的函数比较，不训练参数、不默认求正交 R；
|
||||
注意力分布、完整分支输出、真实网络单处替换和运行收益分别验证。
|
||||
详见 [免训练有界方案](TRAINING_FREE_FUNCTIONAL_PLAN.md) 与
|
||||
[文献核对](FUNCTIONAL_REUSE_LITERATURE.md)。这些是方案，尚未执行。
|
||||
+123
@@ -0,0 +1,123 @@
|
||||
# Parameter similarity SOP — 2B pilot draft
|
||||
|
||||
Status: v0.1, executed for the bounded Qwen3.5-2B pilot. See [findings](PILOT_FINDINGS.md).
|
||||
Changes proposed from this experience require a small equivalence/performance check before
|
||||
the next model; no second model is started automatically.
|
||||
|
||||
## Scope and decision
|
||||
|
||||
Analyze official, unquantized safetensors without instantiating a model, running inference,
|
||||
training, or executing repository-supplied code. Start with Qwen/Qwen3.5-2B only.
|
||||
The question is the number of representative-centered sharing groups under direct weight
|
||||
replacement error limits, and the net byte savings of a theoretical shared representation.
|
||||
Weight error is not task-accuracy loss, and parameter sharing does not imply fewer FLOPs.
|
||||
|
||||
## Prerequisites
|
||||
|
||||
1. Confirm authorized remote host, free disk, memory, CPU, existing work, and access.
|
||||
2. Install prebuilt wheels only. Never compile locally or silently use another runner.
|
||||
3. Freeze the official repository revision and download file manifest including SHA256.
|
||||
4. Allocate checkpoint size plus 3 GB disk margin. Use an experiment-owned directory,
|
||||
not a shared Hugging Face cache. Never delete pre-existing weights.
|
||||
|
||||
## Bounded procedure
|
||||
|
||||
1. Run synthetic correctness tests on the remote host: BF16, 3D indexing, directional
|
||||
replacement error, and non-transitive groups.
|
||||
2. Download one checkpoint sequentially and verify upstream sizes and LFS SHA256.
|
||||
3. Scan each tensor in 262,144-element chunks. Record shape, dtype, axis role,
|
||||
layer, bytes, moments, max magnitude, and finite-value checks. Memory-map weights;
|
||||
evict scanned pages. Do not search for exactly equal weight tensors.
|
||||
4. Whole-tensor screening: same role, same shape, same dtype. Sample 2,048 aligned,
|
||||
evenly spaced elements. Select up to 10 neighbors by each of cosine and relative
|
||||
error. Fairly distribute the 2,000-pair exact-verification cap across role groups.
|
||||
5. Block screening: non-overlapping 64x64 and 128x128 2D partitions. For 3D tensors,
|
||||
use axis-0 slices and bounded volumes adapted to actual dimensions. Keep up to
|
||||
256 uniformly reservoir-sampled blocks per role/shape/scale for the 1.7 GiB pilot host
|
||||
(below the planned maximum of 512). Record edge bytes
|
||||
and sampling coverage. Verify at most 20,000 block pairs. Never extrapolate sampled
|
||||
discoveries to unseen blocks or add savings across alternative scales.
|
||||
6. Exact metrics: full-element cosine, Pearson, directional replacement errors,
|
||||
symmetric error, optimal-scalar residual. Tail diagnostics use samples and say so.
|
||||
Controls include random actual comparable pairs, shuffled sampled elements,
|
||||
and norm-matched Gaussian vectors. Save sampled versus exact error discrepancies.
|
||||
Bounded mode-unfolding SVD probes are diagnostic only, not a replacement test.
|
||||
7. Form deterministic star groups at 1%, 2%, 5%, 10%, 20% relative weight error.
|
||||
Every member requires an exact verified edge to its representative. Unknown
|
||||
pairs remain independent; never use graph connected components as sharing groups.
|
||||
8. Compute original bytes minus representatives, 16-byte mapping per analyzed object,
|
||||
and 64-byte descriptor per shared group. Fall back to unchanged storage if larger.
|
||||
Count savings against the entire model, preserving unsampled blocks and edge regions.
|
||||
9. Save raw data, groups, thresholds, coverage, timings, resource usage, figures
|
||||
(PNG/PDF/SVG), model metadata and report. Validate every group against exact edges.
|
||||
10. Copy results to the deliverable workspace and verify checksums before deleting
|
||||
the experiment-owned checkpoint. Save a cleanup record. Do not delete the venv,
|
||||
result data or any unrelated cache.
|
||||
11. Review full diff, commit and push a work branch with Codex Bot identity, then
|
||||
create a PR for manual merge. Never silently initialize/push main or auto-merge.
|
||||
|
||||
## Budget and stop conditions
|
||||
|
||||
- Real 2B analysis computation: hard wall-time limit 7,200 seconds. Reserve final time
|
||||
for summaries; download and report-generation time are recorded separately.
|
||||
- Use one BLAS thread to avoid interference with existing services.
|
||||
- Save exact-pair tables every 100 pairs; stage deadlines prevent whole-tensor work
|
||||
from consuming the complete block budget.
|
||||
- On timeout or failure, retain source weights until partial results are recoverable
|
||||
and report coverage/failure truthfully. An internal planned cutoff may produce a complete
|
||||
bounded report with partial candidate coverage; always state verified/selected counts.
|
||||
An external kill or missing required report is an incomplete run, not a completed experiment.
|
||||
- Empty or negligible low-error sharing is a valid negative result. Do not add scales,
|
||||
increase error thresholds, align permutations, train or expand to other models automatically.
|
||||
- No runtime capability/performance claim without separately authorized evaluation.
|
||||
|
||||
## Commands (remote, from project root)
|
||||
|
||||
```sh
|
||||
OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 timeout 300 venv/bin/python -B -m unittest discover -s tests -v
|
||||
venv/bin/python -B -m pilot.download --root runs/qwen35-2b
|
||||
OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 timeout 7200 venv/bin/python -B -m pilot.analyze --root runs/qwen35-2b --seconds 6600
|
||||
OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 venv/bin/python -B -m pilot.plot --root runs/qwen35-2b
|
||||
OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 timeout 300 venv/bin/python -B -m unittest discover -s tests -v > runs/qwen35-2b/tests.log 2>&1
|
||||
venv/bin/python -B -m pilot.audit --root runs/qwen35-2b --analysis-code-commit 6121016
|
||||
venv/bin/python -B -m pilot.validate --root runs/qwen35-2b
|
||||
# Copy archived results, verify checksums at the destination, then:
|
||||
venv/bin/python -B -m pilot.validate --root runs/qwen35-2b --cleanup-weights
|
||||
```
|
||||
|
||||
## Follow-up decisions after the pilot
|
||||
|
||||
- The pilot verified 2,000 tensor pairs and 14,778/20,000 selected block pairs before cutoff.
|
||||
It took about 111 minutes; the final six-test suite including revised figure checks passed in about 77 seconds.
|
||||
- Hatch missing heatmap cells and label identity values as constructed. Separate layer-distance
|
||||
curves by dimension/scale; retain warnings about within-stratum module/selection bias.
|
||||
Normalize control histograms, display sample counts, and verify their areas equal one.
|
||||
Keep these diagnostics distinct from task-equivalence or statistical-significance claims.
|
||||
- Avoid concurrent plotting tests and heavy weight IO on the shared host. Record IO wait,
|
||||
swap, CPU and RSS; a low process RSS is not proof of low total resource cost.
|
||||
- Distinguish tensor/group counts from byte contributions: 1D tensors were about half the
|
||||
inventory count but only 0.0197% of weight bytes in this checkpoint.
|
||||
- Keep fair candidate selection separate from IO-local execution order. Benchmark batching
|
||||
by source file/tensor without changing the selected pairs or metrics.
|
||||
- The sampled raw blocks total about 244.5 MiB. Test an explicitly bounded raw BF16/F32
|
||||
cache against the uncached implementation before adopting it; include fingerprint and
|
||||
temporary conversion memory in the budget. This was not enabled in the current run.
|
||||
- Report both pre-metadata and net savings. Tiny blocks can have groups yet lose all savings
|
||||
to mapping overhead. Any alternative sparse encoding needs its own explicit accounting.
|
||||
- Separate low-byte normalization groups from matrix/expert compression opportunities.
|
||||
- Decide whether approximate-neighbor recall and block coverage justify another bounded pass.
|
||||
- If meaningful candidates exist, request a small calibration-forward experiment to measure
|
||||
output distortion, then held-out loss/tasks, then implementation-specific memory and latency.
|
||||
- For MoE, analyze router behavior and joint gate/up/down expert consistency before treating
|
||||
independently similar matrix blocks as interchangeable experts.
|
||||
- Use these observations to revise this SOP before downloading a second model.
|
||||
|
||||
## Training-free functional follow-up (design only)
|
||||
|
||||
Direct coordinate-aligned parameter similarity is a baseline, not a test of Q/K/V or MLP
|
||||
functional interchangeability. The follow-up uses same-input module probes without solving
|
||||
rotations or training any parameters, then single-site intervention and held-out quality/cost
|
||||
validation. See the [bounded plan](TRAINING_FREE_FUNCTIONAL_PLAN.md) and
|
||||
[primary-source literature](FUNCTIONAL_REUSE_LITERATURE.md). No forward pass, new download,
|
||||
or new model is started by adopting this document. The original static pilot remains
|
||||
inference-free; the proposed follow-up requires a separately confirmed execution environment.
|
||||
@@ -0,0 +1,119 @@
|
||||
# 免训练功能复用：下一阶段有界方案
|
||||
|
||||
状态：研究设计，尚未执行。原始参数试验已完成；没有新增前向、下载或训练。
|
||||
目标：寻找并验证至少一个在留出数据上保持约定质量、同时降低明确运行成本的复用实例；
|
||||
若没有通过者，报告有界搜索的阴性结果，不保证一定能找到。
|
||||
|
||||
## 1. 硬约束与研究对象
|
||||
|
||||
- 冻结所有模型参数。不进行优化器更新、微调、蒸馏、LoRA、学习投影或低秩修正。
|
||||
- 默认不求正交 R。校准集只用于选择有限候选和阈值，留出集不参与选择。
|
||||
- 同类型模块配对；Q/K/V 联合作为功能系统，MLP 独立验证。保留归一化、位置处理、门控及头结构。
|
||||
- 已存配置为 24 个文本层，其中 6 个 full attention、18 个线性注意力；执行前按固定 revision 的实际实现复核。
|
||||
full attention 的 15 个无序层对可穷举；线性注意力首轮只检查 6 个预先固定的相邻同类型层对。
|
||||
MLP 首轮检查 6 个预先固定、覆盖浅中深层的相邻层对。具体名单在运行前写入 manifest。
|
||||
- 之前热力图的 in_proj_a 是线性注意力控制投影示例，不是 Q/K/V 功能相似性的测量。
|
||||
|
||||
## 2. 为什么不用先找 R
|
||||
|
||||
给定相同隐藏输入 H，且位置、mask 和缓存/递推状态一致，比较两套模块的实际检索函数：
|
||||
|
||||
S_i(H) = Q_i(H) K_i(H)^T / sqrt(d)，P_i(H) = softmax(S_i(H) + mask)。
|
||||
|
||||
这里 Q/K 指完成该架构实际归一化和位置处理后的量。直接比较 P 可避开内部坐标差异；
|
||||
logit 每行加常数不改变 softmax，因此同时报告去行均值的分数误差。
|
||||
|
||||
这不是“存在共同正交 R”的数学证明。反例 Q_2=2Q_1、K_2=K_1/2 保持所有点积，
|
||||
但非零 Q 的范数改变，不能由正交变换产生。功能等价允许比共同正交旋转更广的关系。
|
||||
相反，在单套 Q/K 上共同右乘 R 保持点积，只说明坐标不唯一，并未说明两个已训练层可复用。
|
||||
实际 RoPE、GQA、归一化、门控也必须包含在函数边界内。
|
||||
|
||||
有限探针只能提供相应输入分布上的经验证据，不是对所有 H 的形式证明。
|
||||
在无位置处理/归一化的纯线性特例中，可研究 W_Q W_K^T 的代数相等性；
|
||||
一般模块不据此简化，首轮不物化所有大算子。
|
||||
|
||||
## 3. 同输入与真实输入分开
|
||||
|
||||
对候选 i/j，分别从原模型 i 和 j 的入口收集 H_i、H_j；分别比较 f_i(H_i) 与 f_j(H_i)，
|
||||
以及 f_i(H_j) 与 f_j(H_j)。避免只选有利于某一层的输入。模块前归一化如位于边界外，
|
||||
必须明确分别测“归一化后核心”和“包含归一化的完整分支”。
|
||||
|
||||
这检验模块函数接近程度。实际网络的 f_i(H_i) 与 f_j(H_j) 使用不同输入，是另一项量。
|
||||
权重共享仍需对每个输入计算；缓存上游输出用于下游则改变了输入，必须实际干预后检验。
|
||||
MLP 测残差相加前的分支输出，防止共同输入 H 掩盖分支差异。
|
||||
|
||||
## 4. 指标和对照
|
||||
|
||||
| 对象 | 首轮指标 | 不能据此宣称的结论 |
|
||||
|---|---|---|
|
||||
| full attention 逐头 P | 平均及 P95 行 JS（自然对数，范围 0–ln2）、top-5 重合率；去行均值 S 相对误差 | P 相似不保证 V/O 输出相似 |
|
||||
| 完整 attention 分支 | 相对 L2、逐 token 误差 P50/P95、余弦；包含 V/O/门控 | 平均低误差不保证长尾任务能力 |
|
||||
| 线性注意力 | 同输入、同合法初始状态的分支输出及最终状态误差；后续状态继续使用测试 | 不构造不存在的 softmax 图；状态数值相似不等于长期稳定 |
|
||||
| MLP | 同输入分支输出相对 L2、逐 token P95、余弦 | 不以各矩阵单独相似代替 gate/up/down 联合验证 |
|
||||
|
||||
相对误差统一为 ||candidate-reference|| / max(||reference||, epsilon)，并记录低范数样本数量。
|
||||
首轮筛选门槛预设：P 平均 JS ≤0.01 且 top-5 重合 ≥0.8；完整分支相对 L2 ≤0.05、
|
||||
逐 token 相对误差 P95 ≤0.1。它们是筛选约定，不是文献保证或能力损失界。
|
||||
线性模块和 MLP 使用输出门槛；只通过 P 的候选不能自动通过完整模块门槛。
|
||||
|
||||
对照：同模块重复测量、随机同类型层对；首 token 保留/移除并重新归一化两个版本；
|
||||
排除自注意与局部窗口主导后再测；按序列/位置/浅中深层分层。零分母与无效行单独标记。
|
||||
逐头记录，不先平均头。默认只用原头序；可选的离散头排列仅在校准集选定后冻结，
|
||||
须满足 GQA 分组和 V/O 一致性，不允许把最优头匹配分数当作可执行替换结果。
|
||||
CKA 仅作辅助筛选；共享位置模板、attention sink 和共同残差都可能制造高相似。
|
||||
|
||||
## 5. 资源与阶段预算
|
||||
|
||||
当前服务器实际约 1.7 GiB RAM，不按“60 GB 内存”安排前向；60 GB 是磁盘扩容。
|
||||
后续需要用户确认可执行前向的远程环境；建议先评估至少 16 GiB RAM/显存的单模型环境，
|
||||
这不是已验证最低配置。先做内存估算及一次短样本容量探测，不足即停止，不自动换机器。
|
||||
不在本机编译、不自动重下已删除权重。本轮只交付设计。
|
||||
|
||||
后续若获执行授权，首轮固定模型/revision/dtype、batch=1、长度最多 256，
|
||||
校准和留出各 16 条固定且去重的文本序列（中文/英文各半），只处理文本路径。
|
||||
数据来源、许可、tokenizer revision、截断、样本 hash 和抽样 seed 必须先入 manifest。
|
||||
每序列至多 32 个固定 query 位置，保留这些行的全部合法 key，避免采样 key 改变 softmax。
|
||||
流式聚合，不保存全部层的 N×N attention，不调用生成式聊天作替代评测。
|
||||
|
||||
| 阶段 | 上限 | 完成条件 / 停止条件 |
|
||||
|---|---|---|
|
||||
| 容量探测及基线 | 15 分钟 | 无 OOM，固定输出重复测量可复现；否则停止 |
|
||||
| 同输入探测与筛选 | 45 分钟 | 固定 27 对候选；每类别最多保留 1 对，总计最多 3 对；无通过者即停止后续干预 |
|
||||
| 单处干预与质量 | 45 分钟 | 逐个候选，不叠加；留出集测 token NLL 和逐序列变化；失败不追加搜索 |
|
||||
| 成本测量和归档 | 15 分钟 | 保存基线/干预的质量、内存和计时，超时则输出已完成覆盖率 |
|
||||
|
||||
总计算上限 120 分钟；下载单独记录，不触发无限重试。16 条留出序列只是先导证据，
|
||||
不能据此宣布广泛能力保持。首轮通过后才另立更大留出任务验证计划。
|
||||
|
||||
## 6. 干预必须对应可节省的对象
|
||||
|
||||
| 干预 | 可以争取的收益 | 必须额外确认 |
|
||||
|---|---|---|
|
||||
| 跨层绑定权重 | 权重存储/加载内存 | 各层输入仍不同，通常不减少投影 FLOPs |
|
||||
| 重用上游 attention 分布，保留下游 V/O | 部分 Q/K 投影与 QK/softmax 计算 | 保留分布的内存及实现开销；decode 未必值得缓存完整图 |
|
||||
| 跨层共享 K/V cache | KV 内存，可能减少 K/V 投影 | GQA、位置、缓存布局与层依赖合法性 |
|
||||
| 重用完整 attention/MLP 输出 | 对应分支计算 | 下游输入不同，必须测试整个原模型中的替换效应 |
|
||||
|
||||
选择的每个候选首先给出一种具体可实现干预；只有同输入指标通过却无实际可省操作，
|
||||
标记“功能相似但暂无复用实现”。不把整模块可替换和其中 Q/K 可替换混为一谈。
|
||||
|
||||
首轮探索性质量门槛：留出平均 token NLL 增量 ≤0.01 nat/token，逐序列增量 P95 ≤0.05，
|
||||
输出有限；同时保存全部逐序列差值，不将该小样本阈值称为统计显著性或普适能力保证。
|
||||
候选只在校准集筛选；留出失败不回调阈值、不反复选择。
|
||||
|
||||
性能分别测 prefill 和固定 32-token decode；固定 batch/长度/dtype/backend，预热 3 次、
|
||||
交错运行基线与干预各 10 次，报告原始时延和中位数/P95、峰值内存、KV/权重字节及理论运算变化。
|
||||
插桩诊断时延不作为优化实现时延。若速度差落在运行波动内，不宣称加速；
|
||||
若只减内存，则结论限定为内存收益。计入缓存、索引与数据移动开销。
|
||||
|
||||
## 7. 分组与最终证据
|
||||
|
||||
单处通过后，才讨论多成员组：每个成员必须通过对同一代表的定向替换检查；
|
||||
不能用 A≈B、B≈C 推出 A 可代表 C。多处同时替换须再测累积误差和性能，
|
||||
单处收益不能相加为最终模型收益。首轮不自动执行多处替换。
|
||||
|
||||
输出每对候选/每序列/每头指标 CSV、覆盖率和失败原因、质量—成本散点图、逐层热图、
|
||||
基线与干预原始计时、环境及代码版本。结论分为：功能接近、单处干预通过、测得实际收益三档。
|
||||
“存在一个免训练复用点”的操作性证据必须同时具有后两项；当前静态试验尚未提供这类证据。
|
||||
|
||||
相关依据与训练限制见 [文献核对](FUNCTIONAL_REUSE_LITERATURE.md)。
|
||||
@@ -0,0 +1 @@
|
||||
"""Bounded, inference-free weight similarity pilot."""
|
||||
@@ -0,0 +1,396 @@
|
||||
"""Bounded CPU-only screening, exact verification and conservative sharing groups."""
|
||||
import argparse
|
||||
import collections
|
||||
import csv
|
||||
import hashlib
|
||||
import itertools
|
||||
import json
|
||||
import math
|
||||
import pathlib
|
||||
import platform
|
||||
import resource
|
||||
import time
|
||||
|
||||
import numpy as np
|
||||
|
||||
from .tensors import decode, finish, load_tensors, metrics
|
||||
|
||||
THRESHOLDS = [.01, .02, .05, .10, .20]
|
||||
|
||||
|
||||
def write_json(path, obj):
|
||||
path.write_text(json.dumps(obj, indent=2, allow_nan=False))
|
||||
|
||||
|
||||
def write_csv(path, rows):
|
||||
if not rows:
|
||||
path.write_text('')
|
||||
return
|
||||
with path.open('w', newline='') as f:
|
||||
writer = csv.DictWriter(f, fieldnames=list(rows[0]))
|
||||
writer.writeheader()
|
||||
writer.writerows(rows)
|
||||
|
||||
|
||||
def stable_seed(key, seed):
|
||||
return int.from_bytes(hashlib.sha256((str(seed) + str(key)).encode()).digest()[:8], 'little')
|
||||
|
||||
|
||||
def read_object(obj, tensors):
|
||||
t = tensors[obj['tensor']]
|
||||
if obj['kind'] == 'tensor':
|
||||
return t.read()
|
||||
slices = tuple(slice(s, s + b) for s, b in zip(obj['start'], obj['shape']))
|
||||
return decode(t.raw.reshape(t.shape)[slices], t.dtype).ravel()
|
||||
|
||||
|
||||
def sample_object(obj, tensors, sample_size=2048):
|
||||
t = tensors[obj['tensor']]
|
||||
n = math.prod(obj['shape'])
|
||||
indices = np.linspace(0, n - 1, min(n, sample_size), dtype=np.int64)
|
||||
if obj['kind'] == 'tensor':
|
||||
return t.read(indices)
|
||||
coords = np.unravel_index(indices, obj['shape'])
|
||||
global_coords = tuple(c + s for c, s in zip(coords, obj['start']))
|
||||
flat = np.ravel_multi_index(global_coords, t.shape)
|
||||
return t.read(flat)
|
||||
|
||||
|
||||
def exact_pair(a, b, tensors, deadline):
|
||||
if a['kind'] != 'tensor':
|
||||
result = metrics(read_object(a, tensors), read_object(b, tensors))
|
||||
tensors[a['tensor']].evict()
|
||||
tensors[b['tensor']].evict()
|
||||
return result
|
||||
ta, tb = tensors[a['tensor']], tensors[b['tensor']]
|
||||
xx = yy = xy = delta = sx = sy = 0.
|
||||
for start in range(0, ta.numel, 262144):
|
||||
if time.monotonic() > deadline:
|
||||
raise TimeoutError('Analysis deadline reached')
|
||||
sl = slice(start, min(start + 262144, ta.numel))
|
||||
x, y = ta.read(sl).astype(np.float64), tb.read(sl).astype(np.float64)
|
||||
xx += float(x @ x)
|
||||
yy += float(y @ y)
|
||||
xy += float(x @ y)
|
||||
d = x - y
|
||||
delta += float(d @ d)
|
||||
sx += float(x.sum())
|
||||
sy += float(y.sum())
|
||||
ta.evict()
|
||||
tb.evict()
|
||||
ta.evict()
|
||||
tb.evict()
|
||||
return finish(xx, yy, xy, delta, sx, sy, ta.numel)
|
||||
|
||||
|
||||
def sharing_groups(objects, pairs, threshold):
|
||||
"""Star groups; every member must have an exact verified edge to its representative."""
|
||||
adjacent = collections.defaultdict(dict)
|
||||
for pair in pairs:
|
||||
a, b = pair['a'], pair['b']
|
||||
if pair['error_member_to_rep'] <= threshold:
|
||||
adjacent[b][a] = pair['error_member_to_rep']
|
||||
if pair['error_rep_to_member'] <= threshold:
|
||||
adjacent[a][b] = pair['error_rep_to_member']
|
||||
unassigned = set(range(len(objects)))
|
||||
order = sorted(unassigned, key=lambda i: (-len(adjacent[i]), objects[i]['id']))
|
||||
groups = []
|
||||
for rep in order:
|
||||
if rep not in unassigned:
|
||||
continue
|
||||
members = sorted(set(adjacent[rep]) & unassigned)
|
||||
members.append(rep)
|
||||
unassigned.difference_update(members)
|
||||
groups.append({'representative': rep, 'members': members,
|
||||
'max_error': max([adjacent[rep][i] for i in members if i != rep] or [0.])})
|
||||
return groups
|
||||
|
||||
|
||||
def nearest_candidates(objects, features, cap, neighbors=10):
|
||||
buckets = collections.defaultdict(list)
|
||||
for i, obj in enumerate(objects):
|
||||
buckets[(obj['role'], tuple(obj['shape']), obj['scale'], obj['dtype'])].append(i)
|
||||
candidates = []
|
||||
eligible_pairs = 0
|
||||
for key, ids in sorted(buckets.items(), key=lambda item: str(item[0])):
|
||||
eligible_pairs += len(ids) * (len(ids) - 1) // 2
|
||||
if len(ids) < 2:
|
||||
continue
|
||||
x = np.stack([features[i] for i in ids]).astype(np.float64)
|
||||
norm = np.linalg.norm(x, axis=1)
|
||||
u = x / np.maximum(norm[:, None], 1e-30)
|
||||
# Cap each bucket's reservoir to bound the screening matrix.
|
||||
cos = u @ u.T
|
||||
sq = np.maximum(0., norm[:, None] ** 2 + norm[None, :] ** 2 - 2 * (x @ x.T))
|
||||
err = np.sqrt(sq / np.maximum(norm[:, None] ** 2, 1e-30))
|
||||
np.fill_diagonal(cos, -np.inf)
|
||||
np.fill_diagonal(err, np.inf)
|
||||
seen = set()
|
||||
for row, i in enumerate(ids):
|
||||
near = set(np.argsort(err[row])[:neighbors]) | set(np.argsort(-cos[row])[:neighbors])
|
||||
for col in near:
|
||||
j = ids[col]
|
||||
edge = (min(i, j), max(i, j))
|
||||
if i == j or edge in seen:
|
||||
continue
|
||||
seen.add(edge)
|
||||
candidates.append({'a': edge[0], 'b': edge[1], 'screen_error': float(min(err[row, col], err[col, row])),
|
||||
'screen_cosine': float(cos[row, col])})
|
||||
# Low-error candidates first, then guarantee role/scale diversity round-robin.
|
||||
queues = collections.defaultdict(list)
|
||||
for c in sorted(candidates, key=lambda c: (c['screen_error'], c['a'], c['b'])):
|
||||
o = objects[c['a']]
|
||||
queues[(o['role'], tuple(o['shape']), o['scale'])].append(c)
|
||||
queues = [collections.deque(q) for _, q in sorted(queues.items(), key=lambda x: str(x[0]))]
|
||||
chosen = []
|
||||
while queues and len(chosen) < cap:
|
||||
for q in queues:
|
||||
if q and len(chosen) < cap:
|
||||
chosen.append(q.popleft())
|
||||
queues = [q for q in queues if q]
|
||||
return chosen, {'eligible_pairs_in_screened_objects': eligible_pairs,
|
||||
'candidate_pairs_before_cap': len(candidates), 'selected_pairs': len(chosen)}
|
||||
|
||||
|
||||
def object_record(t, kind='tensor', shape=None, start=None, scale='whole'):
|
||||
shape = tuple(shape or t.shape)
|
||||
return {'id': t.name if kind == 'tensor' else f'{t.name}|{scale}|{start}',
|
||||
'tensor': t.name, 'role': t.role, 'layer': t.layer, 'kind': kind, 'dtype': t.dtype,
|
||||
'shape': shape, 'start': start, 'scale': scale,
|
||||
'bytes': math.prod(shape) * (t.bytes // t.numel)}
|
||||
|
||||
|
||||
def build_blocks(tensors, seed, deadline, reservoir_limit=256):
|
||||
reservoirs, counts, rngs = {}, collections.Counter(), {}
|
||||
totals = collections.defaultdict(lambda: {'blocks': 0, 'bytes': 0, 'edge_bytes': 0})
|
||||
for t in tensors.values():
|
||||
if len(t.shape) == 2:
|
||||
specs = [(f'2d-{s}', (s, s)) for s in (64, 128) if min(t.shape) >= s]
|
||||
elif len(t.shape) == 3:
|
||||
specs = [('3d-axis0-slice', (1, min(64, t.shape[1]), min(64, t.shape[2]))),
|
||||
('3d-volume', (min(4, t.shape[0]), min(32, t.shape[1]), min(32, t.shape[2])))]
|
||||
else:
|
||||
continue
|
||||
for scale, shape in specs:
|
||||
nblocks = math.prod(d // b for d, b in zip(t.shape, shape))
|
||||
totals[scale]['blocks'] += nblocks
|
||||
block_bytes = math.prod(shape) * (t.bytes // t.numel)
|
||||
totals[scale]['bytes'] += nblocks * block_bytes
|
||||
totals[scale]['edge_bytes'] += t.bytes - nblocks * block_bytes
|
||||
for grid in itertools.product(*(range(d // b) for d, b in zip(t.shape, shape))):
|
||||
if time.monotonic() > deadline:
|
||||
raise TimeoutError('Deadline during block enumeration')
|
||||
start = tuple(i * b for i, b in zip(grid, shape))
|
||||
obj = object_record(t, 'block', shape, start, scale)
|
||||
key = (t.role, shape, scale)
|
||||
if key not in reservoirs:
|
||||
reservoirs[key], rngs[key] = [], np.random.default_rng(stable_seed(key, seed))
|
||||
counts[key] += 1
|
||||
reservoir = reservoirs[key]
|
||||
if len(reservoir) < reservoir_limit:
|
||||
reservoir.append(obj)
|
||||
else:
|
||||
j = int(rngs[key].integers(counts[key]))
|
||||
if j < reservoir_limit:
|
||||
reservoir[j] = obj
|
||||
objects = [obj for key in sorted(reservoirs, key=str) for obj in reservoirs[key]]
|
||||
coverage = [{'role': key[0], 'shape': str(key[1]), 'scale': key[2],
|
||||
'total_blocks': counts[key], 'sampled_blocks': len(reservoirs[key])}
|
||||
for key in sorted(reservoirs, key=str)]
|
||||
return objects, coverage, dict(totals)
|
||||
|
||||
|
||||
def structural_probe(a, b, tensors):
|
||||
"""Bounded mode unfoldings of aligned subvolumes; exploratory, never a grouping criterion."""
|
||||
def small(obj):
|
||||
t = tensors[obj['tensor']]
|
||||
shape = obj['shape']
|
||||
starts = obj['start'] or [0] * len(shape)
|
||||
slices = tuple(slice(s, s + min(d, 32)) for s, d in zip(starts, shape))
|
||||
return decode(t.raw.reshape(t.shape)[slices], t.dtype)
|
||||
x, y = small(a), small(b)
|
||||
result = []
|
||||
if x.ndim < 2:
|
||||
return result
|
||||
for axis in range(x.ndim):
|
||||
mx = np.moveaxis(x, axis, 0).reshape(x.shape[axis], -1)
|
||||
my = np.moveaxis(y, axis, 0).reshape(y.shape[axis], -1)
|
||||
if min(mx.shape) < 2:
|
||||
result.append({'axis': axis, 'probe_shape': list(x.shape), 'trivial_mode_skipped': True})
|
||||
continue
|
||||
ux, sx, _ = np.linalg.svd(mx, full_matrices=False)
|
||||
uy, sy, _ = np.linalg.svd(my, full_matrices=False)
|
||||
k = min(4, max(1, min(len(sx), len(sy)) // 2))
|
||||
result.append({'axis': axis, 'probe_shape': list(x.shape),
|
||||
'k': k,
|
||||
'normalized_spectrum_distance': float(np.linalg.norm(sx / max(np.linalg.norm(sx), 1e-30) - sy / max(np.linalg.norm(sy), 1e-30))),
|
||||
'left_subspace_overlap': float(np.linalg.norm(ux[:, :k].T @ uy[:, :k]) ** 2 / k)})
|
||||
return result
|
||||
|
||||
|
||||
def analyze_mode(mode, objects, tensors, root, deadline, cap, seed, baseline_cap=100):
|
||||
print(f'{mode}: sampling {len(objects)} objects', flush=True)
|
||||
features = []
|
||||
for obj in objects:
|
||||
features.append(sample_object(obj, tensors))
|
||||
tensors[obj['tensor']].evict()
|
||||
candidates, coverage = nearest_candidates(objects, features, cap)
|
||||
write_json(root / f'{mode}_objects.json', objects)
|
||||
pairs, nulls, structures = [], [], []
|
||||
rng = np.random.default_rng(seed)
|
||||
random_pairs = []
|
||||
comparable = collections.defaultdict(list)
|
||||
for j, obj in enumerate(objects):
|
||||
comparable[(obj['role'], tuple(obj['shape']), obj['scale'], obj['dtype'])].append(j)
|
||||
for key, ids in sorted(comparable.items(), key=lambda item: str(item[0])):
|
||||
if len(ids) < 2:
|
||||
continue
|
||||
for _ in range(3):
|
||||
a, b = rng.choice(ids, size=2, replace=False).tolist()
|
||||
random_pairs.append({'a': a, 'b': b, 'role': key[0], 'sampled': True,
|
||||
**metrics(features[a], features[b])})
|
||||
write_json(root / f'{mode}_random_pairs.json', random_pairs)
|
||||
timeout = False
|
||||
structural_counts = collections.Counter()
|
||||
for i, candidate in enumerate(candidates):
|
||||
if time.monotonic() > deadline:
|
||||
timeout = True
|
||||
break
|
||||
a, b = candidate['a'], candidate['b']
|
||||
try:
|
||||
measured = exact_pair(objects[a], objects[b], tensors, deadline)
|
||||
except TimeoutError:
|
||||
timeout = True
|
||||
break
|
||||
row = {**candidate, **measured,
|
||||
'a_id': objects[a]['id'], 'b_id': objects[b]['id'],
|
||||
'role': objects[a]['role'], 'scale_name': objects[a]['scale'],
|
||||
'layer_distance': abs(objects[a]['layer'] - objects[b]['layer'])}
|
||||
row['screen_vs_exact_min_error_absolute_difference'] = abs(candidate['screen_error'] - min(measured['error_member_to_rep'], measured['error_rep_to_member']))
|
||||
# Tail diagnostics are sampled, explicitly labelled, not claimed exact.
|
||||
fx, fy = features[a].astype(np.float64), features[b].astype(np.float64)
|
||||
row['sampled_error_q99_over_member_rms'] = float(np.quantile(np.abs(fx - fy), .99) / max(np.sqrt(np.mean(fx * fx)), 1e-30))
|
||||
pairs.append(row)
|
||||
if i < baseline_cap:
|
||||
for method, y in [('shuffled', rng.permutation(fy)), ('gaussian_norm_matched', rng.normal(size=fy.size))]:
|
||||
if method == 'gaussian_norm_matched':
|
||||
y *= np.linalg.norm(fy) / max(np.linalg.norm(y), 1e-30)
|
||||
nulls.append({'a': a, 'b': b, 'method': method, 'sampled': True, **metrics(fx, y)})
|
||||
ndim = len(objects[a]['shape'])
|
||||
if ndim >= 2 and structural_counts[ndim] < 20:
|
||||
structures.append({'a': a, 'b': b, 'modes': structural_probe(objects[a], objects[b], tensors)})
|
||||
tensors[objects[a]['tensor']].evict()
|
||||
tensors[objects[b]['tensor']].evict()
|
||||
structural_counts[ndim] += 1
|
||||
if (i + 1) % 100 == 0:
|
||||
write_csv(root / f'{mode}_pairs.csv', pairs)
|
||||
print(f'{mode}: verified {i + 1}/{len(candidates)}', flush=True)
|
||||
write_csv(root / f'{mode}_pairs.csv', pairs)
|
||||
write_json(root / f'{mode}_nulls.json', nulls)
|
||||
write_json(root / f'{mode}_structure.json', structures)
|
||||
coverage.update({'objects_screened': len(objects), 'pairs_verified': len(pairs), 'timed_out': timeout,
|
||||
'screen_sample_elements': 2048, 'candidate_neighbors_per_metric': 10})
|
||||
write_json(root / f'{mode}_coverage.json', coverage)
|
||||
return pairs
|
||||
|
||||
|
||||
def summarize(mode, objects, pairs, model_bytes, root, partition_totals=None):
|
||||
rows = []
|
||||
scales = sorted({o['scale'] for o in objects})
|
||||
for scale in scales:
|
||||
ids = [i for i, o in enumerate(objects) if o['scale'] == scale]
|
||||
selected = set(ids)
|
||||
for threshold in THRESHOLDS:
|
||||
groups = sharing_groups(objects, [p for p in pairs if p['a'] in selected], threshold)
|
||||
groups = [g for g in groups if g['representative'] in selected]
|
||||
original = sum(objects[i]['bytes'] for i in ids)
|
||||
rep_bytes = sum(objects[g['representative']]['bytes'] for g in groups)
|
||||
# Concrete theoretical layout: 16 bytes per object mapping, plus 64 bytes per shared group.
|
||||
# Identity case falls back to original checkpoint at zero additional cost.
|
||||
mapping_bytes = 16 * len(ids) + 64 * sum(len(g['members']) > 1 for g in groups)
|
||||
proposed = rep_bytes + mapping_bytes
|
||||
saved = max(0, original - proposed)
|
||||
partition_count = partition_totals[scale]['blocks'] if partition_totals else len(ids)
|
||||
rows.append({'mode': mode, 'scale': scale, 'threshold': threshold,
|
||||
'objects': len(ids), 'groups': len(groups),
|
||||
'total_partition_objects': partition_count,
|
||||
'unexamined_partition_objects': partition_count - len(ids),
|
||||
'conservative_partition_groups': partition_count - len(ids) + len(groups),
|
||||
'shared_groups': sum(len(g['members']) > 1 for g in groups),
|
||||
'max_group_size': max([len(g['members']) for g in groups] or [0]),
|
||||
'verified_shared_bytes': sum(sum(objects[i]['bytes'] for i in g['members']) for g in groups if len(g['members']) > 1),
|
||||
'examined_object_bytes': original, 'representative_bytes': rep_bytes,
|
||||
'mapping_bytes_if_encoded': mapping_bytes, 'net_saved_bytes': saved,
|
||||
'net_saved_model_fraction': saved / model_bytes,
|
||||
'max_replacement_error': max([g['max_error'] for g in groups] or [0.])})
|
||||
write_json(root / f'groups_{mode}_{scale}_{threshold:.2f}.json', groups)
|
||||
return rows
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', type=pathlib.Path, required=True)
|
||||
p.add_argument('--seconds', type=int, default=7200)
|
||||
p.add_argument('--seed', type=int, default=20261109)
|
||||
args = p.parse_args()
|
||||
started = time.monotonic()
|
||||
deadline = started + args.seconds
|
||||
root = args.root
|
||||
output = root / 'analysis'
|
||||
output.mkdir(exist_ok=True)
|
||||
tensors = load_tensors(root / 'weights')
|
||||
inventory = []
|
||||
for t in tensors.values():
|
||||
if time.monotonic() > deadline:
|
||||
raise TimeoutError('Deadline during inventory')
|
||||
n = 0
|
||||
total = squared = 0.
|
||||
max_abs = 0.
|
||||
for x in t.chunked():
|
||||
if not np.isfinite(x).all():
|
||||
raise ValueError(f'Nonfinite weights: {t.name}')
|
||||
z = x.astype(np.float64)
|
||||
n += z.size
|
||||
total += float(z.sum())
|
||||
squared += float(z @ z)
|
||||
max_abs = max(max_abs, float(np.abs(z).max()))
|
||||
t.evict()
|
||||
inventory.append({'name': t.name, 'shape': str(t.shape), 'ndim': len(t.shape), 'dtype': t.dtype,
|
||||
'numel': t.numel, 'bytes': t.bytes, 'role': t.role, 'layer': t.layer,
|
||||
'mean': total / n, 'rms': math.sqrt(squared / n), 'max_abs': max_abs})
|
||||
t.evict()
|
||||
write_csv(output / 'inventory.csv', inventory)
|
||||
print(f'Inventory: {len(tensors)} tensors, {sum(t.bytes for t in tensors.values())} bytes', flush=True)
|
||||
whole = [object_record(t) for t in tensors.values()]
|
||||
# Stage budgets prevent the whole-tensor stage from consuming block-analysis time.
|
||||
whole_deadline = min(deadline, started + args.seconds * .50)
|
||||
whole_pairs = analyze_mode('tensor', whole, tensors, output, whole_deadline, 2000, args.seed)
|
||||
blocks, block_coverage, block_totals = build_blocks(tensors, args.seed, deadline)
|
||||
write_csv(output / 'block_sampling.csv', block_coverage)
|
||||
write_json(output / 'block_partitions.json', block_totals)
|
||||
block_pairs = analyze_mode('block', blocks, tensors, output, deadline, 20000, args.seed)
|
||||
model_bytes = sum(t.bytes for t in tensors.values())
|
||||
rows = summarize('tensor', whole, whole_pairs, model_bytes, output)
|
||||
rows += summarize('block', blocks, block_pairs, model_bytes, output, block_totals)
|
||||
write_csv(output / 'compression_curve.csv', rows)
|
||||
summary = {'model': json.loads((root / 'manifest.json').read_text())['repo'],
|
||||
'revision': json.loads((root / 'manifest.json').read_text())['revision'],
|
||||
'tensor_count': len(tensors), 'parameter_count': sum(t.numel for t in tensors.values()),
|
||||
'model_weight_bytes': model_bytes, 'ndim_counts': dict(collections.Counter(len(t.shape) for t in tensors.values())),
|
||||
'elapsed_seconds': time.monotonic() - started, 'analysis_budget_seconds': args.seconds,
|
||||
'peak_rss_kib': resource.getrusage(resource.RUSAGE_SELF).ru_maxrss,
|
||||
'python': platform.python_version(), 'numpy': np.__version__, 'seed': args.seed,
|
||||
'limitations': ['No inference or functional equivalence measured', 'No exhaustive pair search',
|
||||
'Block search uses per-role/shape/scale reservoirs of 256',
|
||||
'Different block scales overlap; savings must not be added',
|
||||
'Block savings count only verified disjoint sampled blocks, no extrapolation',
|
||||
'Spectral probes use aligned bounded subvolumes, not whole tensors',
|
||||
'Shared encoding is theoretical, not a runnable compressed checkpoint',
|
||||
'No cross-role grouping, neuron permutation alignment, or training'],
|
||||
'completed': True}
|
||||
write_json(output / 'summary.json', summary)
|
||||
print(json.dumps(summary), flush=True)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -0,0 +1,75 @@
|
||||
"""Audit saved numerical data and freeze provenance before archive/cleanup."""
|
||||
import argparse
|
||||
import csv
|
||||
import hashlib
|
||||
import json
|
||||
import math
|
||||
import os
|
||||
import pathlib
|
||||
import platform
|
||||
import re
|
||||
import shutil
|
||||
import time
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', required=True, type=pathlib.Path)
|
||||
p.add_argument('--analysis-code-commit', help='Commit actually used for the weight analysis; omit if unknown')
|
||||
args = p.parse_args()
|
||||
root = args.root.resolve()
|
||||
if not __debug__:
|
||||
raise RuntimeError('Validation requires Python assertions enabled')
|
||||
checked = {}
|
||||
for mode in ('tensor', 'block'):
|
||||
with (root / 'analysis' / f'{mode}_pairs.csv').open() as f:
|
||||
rows = list(csv.DictReader(f))
|
||||
coverage = json.loads((root / 'analysis' / f'{mode}_coverage.json').read_text())
|
||||
if len(rows) != coverage['pairs_verified']:
|
||||
raise ValueError('Coverage count mismatch')
|
||||
for row in rows:
|
||||
for key in ('cosine', 'pearson'):
|
||||
if row[key] and (not math.isfinite(float(row[key])) or abs(float(row[key])) > 1 + 1e-7):
|
||||
raise ValueError(f'Invalid correlation: {mode}/{key}')
|
||||
for key in ('error_member_to_rep', 'error_rep_to_member', 'symmetric_error', 'scaled_error',
|
||||
'sampled_error_q99_over_member_rms', 'screen_vs_exact_min_error_absolute_difference'):
|
||||
value = float(row[key])
|
||||
if not math.isfinite(value) or value < 0:
|
||||
raise ValueError(f'Invalid error: {mode}/{key}')
|
||||
if not math.isfinite(float(row['scale'])):
|
||||
raise ValueError('Invalid scaling diagnostic')
|
||||
checked[mode] = {'rows': len(rows), 'finite_metrics': True, 'coverage_matches': True}
|
||||
test_log = (root / 'tests.log').read_text()
|
||||
match = re.search(r'Ran (\d+) tests in ([\d.]+)s', test_log)
|
||||
if not match or '\nOK\n' not in test_log:
|
||||
raise ValueError('Final tests did not pass')
|
||||
qc = {'passed': True, 'checked_unix': time.time(), 'numerical_tables': checked,
|
||||
'final_tests': int(match[1]), 'final_test_seconds': float(match[2]),
|
||||
'earlier_parallel_test_timeout_seconds': 300 if (root / 'tests-during-io-timeout.txt').exists() else None}
|
||||
(root / 'numerical-qc.json').write_text(json.dumps(qc, indent=2))
|
||||
for src, dest in [('tests.log', 'test-results.txt'), ('analysis.log', 'analysis-run.txt'),
|
||||
('plot.log', 'plot-run.txt'), ('download-ranged.log', 'download-progress.txt'),
|
||||
('supervisor.log', 'supervisor-run.txt')]:
|
||||
shutil.copy2(root / src, root / dest)
|
||||
mem = {}
|
||||
for line in pathlib.Path('/proc/meminfo').read_text().splitlines():
|
||||
key, value = line.split(':', 1)
|
||||
if key in ('MemTotal', 'MemAvailable', 'SwapTotal', 'SwapFree'):
|
||||
mem[key + '_KiB'] = int(value.split()[0])
|
||||
environment = {'recorded_unix': time.time(), 'python': platform.python_version(),
|
||||
'platform': platform.platform(), 'machine': platform.machine(),
|
||||
'logical_cpus': os.cpu_count(), 'memory_snapshot': mem,
|
||||
'blas_threads': 1, 'assertions_enabled': __debug__,
|
||||
'analysis_io_priority': 'best-effort 7, applied during block verification',
|
||||
'analysis_code_commit': args.analysis_code_commit,
|
||||
'note': 'Shared host; see resource-observation.txt. Memory snapshot is post-analysis.'}
|
||||
(root / 'execution-environment.json').write_text(json.dumps(environment, indent=2))
|
||||
project = pathlib.Path(__file__).resolve().parents[1]
|
||||
files = sorted((project / 'pilot').glob('*.py')) + [project / 'requirements.txt', project / 'tests/test_pilot.py']
|
||||
hashes = [f'{hashlib.sha256(file.read_bytes()).hexdigest()} {file.relative_to(project)}' for file in files]
|
||||
(root / 'code_checksums.txt').write_text('\n'.join(hashes) + '\n')
|
||||
print(json.dumps(qc))
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -0,0 +1,140 @@
|
||||
"""Download pinned official safetensors, no framework or shared cache."""
|
||||
import argparse
|
||||
import concurrent.futures
|
||||
import hashlib
|
||||
import json
|
||||
import pathlib
|
||||
import os
|
||||
import shutil
|
||||
import time
|
||||
import threading
|
||||
import urllib.parse
|
||||
import urllib.request
|
||||
|
||||
|
||||
def fetch_json(url):
|
||||
with urllib.request.urlopen(url, timeout=90) as response:
|
||||
return json.load(response)
|
||||
|
||||
|
||||
def ranged_download(url, part, total, workers=8):
|
||||
"""Resume a sequential prefix; journal completed byte ranges before final hashing."""
|
||||
journal = part.with_suffix(part.suffix + '.ranges.json')
|
||||
if journal.exists():
|
||||
state = json.loads(journal.read_text())
|
||||
if state['total'] != total:
|
||||
raise RuntimeError('Range journal size mismatch')
|
||||
else:
|
||||
state = {'total': total, 'prefix': part.stat().st_size if part.exists() else 0, 'done': []}
|
||||
journal.write_text(json.dumps(state))
|
||||
lock = threading.Lock()
|
||||
fd = os.open(part, os.O_CREAT | os.O_RDWR, 0o600)
|
||||
chunk_size = 64 * 1024 * 1024
|
||||
ranges = [(start, min(start + chunk_size, total) - 1)
|
||||
for start in range(state['prefix'], total, chunk_size)]
|
||||
|
||||
def one(bounds):
|
||||
start, end = bounds
|
||||
if start in state['done']:
|
||||
return
|
||||
for attempt in range(3):
|
||||
try:
|
||||
req = urllib.request.Request(url, headers={'Range': f'bytes={start}-{end}'})
|
||||
with urllib.request.urlopen(req, timeout=180) as response:
|
||||
expected = f'bytes {start}-{end}/{total}'
|
||||
if response.status != 206 or response.headers.get('Content-Range') != expected:
|
||||
raise RuntimeError('Server did not honor exact requested range')
|
||||
offset = start
|
||||
while chunk := response.read(min(4 * 1024 * 1024, end - offset + 1)):
|
||||
written = os.pwrite(fd, chunk, offset)
|
||||
if written != len(chunk):
|
||||
raise RuntimeError('Short positional write')
|
||||
offset += written
|
||||
if offset > end:
|
||||
break
|
||||
if offset != end + 1:
|
||||
raise RuntimeError('Incomplete range')
|
||||
with lock:
|
||||
os.fsync(fd)
|
||||
state['done'].append(start)
|
||||
tmp = journal.with_suffix('.tmp')
|
||||
tmp.write_text(json.dumps(state))
|
||||
tmp.replace(journal)
|
||||
done = state['prefix'] + sum(b - a + 1 for a, b in ranges if a in state['done'])
|
||||
print(f'Weight download {done}/{total} bytes ({100 * done / total:.1f}%)', flush=True)
|
||||
return
|
||||
except Exception:
|
||||
if attempt == 2:
|
||||
raise
|
||||
time.sleep(2 ** attempt)
|
||||
try:
|
||||
with concurrent.futures.ThreadPoolExecutor(max_workers=workers) as pool:
|
||||
list(pool.map(one, ranges))
|
||||
finally:
|
||||
os.close(fd)
|
||||
journal.unlink()
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', type=pathlib.Path, required=True)
|
||||
p.add_argument('--repo', default='Qwen/Qwen3.5-2B')
|
||||
args = p.parse_args()
|
||||
root = args.root.resolve()
|
||||
root.mkdir(parents=True, exist_ok=True)
|
||||
weights = root / 'weights'
|
||||
weights.mkdir(exist_ok=True)
|
||||
marker = weights / '.pilot-owned'
|
||||
marker.write_text(args.repo)
|
||||
old_manifest = root / 'manifest.json'
|
||||
pinned = json.loads(old_manifest.read_text())['revision'] if old_manifest.exists() else None
|
||||
endpoint = f'https://huggingface.co/api/models/{args.repo}'
|
||||
if pinned:
|
||||
endpoint += f'/revision/{pinned}'
|
||||
info = fetch_json(endpoint + '?blobs=true')
|
||||
revision = info['sha']
|
||||
files = [f for f in info['siblings'] if f['rfilename'].endswith('.safetensors')
|
||||
or f['rfilename'] in ('config.json', 'model.safetensors.index.json', 'README.md')]
|
||||
total = sum(f.get('size', 0) for f in files)
|
||||
if shutil.disk_usage(root).free < total + 3_000_000_000:
|
||||
raise RuntimeError('Insufficient space for checkpoint plus 3 GB margin')
|
||||
manifest = {'repo': args.repo, 'revision': revision, 'expected_bytes': total,
|
||||
'started_unix': time.time(), 'files': [], 'license': info.get('cardData', {}).get('license')}
|
||||
(root / 'source_metadata.json').write_text(json.dumps(info, indent=2))
|
||||
for entry in files:
|
||||
name = entry['rfilename']
|
||||
if pathlib.PurePosixPath(name).name != name:
|
||||
raise RuntimeError(f'Unexpected nested path: {name}')
|
||||
dest = weights / name
|
||||
part = dest.with_suffix(dest.suffix + '.partial')
|
||||
digest = hashlib.sha256()
|
||||
url = f'https://huggingface.co/{args.repo}/resolve/{revision}/{urllib.parse.quote(name)}?download=true'
|
||||
print(f'Downloading {name}, expected {entry.get("size")} bytes', flush=True)
|
||||
if entry.get('size', 0) > 128 * 1024 * 1024:
|
||||
ranged_download(url, part, entry['size'])
|
||||
with part.open('rb') as stream:
|
||||
while chunk := stream.read(4 * 1024 * 1024):
|
||||
digest.update(chunk)
|
||||
else:
|
||||
with urllib.request.urlopen(url, timeout=180) as response, part.open('wb') as out:
|
||||
while chunk := response.read(4 * 1024 * 1024):
|
||||
out.write(chunk)
|
||||
digest.update(chunk)
|
||||
sha = digest.hexdigest()
|
||||
if entry.get('size') and part.stat().st_size != entry['size']:
|
||||
raise RuntimeError(f'Size mismatch: {name}')
|
||||
expected_sha = entry.get('lfs', {}).get('sha256')
|
||||
if expected_sha and sha != expected_sha:
|
||||
raise RuntimeError(f'SHA256 mismatch: {name}')
|
||||
part.rename(dest)
|
||||
manifest['files'].append({'name': name, 'bytes': dest.stat().st_size,
|
||||
'sha256': sha, 'upstream_sha256': expected_sha})
|
||||
(root / 'manifest.json').write_text(json.dumps(manifest, indent=2))
|
||||
manifest['completed_unix'] = time.time()
|
||||
(root / 'manifest.json').write_text(json.dumps(manifest, indent=2))
|
||||
shutil.copy2(weights / 'config.json', root / 'model_config.json')
|
||||
print(json.dumps({'revision': revision, 'bytes': sum(f['bytes'] for f in manifest['files'])}), flush=True)
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
+261
@@ -0,0 +1,261 @@
|
||||
"""Scientific figures and a compact result report from saved data only."""
|
||||
import argparse
|
||||
import csv
|
||||
import json
|
||||
import pathlib
|
||||
|
||||
import matplotlib
|
||||
matplotlib.use('Agg')
|
||||
import matplotlib.pyplot as plt
|
||||
from matplotlib.patches import Patch, Rectangle
|
||||
import numpy as np
|
||||
|
||||
|
||||
def read_csv(path):
|
||||
with path.open() as f:
|
||||
return list(csv.DictReader(f))
|
||||
|
||||
|
||||
def write_csv(path, rows):
|
||||
with path.open('w', newline='') as f:
|
||||
if rows:
|
||||
writer = csv.DictWriter(f, fieldnames=list(rows[0]))
|
||||
writer.writeheader()
|
||||
writer.writerows(rows)
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', type=pathlib.Path, required=True)
|
||||
args = p.parse_args()
|
||||
root = args.root
|
||||
out = root / 'analysis'
|
||||
figures = root / 'figures'
|
||||
figures.mkdir(exist_ok=True)
|
||||
plt.rcParams.update({'font.size': 10, 'axes.spines.top': False, 'axes.spines.right': False,
|
||||
'pdf.fonttype': 42, 'svg.fonttype': 'none', 'figure.dpi': 120})
|
||||
def save(fig, name):
|
||||
fig.tight_layout()
|
||||
for ext in ('png', 'pdf', 'svg'):
|
||||
fig.savefig(figures / f'{name}.{ext}', dpi=200, bbox_inches='tight')
|
||||
plt.close(fig)
|
||||
|
||||
curves = read_csv(out / 'compression_curve.csv')
|
||||
summary = json.loads((out / 'summary.json').read_text())
|
||||
dimensional = {}
|
||||
for row in read_csv(out / 'inventory.csv'):
|
||||
ndim = int(row['ndim'])
|
||||
item = dimensional.setdefault(ndim, {'ndim': ndim, 'tensor_count': 0, 'weight_bytes': 0})
|
||||
item['tensor_count'] += 1
|
||||
item['weight_bytes'] += int(row['bytes'])
|
||||
dimensional_rows = [dimensional[key] for key in sorted(dimensional)]
|
||||
for item in dimensional_rows:
|
||||
item['model_byte_fraction'] = item['weight_bytes'] / summary['model_weight_bytes']
|
||||
with (out / 'dimensional_bytes.csv').open('w', newline='') as f:
|
||||
writer = csv.DictWriter(f, fieldnames=list(dimensional_rows[0]))
|
||||
writer.writeheader()
|
||||
writer.writerows(dimensional_rows)
|
||||
fig, axes = plt.subplots(1, 2, figsize=(11, 4))
|
||||
for scale in sorted({r['scale'] for r in curves}):
|
||||
rows = [r for r in curves if r['scale'] == scale]
|
||||
x = [100 * float(r['threshold']) for r in rows]
|
||||
axes[0].plot(x, [100 * float(r['net_saved_model_fraction']) for r in rows], 'o-', label=scale)
|
||||
axes[1].plot(x, [int(r['shared_groups']) for r in rows], 'o-', label=scale)
|
||||
axes[0].set(xlabel='Maximum relative weight error (%)', ylabel='Verified net savings / model bytes (%)')
|
||||
axes[1].set(xlabel='Maximum relative weight error (%)', ylabel='Non-singleton sharing groups')
|
||||
axes[0].legend(fontsize=8, loc='upper left')
|
||||
fig.suptitle('Conservative verified sharing; block scales are separate alternatives')
|
||||
save(fig, 'compression_frontier')
|
||||
|
||||
pair_sets = {mode: read_csv(out / f'{mode}_pairs.csv') for mode in ('tensor', 'block')}
|
||||
role_rows = []
|
||||
for mode, pairs in pair_sets.items():
|
||||
mode_objects = json.loads((out / f'{mode}_objects.json').read_text())
|
||||
buckets = {}
|
||||
for row in pairs:
|
||||
obj = mode_objects[int(row['a'])]
|
||||
key = (obj['role'], obj['scale'], tuple(obj['shape']), obj['dtype'])
|
||||
buckets.setdefault(key, []).append(row)
|
||||
for (role_name, scale_name, shape, dtype), rows in sorted(buckets.items(), key=lambda item: str(item[0])):
|
||||
errors = [float(r['error_member_to_rep']) for r in rows]
|
||||
cosines = [float(r['cosine']) for r in rows if r['cosine']]
|
||||
role_rows.append({'mode': mode, 'role': role_name, 'scale': scale_name,
|
||||
'shape': str(shape), 'ndim': len(shape), 'dtype': dtype,
|
||||
'verified_pairs': len(rows), 'minimum_direct_error': min(errors),
|
||||
'median_direct_error': float(np.median(errors)),
|
||||
'maximum_cosine': max(cosines) if cosines else '',
|
||||
'minimum_scaled_error': min(float(r['scaled_error']) for r in rows),
|
||||
'pairs_under_5pct_direct_error': sum(e <= .05 for e in errors),
|
||||
'pairs_under_20pct_direct_error': sum(e <= .20 for e in errors)})
|
||||
with (out / 'role_summary.csv').open('w', newline='') as f:
|
||||
if role_rows:
|
||||
writer = csv.DictWriter(f, fieldnames=list(role_rows[0]))
|
||||
writer.writeheader()
|
||||
writer.writerows(role_rows)
|
||||
fig, axes = plt.subplots(2, 2, figsize=(11, 7))
|
||||
for row, (mode, pairs) in enumerate(pair_sets.items()):
|
||||
for scale in sorted({r['scale_name'] for r in pairs}):
|
||||
group = [r for r in pairs if r['scale_name'] == scale]
|
||||
axes[row, 0].hist([float(r['cosine']) for r in group if r['cosine']], bins=np.linspace(-1, 1, 46), histtype='step', label=scale)
|
||||
errors = [float(r['error_member_to_rep']) for r in group]
|
||||
axes[row, 1].hist(errors, bins=np.linspace(0, max(1e-6, max(errors, default=0.) * 1.01), 46), histtype='step', label=scale)
|
||||
axes[row, 0].set(title=f'{mode}: exact cosine of screened candidates', xlabel='Cosine', ylabel='Candidate pairs')
|
||||
axes[row, 1].set(title=f'{mode}: exact direct replacement error', xlabel='Relative Frobenius error', ylabel='Candidate pairs')
|
||||
axes[row, 0].legend(fontsize=7, loc='upper right')
|
||||
save(fig, 'candidate_distributions')
|
||||
|
||||
pairs = pair_sets['tensor']
|
||||
roles = sorted({r['role'] for r in pairs})
|
||||
# Illustrate one repeated 2D projection; this is not a byte-weighted model summary.
|
||||
objects = json.loads((out / 'tensor_objects.json').read_text())
|
||||
candidates = [role for role in roles if any(o['role'] == role and len(o['shape']) == 2 for o in objects)]
|
||||
role = max(candidates or roles, key=lambda k: sum(r['role'] == k for r in pairs))
|
||||
chosen = [(i, o) for i, o in enumerate(objects) if o['role'] == role]
|
||||
chosen.sort(key=lambda item: item[1]['layer'])
|
||||
indices = {i: j for j, (i, _) in enumerate(chosen)}
|
||||
matrix = np.full((len(chosen), len(chosen)), np.nan)
|
||||
np.fill_diagonal(matrix, 1.)
|
||||
for r in pairs:
|
||||
a, b = int(r['a']), int(r['b'])
|
||||
if a in indices and b in indices and r['cosine']:
|
||||
matrix[indices[a], indices[b]] = matrix[indices[b], indices[a]] = float(r['cosine'])
|
||||
fig, ax = plt.subplots(figsize=(7, 6))
|
||||
cmap = plt.get_cmap('coolwarm').with_extremes(bad='white')
|
||||
im = ax.imshow(matrix, cmap=cmap, vmin=-1, vmax=1)
|
||||
for y, x in zip(*np.where(np.isnan(matrix))):
|
||||
ax.add_patch(Rectangle((x - .5, y - .5), 1, 1, facecolor='none',
|
||||
edgecolor='#999999', hatch='///', linewidth=0))
|
||||
labels = [o['layer'] for _, o in chosen]
|
||||
ticks = np.arange(0, len(labels), max(1, len(labels) // 10))
|
||||
ax.set_xticks(ticks, [labels[i] for i in ticks])
|
||||
ax.set_yticks(ticks, [labels[i] for i in ticks])
|
||||
ax.set(xlabel='Layer', ylabel='Layer', title=role.removeprefix('model.') + '\nHatching = unverified; diagonal is identity')
|
||||
ax.legend(handles=[Patch(facecolor='white', edgecolor='#999999', hatch='///', label='Unverified pair')],
|
||||
loc='upper center', bbox_to_anchor=(.5, -.12), fontsize=8, frameon=False)
|
||||
fig.colorbar(im, ax=ax, label='Exact cosine')
|
||||
save(fig, 'layer_heatmap')
|
||||
heatmap_rows = [{'a_layer': oa['layer'], 'b_layer': ob['layer'], 'role': role,
|
||||
'cosine': float(matrix[i, j]) if not np.isnan(matrix[i, j]) else '',
|
||||
'verified': bool(i != j and not np.isnan(matrix[i, j])), 'identity': i == j}
|
||||
for i, (_, oa) in enumerate(chosen) for j, (_, ob) in enumerate(chosen)]
|
||||
write_csv(out / 'heatmap_pairs.csv', heatmap_rows)
|
||||
|
||||
distance_rows = []
|
||||
fig, axes = plt.subplots(1, 2, figsize=(12, 4), sharey=True)
|
||||
for ax, mode in zip(axes, ('tensor', 'block')):
|
||||
mode_objects = json.loads((out / f'{mode}_objects.json').read_text())
|
||||
buckets = {}
|
||||
for row in pair_sets[mode]:
|
||||
if not row['cosine']:
|
||||
continue
|
||||
obj = mode_objects[int(row['a'])]
|
||||
stratum = f"{len(obj['shape'])}D" if mode == 'tensor' else obj['scale']
|
||||
buckets.setdefault(stratum, {}).setdefault(int(row['layer_distance']), []).append(float(row['cosine']))
|
||||
for stratum, by_distance in sorted(buckets.items()):
|
||||
distances = sorted(by_distance)
|
||||
medians = []
|
||||
for distance in distances:
|
||||
values = by_distance[distance]
|
||||
median = float(np.median(values))
|
||||
medians.append(median)
|
||||
distance_rows.append({'mode': mode, 'stratum': stratum, 'layer_distance': distance,
|
||||
'pairs': len(values), 'median_cosine': median,
|
||||
'q25': float(np.quantile(values, .25)), 'q75': float(np.quantile(values, .75))})
|
||||
ax.plot(distances, medians, 'o-', markersize=3, label=stratum)
|
||||
ax.set(xlabel='Absolute layer distance', ylabel='Median exact cosine',
|
||||
title=f'{mode}: dimension / scale strata', ylim=(-.1, 1.05))
|
||||
ax.legend(fontsize=8, loc='upper center', bbox_to_anchor=(.5, -.23), ncol=2, frameon=False)
|
||||
fig.suptitle('Screened candidates: module mixture and selection bias still apply')
|
||||
write_csv(out / 'layer_distance_stratified.csv', distance_rows)
|
||||
save(fig, 'layer_distance')
|
||||
|
||||
null_rows = []
|
||||
fig, axes = plt.subplots(3, 2, figsize=(12, 10))
|
||||
bins = np.linspace(-1, 1, 26)
|
||||
for col, mode in enumerate(('tensor', 'block')):
|
||||
mode_objects = json.loads((out / f'{mode}_objects.json').read_text())
|
||||
nulls = json.loads((out / f'{mode}_nulls.json').read_text())
|
||||
real_random = json.loads((out / f'{mode}_random_pairs.json').read_text())
|
||||
cohorts = [('shuffled', [r for r in nulls if r['method'] == 'shuffled']),
|
||||
('gaussian_norm_matched', [r for r in nulls if r['method'] == 'gaussian_norm_matched']),
|
||||
('random_actual_same_role', real_random)]
|
||||
for row, ndim in enumerate((1, 2, 3)):
|
||||
ax = axes[row, col]
|
||||
any_values = False
|
||||
for method, records in cohorts:
|
||||
values = np.asarray([r['cosine'] for r in records if r['cosine'] is not None
|
||||
and len(mode_objects[r['a']]['shape']) == ndim])
|
||||
if not len(values):
|
||||
continue
|
||||
any_values = True
|
||||
values = np.clip(values, -1, 1)
|
||||
ax.hist(values, bins=bins, density=True, histtype='step', label=f'{method} (n={len(values)})')
|
||||
counts, _ = np.histogram(values, bins=bins)
|
||||
for i, count in enumerate(counts):
|
||||
null_rows.append({'mode': mode, 'ndim': ndim, 'method': method, 'samples': len(values),
|
||||
'bin_left': float(bins[i]), 'bin_right': float(bins[i + 1]),
|
||||
'count': int(count), 'density': float(count / (len(values) * (bins[i + 1] - bins[i])))})
|
||||
ax.set(title=f'{mode}: {ndim}D sampled controls', xlabel='Sampled cosine', ylabel='Density')
|
||||
if any_values:
|
||||
ax.legend(fontsize=7, loc='upper right')
|
||||
else:
|
||||
ax.text(.5, .5, 'No eligible samples', transform=ax.transAxes, ha='center')
|
||||
fig.suptitle('Dimension-stratified controls: density normalized; role mixtures may differ')
|
||||
write_csv(out / 'null_histograms.csv', null_rows)
|
||||
save(fig, 'null_controls')
|
||||
|
||||
fig, axes = plt.subplots(1, 2, figsize=(11, 4))
|
||||
structure = json.loads((out / 'block_structure.json').read_text())
|
||||
for ax, ndim in zip(axes, (2, 3)):
|
||||
probes = [m for pair in structure for m in pair['modes'] if len(m['probe_shape']) == ndim and not m.get('trivial_mode_skipped')]
|
||||
for axis in sorted({m['axis'] for m in probes}):
|
||||
selected = [m for m in probes if m['axis'] == axis]
|
||||
ax.scatter([m['normalized_spectrum_distance'] for m in selected],
|
||||
[m['left_subspace_overlap'] for m in selected], s=18, alpha=.65, label=f'mode {axis}')
|
||||
ax.set(title=f'{ndim}D blocks: bounded mode-unfolding probes', xlabel='Normalized spectrum distance', ylabel='Leading left subspace overlap (k <= 4)')
|
||||
if probes:
|
||||
ax.legend(fontsize=8, loc='upper right')
|
||||
else:
|
||||
ax.text(.5, .5, 'No eligible probes', transform=ax.transAxes, ha='center')
|
||||
save(fig, 'block_structure')
|
||||
|
||||
parts = json.loads((out / 'block_partitions.json').read_text())
|
||||
lines = ['# Qwen3.5-2B parameter similarity pilot', '',
|
||||
f"- Revision: `{summary['revision']}`", f"- Stored parameters: {summary['parameter_count']:,}",
|
||||
f"- Weight bytes: {summary['model_weight_bytes']:,}",
|
||||
f"- Tensor dimensionalities: {summary['ndim_counts']}",
|
||||
f"- Analysis and grouping/export wall time: {summary['elapsed_seconds']:.1f} seconds",
|
||||
f"- Pair-comparison cutoff: {summary['analysis_budget_seconds']} seconds; final grouping/export is additional and remains under the supervisor hard limit.",
|
||||
f"- Peak analysis RSS: {summary['peak_rss_kib'] / 1024:.1f} MiB", '',
|
||||
'## Verified sharing estimates', '',
|
||||
'| Representation | Weight-error limit | Verified sharing groups | Conservative total groups* | Largest group | Net model savings |',
|
||||
'|---|---:|---:|---:|---:|---:|---:|']
|
||||
for r in curves:
|
||||
lines.append(f"| {r['scale']} | {float(r['threshold']):.0%} | {r['shared_groups']} | {r['conservative_partition_groups']} | {r['max_group_size']} | {float(r['net_saved_model_fraction']):.6%} |")
|
||||
lines += ['', '*Conservative total groups counts unexamined full blocks as singletons; boundary regions and ineligible tensors stay unchanged and are outside that block count.',
|
||||
'Savings count only verified replacement edges. Unsampled blocks remain unchanged. Different block scales must not be summed.',
|
||||
'Theoretical encoding: one stored representative per group, 16 bytes per object mapping and 64 bytes per shared-group descriptor; identity fallback if larger.',
|
||||
'Exact grouping uses direct replacement only. Scaled and spectral results are diagnostic.', '',
|
||||
'## Stratified diagnostics', '',
|
||||
'See `analysis/role_summary.csv` for same-role, same-shape statistics, separated by tensor dimensionality and block scale.',
|
||||
'See `analysis/dimensional_bytes.csv` for tensor counts versus weight-byte contributions by dimensionality.',
|
||||
'Heatmap missing pairs are hatched (`heatmap_pairs.csv`); identity entries are not empirical matches.',
|
||||
'Layer-distance medians are split by dimension/scale (`layer_distance_stratified.csv`); role and selection bias remain.',
|
||||
'Controls are split by dimension and density-normalized, with sample counts (`null_histograms.csv`); these are not significance tests.',
|
||||
'These are screened-candidate statistics, not estimates of all possible pairs. Small normalization vectors must not be conflated with large matrices.', '',
|
||||
'## Coverage', '']
|
||||
for mode in ('tensor', 'block'):
|
||||
coverage = json.loads((out / f'{mode}_coverage.json').read_text())
|
||||
lines.append(f'- {mode}: `{json.dumps(coverage)}`')
|
||||
lines += [f'- Complete-block partition totals (per alternative): `{json.dumps(parts)}`', '',
|
||||
'## Interpretation limits', '', *['- ' + s for s in summary['limitations']],
|
||||
'- Candidate histograms are selection-biased and do not represent the distribution of all tensor pairs.',
|
||||
'- 1D normalization weights may cluster strongly but carry few bytes; do not infer material compression from group counts.',
|
||||
'- No task accuracy, perplexity, inference speed or runtime-memory improvement has been measured.',
|
||||
'- Stored parameter counts may omit tied aliases; no duplicate-weight equality search is performed.', '',
|
||||
'## Figures', '', *[f'' for name in ['compression_frontier', 'candidate_distributions', 'layer_heatmap', 'layer_distance', 'null_controls', 'block_structure']]]
|
||||
(root / 'REPORT.md').write_text('\n'.join(lines) + '\n')
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -0,0 +1,58 @@
|
||||
"""Supervise the one-model job; no automatic download of subsequent models."""
|
||||
import argparse
|
||||
import json
|
||||
import os
|
||||
import pathlib
|
||||
import subprocess
|
||||
import sys
|
||||
import time
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', type=pathlib.Path, required=True)
|
||||
args = p.parse_args()
|
||||
root = args.root.resolve()
|
||||
root.mkdir(exist_ok=True, parents=True)
|
||||
status_file = root / 'status.json'
|
||||
start = time.time()
|
||||
env = dict(os.environ, OPENBLAS_NUM_THREADS='1', OMP_NUM_THREADS='1')
|
||||
|
||||
def status(phase, **extra):
|
||||
status_file.write_text(json.dumps({'phase': phase, 'updated_unix': time.time(),
|
||||
'supervisor_started_unix': start, **extra}, indent=2))
|
||||
print(phase, flush=True)
|
||||
|
||||
def command(module, log, timeout, extra=()):
|
||||
with (root / log).open('w') as stream:
|
||||
subprocess.run([sys.executable, '-B', '-m', module, '--root', str(root), *extra],
|
||||
check=True, timeout=timeout, env=env, stdout=stream, stderr=subprocess.STDOUT)
|
||||
|
||||
try:
|
||||
status('waiting_for_download')
|
||||
download_wait_start = time.monotonic()
|
||||
while True:
|
||||
manifest_path = root / 'manifest.json'
|
||||
if manifest_path.exists():
|
||||
manifest = json.loads(manifest_path.read_text())
|
||||
if manifest.get('completed_unix') and all((root / 'weights' / f['name']).exists() for f in manifest['files']):
|
||||
break
|
||||
if time.monotonic() - download_wait_start > 7200:
|
||||
raise TimeoutError('Download wait exceeded 2 hours; no analysis started')
|
||||
time.sleep(10)
|
||||
status('analyzing', analysis_hard_limit_seconds=7200)
|
||||
command('pilot.analyze', 'analysis.log', 7200, ('--seconds', '6600'))
|
||||
status('plotting')
|
||||
command('pilot.plot', 'plot.log', 300)
|
||||
status('validating')
|
||||
command('pilot.validate', 'validate.log', 120)
|
||||
with (root / 'environment.txt').open('w') as f:
|
||||
subprocess.run([sys.executable, '-B', '-m', 'pip', 'freeze'], check=True, stdout=f, env=env)
|
||||
status('results_ready_weights_retained_until_archive_copy')
|
||||
except Exception as e:
|
||||
status('failed', error_type=type(e).__name__, error=str(e))
|
||||
raise
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -0,0 +1,82 @@
|
||||
"""Read safetensors BF16 directly; never instantiate a model or use pickle."""
|
||||
import json
|
||||
import math
|
||||
import mmap
|
||||
import re
|
||||
import struct
|
||||
|
||||
import numpy as np
|
||||
|
||||
DTYPES = {'BF16': ('<u2', 2), 'F16': ('<f2', 2), 'F32': ('<f4', 4), 'F64': ('<f8', 8)}
|
||||
|
||||
|
||||
def decode(raw, dtype):
|
||||
if dtype == 'BF16':
|
||||
return (np.asarray(raw).astype(np.uint32) << 16).view(np.float32)
|
||||
return np.asarray(raw, dtype=np.float32)
|
||||
|
||||
|
||||
class Tensor:
|
||||
def __init__(self, file, base, name, meta):
|
||||
self.name, self.shape, self.dtype = name, tuple(meta['shape']), meta['dtype']
|
||||
self.numel = math.prod(self.shape)
|
||||
self.bytes = meta['data_offsets'][1] - meta['data_offsets'][0]
|
||||
code, itemsize = DTYPES[self.dtype]
|
||||
if self.bytes != self.numel * itemsize:
|
||||
raise ValueError(f'Invalid byte length: {name}')
|
||||
self.raw = np.memmap(file, mode='r', dtype=code,
|
||||
offset=base + meta['data_offsets'][0], shape=(self.numel,))
|
||||
self.role = re.sub(r'\.(?:layers|blocks)\.\d+', '.layers.*', name)
|
||||
self.role = re.sub(r'\.experts\.\d+', '.experts.*', self.role)
|
||||
m = re.search(r'\.(?:layers|blocks)\.(\d+)', name)
|
||||
self.layer = int(m[1]) if m else -1
|
||||
|
||||
def read(self, selector=slice(None)):
|
||||
return decode(self.raw[selector], self.dtype)
|
||||
|
||||
def chunked(self, size=262144):
|
||||
for start in range(0, self.numel, size):
|
||||
yield self.read(slice(start, min(start + size, self.numel)))
|
||||
|
||||
def evict(self):
|
||||
if hasattr(self.raw._mmap, 'madvise'):
|
||||
self.raw._mmap.madvise(mmap.MADV_DONTNEED)
|
||||
|
||||
|
||||
def load_tensors(directory):
|
||||
tensors = {}
|
||||
for file in sorted(directory.glob('*.safetensors')):
|
||||
with file.open('rb') as f:
|
||||
size = struct.unpack('<Q', f.read(8))[0]
|
||||
if size > 100_000_000:
|
||||
raise ValueError('Oversized safetensors header')
|
||||
header = json.loads(f.read(size))
|
||||
for name, meta in header.items():
|
||||
if name == '__metadata__':
|
||||
continue
|
||||
if name in tensors:
|
||||
raise ValueError(f'Duplicate tensor name: {name}')
|
||||
tensors[name] = Tensor(file, 8 + size, name, meta)
|
||||
if not tensors:
|
||||
raise ValueError('No safetensors found')
|
||||
return tensors
|
||||
|
||||
|
||||
def metrics(x, y):
|
||||
"""x is member; y is representative. Float64 accumulation."""
|
||||
x, y = x.astype(np.float64), y.astype(np.float64)
|
||||
xx, yy, xy = float(x @ x), float(y @ y), float(x @ y)
|
||||
delta = float((x - y) @ (x - y))
|
||||
return finish(xx, yy, xy, delta, float(x.sum()), float(y.sum()), x.size)
|
||||
|
||||
|
||||
def finish(xx, yy, xy, delta, sx, sy, n):
|
||||
tiny = 1e-30
|
||||
vx, vy = max(0., xx - sx * sx / n), max(0., yy - sy * sy / n)
|
||||
return {'cosine': xy / np.sqrt(xx * yy) if xx * yy > tiny else None,
|
||||
'pearson': (xy - sx * sy / n) / np.sqrt(vx * vy) if vx * vy > tiny else None,
|
||||
'error_member_to_rep': np.sqrt(delta / max(xx, tiny)),
|
||||
'error_rep_to_member': np.sqrt(delta / max(yy, tiny)),
|
||||
'symmetric_error': np.sqrt(2 * delta / max(xx + yy, tiny)),
|
||||
'scaled_error': np.sqrt(max(0., xx - xy * xy / max(yy, tiny)) / max(xx, tiny)),
|
||||
'scale': xy / max(yy, tiny)}
|
||||
@@ -0,0 +1,85 @@
|
||||
"""Validate archived analysis before allowing deletion of downloaded weights."""
|
||||
import argparse
|
||||
import csv
|
||||
import hashlib
|
||||
import json
|
||||
import pathlib
|
||||
import shutil
|
||||
import time
|
||||
|
||||
|
||||
def validate(root):
|
||||
analysis = root / 'analysis'
|
||||
summary = json.loads((analysis / 'summary.json').read_text())
|
||||
assert summary['completed']
|
||||
manifest = json.loads((root / 'manifest.json').read_text())
|
||||
assert summary['revision'] == manifest['revision']
|
||||
for mode in ('tensor', 'block'):
|
||||
objects = json.loads((analysis / f'{mode}_objects.json').read_text())
|
||||
with (analysis / f'{mode}_pairs.csv').open() as f:
|
||||
pairs = list(csv.DictReader(f))
|
||||
coverage = json.loads((analysis / f'{mode}_coverage.json').read_text())
|
||||
assert len(pairs) == coverage['pairs_verified']
|
||||
edges = {}
|
||||
for p in pairs:
|
||||
a, b = int(p['a']), int(p['b'])
|
||||
assert objects[a]['shape'] == objects[b]['shape']
|
||||
assert objects[a]['role'] == objects[b]['role']
|
||||
edges[(b, a)] = float(p['error_member_to_rep'])
|
||||
edges[(a, b)] = float(p['error_rep_to_member'])
|
||||
for file in analysis.glob(f'groups_{mode}_*.json'):
|
||||
threshold = float(file.stem.rsplit('_', 1)[1])
|
||||
groups = json.loads(file.read_text())
|
||||
members = [i for g in groups for i in g['members']]
|
||||
assert len(set(members)) == len(members), file
|
||||
for g in groups:
|
||||
assert g['representative'] in g['members']
|
||||
for member in g['members']:
|
||||
if member != g['representative']:
|
||||
assert edges[(g['representative'], member)] <= threshold + 1e-12
|
||||
assert g['max_error'] <= threshold + 1e-12
|
||||
assert (root / 'REPORT.md').is_file()
|
||||
images = list((root / 'figures').glob('*.png'))
|
||||
assert len(images) >= 5
|
||||
for file in images:
|
||||
assert file.read_bytes()[:8] == b'\x89PNG\r\n\x1a\n'
|
||||
checksums = {}
|
||||
for file in sorted(root.rglob('*')):
|
||||
if (file.is_file() and 'weights' not in file.relative_to(root).parts
|
||||
and file.name not in ('checksums.json', 'validation.json', 'cleanup.json', 'status.json')
|
||||
and not file.name.endswith('.log')):
|
||||
checksums[str(file.relative_to(root))] = hashlib.sha256(file.read_bytes()).hexdigest()
|
||||
(root / 'checksums.json').write_text(json.dumps(checksums, indent=2))
|
||||
result = {'validated_unix': time.time(), 'files_checksummed': len(checksums),
|
||||
'group_edges_checked': True, 'passed': True}
|
||||
(root / 'validation.json').write_text(json.dumps(result, indent=2))
|
||||
return result
|
||||
|
||||
|
||||
def main():
|
||||
p = argparse.ArgumentParser()
|
||||
p.add_argument('--root', required=True, type=pathlib.Path)
|
||||
p.add_argument('--cleanup-weights', action='store_true')
|
||||
args = p.parse_args()
|
||||
root = args.root.resolve()
|
||||
result = validate(root)
|
||||
if args.cleanup_weights:
|
||||
weights = root / 'weights'
|
||||
manifest = json.loads((root / 'manifest.json').read_text())
|
||||
assert not weights.is_symlink()
|
||||
assert (weights / '.pilot-owned').read_text() == manifest['repo']
|
||||
expected_names = {entry['name'] for entry in manifest['files']} | {'.pilot-owned'}
|
||||
assert {file.name for file in weights.iterdir()} == expected_names
|
||||
assert all(file.is_file() and not file.is_symlink() for file in weights.iterdir())
|
||||
deleted_bytes = sum(file.stat().st_size for file in weights.iterdir())
|
||||
shutil.rmtree(weights)
|
||||
(root / 'cleanup.json').write_text(json.dumps({'deleted_unix': time.time(),
|
||||
'deleted_bytes': deleted_bytes,
|
||||
'deleted_directory': 'weights',
|
||||
'reason': 'User-authorized delete-after-analysis; archived results validated'}, indent=2))
|
||||
result['weights_removed'] = not weights.exists()
|
||||
print(json.dumps(result))
|
||||
|
||||
|
||||
if __name__ == '__main__':
|
||||
main()
|
||||
@@ -0,0 +1,2 @@
|
||||
numpy==2.5.4
|
||||
matplotlib==3.11.2
|
||||
@@ -0,0 +1,83 @@
|
||||
# Qwen3.5-2B parameter similarity pilot
|
||||
|
||||
- Revision: `15852e8c16360a2fea060d615a32b45270f8a8fc`
|
||||
- Stored parameters: 2,274,069,824
|
||||
- Weight bytes: 4,548,144,832
|
||||
- Tensor dimensionalities: {'1': 319, '2': 294, '3': 18, '5': 1}
|
||||
- Analysis and grouping/export wall time: 6661.1 seconds
|
||||
- Pair-comparison cutoff: 6600 seconds; final grouping/export is additional and remains under the supervisor hard limit.
|
||||
- Peak analysis RSS: 201.3 MiB
|
||||
|
||||
## Verified sharing estimates
|
||||
|
||||
| Representation | Weight-error limit | Verified sharing groups | Conservative total groups* | Largest group | Net model savings |
|
||||
|---|---:|---:|---:|---:|---:|---:|
|
||||
| whole | 1% | 0 | 632 | 1 | 0.000000% |
|
||||
| whole | 2% | 0 | 632 | 1 | 0.000000% |
|
||||
| whole | 5% | 1 | 631 | 2 | 0.000000% |
|
||||
| whole | 10% | 11 | 613 | 6 | 0.000719% |
|
||||
| whole | 20% | 16 | 572 | 15 | 0.002626% |
|
||||
| 2d-128 | 1% | 0 | 138576 | 1 | 0.000000% |
|
||||
| 2d-128 | 2% | 0 | 138576 | 1 | 0.000000% |
|
||||
| 2d-128 | 5% | 0 | 138576 | 1 | 0.000000% |
|
||||
| 2d-128 | 10% | 0 | 138576 | 1 | 0.000000% |
|
||||
| 2d-128 | 20% | 1 | 138575 | 2 | 0.000000% |
|
||||
| 2d-64 | 1% | 0 | 554304 | 1 | 0.000000% |
|
||||
| 2d-64 | 2% | 0 | 554304 | 1 | 0.000000% |
|
||||
| 2d-64 | 5% | 0 | 554304 | 1 | 0.000000% |
|
||||
| 2d-64 | 10% | 1 | 554303 | 2 | 0.000000% |
|
||||
| 2d-64 | 20% | 3 | 554299 | 3 | 0.000000% |
|
||||
| 3d-axis0-slice | 1% | 0 | 110592 | 1 | 0.000000% |
|
||||
| 3d-axis0-slice | 2% | 0 | 110592 | 1 | 0.000000% |
|
||||
| 3d-axis0-slice | 5% | 9 | 110581 | 3 | 0.000000% |
|
||||
| 3d-axis0-slice | 10% | 33 | 110542 | 7 | 0.000000% |
|
||||
| 3d-axis0-slice | 20% | 45 | 110483 | 12 | 0.000000% |
|
||||
| 3d-volume | 1% | 0 | 27648 | 1 | 0.000000% |
|
||||
| 3d-volume | 2% | 0 | 27648 | 1 | 0.000000% |
|
||||
| 3d-volume | 5% | 0 | 27648 | 1 | 0.000000% |
|
||||
| 3d-volume | 10% | 0 | 27648 | 1 | 0.000000% |
|
||||
| 3d-volume | 20% | 2 | 27646 | 2 | 0.000000% |
|
||||
|
||||
*Conservative total groups counts unexamined full blocks as singletons; boundary regions and ineligible tensors stay unchanged and are outside that block count.
|
||||
Savings count only verified replacement edges. Unsampled blocks remain unchanged. Different block scales must not be summed.
|
||||
Theoretical encoding: one stored representative per group, 16 bytes per object mapping and 64 bytes per shared-group descriptor; identity fallback if larger.
|
||||
Exact grouping uses direct replacement only. Scaled and spectral results are diagnostic.
|
||||
|
||||
## Stratified diagnostics
|
||||
|
||||
See `analysis/role_summary.csv` for same-role, same-shape statistics, separated by tensor dimensionality and block scale.
|
||||
See `analysis/dimensional_bytes.csv` for tensor counts versus weight-byte contributions by dimensionality.
|
||||
Heatmap missing pairs are hatched (`heatmap_pairs.csv`); identity entries are not empirical matches.
|
||||
Layer-distance medians are split by dimension/scale (`layer_distance_stratified.csv`); role and selection bias remain.
|
||||
Controls are split by dimension and density-normalized, with sample counts (`null_histograms.csv`); these are not significance tests.
|
||||
These are screened-candidate statistics, not estimates of all possible pairs. Small normalization vectors must not be conflated with large matrices.
|
||||
|
||||
## Coverage
|
||||
|
||||
- tensor: `{"eligible_pairs_in_screened_objects": 6159, "candidate_pairs_before_cap": 4724, "selected_pairs": 2000, "objects_screened": 632, "pairs_verified": 2000, "timed_out": false, "screen_sample_elements": 2048, "candidate_neighbors_per_metric": 10}`
|
||||
- block: `{"eligible_pairs_in_screened_objects": 1678968, "candidate_pairs_before_cap": 175036, "selected_pairs": 20000, "objects_screened": 13328, "pairs_verified": 14778, "timed_out": true, "screen_sample_elements": 2048, "candidate_neighbors_per_metric": 10}`
|
||||
- Complete-block partition totals (per alternative): `{"2d-64": {"blocks": 554304, "bytes": 4540858368, "edge_bytes": 0}, "2d-128": {"blocks": 138576, "bytes": 4540858368, "edge_bytes": 0}, "3d-axis0-slice": {"blocks": 110592, "bytes": 884736, "edge_bytes": 0}, "3d-volume": {"blocks": 27648, "bytes": 884736, "edge_bytes": 0}}`
|
||||
|
||||
## Interpretation limits
|
||||
|
||||
- No inference or functional equivalence measured
|
||||
- No exhaustive pair search
|
||||
- Block search uses per-role/shape/scale reservoirs of 256
|
||||
- Different block scales overlap; savings must not be added
|
||||
- Block savings count only verified disjoint sampled blocks, no extrapolation
|
||||
- Spectral probes use aligned bounded subvolumes, not whole tensors
|
||||
- Shared encoding is theoretical, not a runnable compressed checkpoint
|
||||
- No cross-role grouping, neuron permutation alignment, or training
|
||||
- Candidate histograms are selection-biased and do not represent the distribution of all tensor pairs.
|
||||
- 1D normalization weights may cluster strongly but carry few bytes; do not infer material compression from group counts.
|
||||
- No task accuracy, perplexity, inference speed or runtime-memory improvement has been measured.
|
||||
- Stored parameter counts may omit tied aliases; no duplicate-weight equality search is performed.
|
||||
|
||||
## Figures
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
|
||||
@@ -0,0 +1,171 @@
|
||||
Inventory: 632 tensors, 4548144832 bytes
|
||||
tensor: sampling 632 objects
|
||||
tensor: verified 100/2000
|
||||
tensor: verified 200/2000
|
||||
tensor: verified 300/2000
|
||||
tensor: verified 400/2000
|
||||
tensor: verified 500/2000
|
||||
tensor: verified 600/2000
|
||||
tensor: verified 700/2000
|
||||
tensor: verified 800/2000
|
||||
tensor: verified 900/2000
|
||||
tensor: verified 1000/2000
|
||||
tensor: verified 1100/2000
|
||||
tensor: verified 1200/2000
|
||||
tensor: verified 1300/2000
|
||||
tensor: verified 1400/2000
|
||||
tensor: verified 1500/2000
|
||||
tensor: verified 1600/2000
|
||||
tensor: verified 1700/2000
|
||||
tensor: verified 1800/2000
|
||||
tensor: verified 1900/2000
|
||||
tensor: verified 2000/2000
|
||||
block: sampling 13328 objects
|
||||
block: verified 100/20000
|
||||
block: verified 200/20000
|
||||
block: verified 300/20000
|
||||
block: verified 400/20000
|
||||
block: verified 500/20000
|
||||
block: verified 600/20000
|
||||
block: verified 700/20000
|
||||
block: verified 800/20000
|
||||
block: verified 900/20000
|
||||
block: verified 1000/20000
|
||||
block: verified 1100/20000
|
||||
block: verified 1200/20000
|
||||
block: verified 1300/20000
|
||||
block: verified 1400/20000
|
||||
block: verified 1500/20000
|
||||
block: verified 1600/20000
|
||||
block: verified 1700/20000
|
||||
block: verified 1800/20000
|
||||
block: verified 1900/20000
|
||||
block: verified 2000/20000
|
||||
block: verified 2100/20000
|
||||
block: verified 2200/20000
|
||||
block: verified 2300/20000
|
||||
block: verified 2400/20000
|
||||
block: verified 2500/20000
|
||||
block: verified 2600/20000
|
||||
block: verified 2700/20000
|
||||
block: verified 2800/20000
|
||||
block: verified 2900/20000
|
||||
block: verified 3000/20000
|
||||
block: verified 3100/20000
|
||||
block: verified 3200/20000
|
||||
block: verified 3300/20000
|
||||
block: verified 3400/20000
|
||||
block: verified 3500/20000
|
||||
block: verified 3600/20000
|
||||
block: verified 3700/20000
|
||||
block: verified 3800/20000
|
||||
block: verified 3900/20000
|
||||
block: verified 4000/20000
|
||||
block: verified 4100/20000
|
||||
block: verified 4200/20000
|
||||
block: verified 4300/20000
|
||||
block: verified 4400/20000
|
||||
block: verified 4500/20000
|
||||
block: verified 4600/20000
|
||||
block: verified 4700/20000
|
||||
block: verified 4800/20000
|
||||
block: verified 4900/20000
|
||||
block: verified 5000/20000
|
||||
block: verified 5100/20000
|
||||
block: verified 5200/20000
|
||||
block: verified 5300/20000
|
||||
block: verified 5400/20000
|
||||
block: verified 5500/20000
|
||||
block: verified 5600/20000
|
||||
block: verified 5700/20000
|
||||
block: verified 5800/20000
|
||||
block: verified 5900/20000
|
||||
block: verified 6000/20000
|
||||
block: verified 6100/20000
|
||||
block: verified 6200/20000
|
||||
block: verified 6300/20000
|
||||
block: verified 6400/20000
|
||||
block: verified 6500/20000
|
||||
block: verified 6600/20000
|
||||
block: verified 6700/20000
|
||||
block: verified 6800/20000
|
||||
block: verified 6900/20000
|
||||
block: verified 7000/20000
|
||||
block: verified 7100/20000
|
||||
block: verified 7200/20000
|
||||
block: verified 7300/20000
|
||||
block: verified 7400/20000
|
||||
block: verified 7500/20000
|
||||
block: verified 7600/20000
|
||||
block: verified 7700/20000
|
||||
block: verified 7800/20000
|
||||
block: verified 7900/20000
|
||||
block: verified 8000/20000
|
||||
block: verified 8100/20000
|
||||
block: verified 8200/20000
|
||||
block: verified 8300/20000
|
||||
block: verified 8400/20000
|
||||
block: verified 8500/20000
|
||||
block: verified 8600/20000
|
||||
block: verified 8700/20000
|
||||
block: verified 8800/20000
|
||||
block: verified 8900/20000
|
||||
block: verified 9000/20000
|
||||
block: verified 9100/20000
|
||||
block: verified 9200/20000
|
||||
block: verified 9300/20000
|
||||
block: verified 9400/20000
|
||||
block: verified 9500/20000
|
||||
block: verified 9600/20000
|
||||
block: verified 9700/20000
|
||||
block: verified 9800/20000
|
||||
block: verified 9900/20000
|
||||
block: verified 10000/20000
|
||||
block: verified 10100/20000
|
||||
block: verified 10200/20000
|
||||
block: verified 10300/20000
|
||||
block: verified 10400/20000
|
||||
block: verified 10500/20000
|
||||
block: verified 10600/20000
|
||||
block: verified 10700/20000
|
||||
block: verified 10800/20000
|
||||
block: verified 10900/20000
|
||||
block: verified 11000/20000
|
||||
block: verified 11100/20000
|
||||
block: verified 11200/20000
|
||||
block: verified 11300/20000
|
||||
block: verified 11400/20000
|
||||
block: verified 11500/20000
|
||||
block: verified 11600/20000
|
||||
block: verified 11700/20000
|
||||
block: verified 11800/20000
|
||||
block: verified 11900/20000
|
||||
block: verified 12000/20000
|
||||
block: verified 12100/20000
|
||||
block: verified 12200/20000
|
||||
block: verified 12300/20000
|
||||
block: verified 12400/20000
|
||||
block: verified 12500/20000
|
||||
block: verified 12600/20000
|
||||
block: verified 12700/20000
|
||||
block: verified 12800/20000
|
||||
block: verified 12900/20000
|
||||
block: verified 13000/20000
|
||||
block: verified 13100/20000
|
||||
block: verified 13200/20000
|
||||
block: verified 13300/20000
|
||||
block: verified 13400/20000
|
||||
block: verified 13500/20000
|
||||
block: verified 13600/20000
|
||||
block: verified 13700/20000
|
||||
block: verified 13800/20000
|
||||
block: verified 13900/20000
|
||||
block: verified 14000/20000
|
||||
block: verified 14100/20000
|
||||
block: verified 14200/20000
|
||||
block: verified 14300/20000
|
||||
block: verified 14400/20000
|
||||
block: verified 14500/20000
|
||||
block: verified 14600/20000
|
||||
block: verified 14700/20000
|
||||
{"model": "Qwen/Qwen3.5-2B", "revision": "15852e8c16360a2fea060d615a32b45270f8a8fc", "tensor_count": 632, "parameter_count": 2274069824, "model_weight_bytes": 4548144832, "ndim_counts": {"1": 319, "2": 294, "3": 18, "5": 1}, "elapsed_seconds": 6661.129063449334, "analysis_budget_seconds": 6600, "peak_rss_kib": 206180, "python": "3.14.4", "numpy": "2.5.4", "seed": 20261109, "limitations": ["No inference or functional equivalence measured", "No exhaustive pair search", "Block search uses per-role/shape/scale reservoirs of 256", "Different block scales overlap; savings must not be added", "Block savings count only verified disjoint sampled blocks, no extrapolation", "Spectral probes use aligned bounded subvolumes, not whole tensors", "Shared encoding is theoretical, not a runnable compressed checkpoint", "No cross-role grouping, neuron permutation alignment, or training"], "completed": true}
|
||||
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"eligible_pairs_in_screened_objects": 1678968,
|
||||
"candidate_pairs_before_cap": 175036,
|
||||
"selected_pairs": 20000,
|
||||
"objects_screened": 13328,
|
||||
"pairs_verified": 14778,
|
||||
"timed_out": true,
|
||||
"screen_sample_elements": 2048,
|
||||
"candidate_neighbors_per_metric": 10
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,22 @@
|
||||
{
|
||||
"2d-64": {
|
||||
"blocks": 554304,
|
||||
"bytes": 4540858368,
|
||||
"edge_bytes": 0
|
||||
},
|
||||
"2d-128": {
|
||||
"blocks": 138576,
|
||||
"bytes": 4540858368,
|
||||
"edge_bytes": 0
|
||||
},
|
||||
"3d-axis0-slice": {
|
||||
"blocks": 110592,
|
||||
"bytes": 884736,
|
||||
"edge_bytes": 0
|
||||
},
|
||||
"3d-volume": {
|
||||
"blocks": 27648,
|
||||
"bytes": 884736,
|
||||
"edge_bytes": 0
|
||||
}
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,55 @@
|
||||
role,shape,scale,total_blocks,sampled_blocks
|
||||
model.language_model.embed_tokens.weight,"(128, 128)",2d-128,31040,256
|
||||
model.language_model.embed_tokens.weight,"(64, 64)",2d-64,124160,256
|
||||
model.language_model.layers.*.linear_attn.conv1d.weight,"(1, 1, 4)",3d-axis0-slice,110592,256
|
||||
model.language_model.layers.*.linear_attn.conv1d.weight,"(4, 1, 4)",3d-volume,27648,256
|
||||
model.language_model.layers.*.linear_attn.in_proj_qkv.weight,"(128, 128)",2d-128,13824,256
|
||||
model.language_model.layers.*.linear_attn.in_proj_qkv.weight,"(64, 64)",2d-64,55296,256
|
||||
model.language_model.layers.*.linear_attn.in_proj_z.weight,"(128, 128)",2d-128,4608,256
|
||||
model.language_model.layers.*.linear_attn.in_proj_z.weight,"(64, 64)",2d-64,18432,256
|
||||
model.language_model.layers.*.linear_attn.out_proj.weight,"(128, 128)",2d-128,4608,256
|
||||
model.language_model.layers.*.linear_attn.out_proj.weight,"(64, 64)",2d-64,18432,256
|
||||
model.language_model.layers.*.mlp.down_proj.weight,"(128, 128)",2d-128,18432,256
|
||||
model.language_model.layers.*.mlp.down_proj.weight,"(64, 64)",2d-64,73728,256
|
||||
model.language_model.layers.*.mlp.gate_proj.weight,"(128, 128)",2d-128,18432,256
|
||||
model.language_model.layers.*.mlp.gate_proj.weight,"(64, 64)",2d-64,73728,256
|
||||
model.language_model.layers.*.mlp.up_proj.weight,"(128, 128)",2d-128,18432,256
|
||||
model.language_model.layers.*.mlp.up_proj.weight,"(64, 64)",2d-64,73728,256
|
||||
model.language_model.layers.*.self_attn.k_proj.weight,"(128, 128)",2d-128,384,256
|
||||
model.language_model.layers.*.self_attn.k_proj.weight,"(64, 64)",2d-64,1536,256
|
||||
model.language_model.layers.*.self_attn.o_proj.weight,"(128, 128)",2d-128,1536,256
|
||||
model.language_model.layers.*.self_attn.o_proj.weight,"(64, 64)",2d-64,6144,256
|
||||
model.language_model.layers.*.self_attn.q_proj.weight,"(128, 128)",2d-128,3072,256
|
||||
model.language_model.layers.*.self_attn.q_proj.weight,"(64, 64)",2d-64,12288,256
|
||||
model.language_model.layers.*.self_attn.v_proj.weight,"(128, 128)",2d-128,384,256
|
||||
model.language_model.layers.*.self_attn.v_proj.weight,"(64, 64)",2d-64,1536,256
|
||||
model.visual.layers.*.attn.proj.weight,"(128, 128)",2d-128,1536,256
|
||||
model.visual.layers.*.attn.proj.weight,"(64, 64)",2d-64,6144,256
|
||||
model.visual.layers.*.attn.qkv.weight,"(128, 128)",2d-128,4608,256
|
||||
model.visual.layers.*.attn.qkv.weight,"(64, 64)",2d-64,18432,256
|
||||
model.visual.layers.*.mlp.linear_fc1.weight,"(128, 128)",2d-128,6144,256
|
||||
model.visual.layers.*.mlp.linear_fc1.weight,"(64, 64)",2d-64,24576,256
|
||||
model.visual.layers.*.mlp.linear_fc2.weight,"(128, 128)",2d-128,6144,256
|
||||
model.visual.layers.*.mlp.linear_fc2.weight,"(64, 64)",2d-64,24576,256
|
||||
model.visual.merger.linear_fc1.weight,"(128, 128)",2d-128,1024,256
|
||||
model.visual.merger.linear_fc1.weight,"(64, 64)",2d-64,4096,256
|
||||
model.visual.merger.linear_fc2.weight,"(128, 128)",2d-128,512,256
|
||||
model.visual.merger.linear_fc2.weight,"(64, 64)",2d-64,2048,256
|
||||
model.visual.pos_embed.weight,"(128, 128)",2d-128,144,144
|
||||
model.visual.pos_embed.weight,"(64, 64)",2d-64,576,256
|
||||
mtp.fc.weight,"(128, 128)",2d-128,512,256
|
||||
mtp.fc.weight,"(64, 64)",2d-64,2048,256
|
||||
mtp.layers.*.mlp.down_proj.weight,"(128, 128)",2d-128,768,256
|
||||
mtp.layers.*.mlp.down_proj.weight,"(64, 64)",2d-64,3072,256
|
||||
mtp.layers.*.mlp.gate_proj.weight,"(128, 128)",2d-128,768,256
|
||||
mtp.layers.*.mlp.gate_proj.weight,"(64, 64)",2d-64,3072,256
|
||||
mtp.layers.*.mlp.up_proj.weight,"(128, 128)",2d-128,768,256
|
||||
mtp.layers.*.mlp.up_proj.weight,"(64, 64)",2d-64,3072,256
|
||||
mtp.layers.*.self_attn.k_proj.weight,"(128, 128)",2d-128,64,64
|
||||
mtp.layers.*.self_attn.k_proj.weight,"(64, 64)",2d-64,256,256
|
||||
mtp.layers.*.self_attn.o_proj.weight,"(128, 128)",2d-128,256,256
|
||||
mtp.layers.*.self_attn.o_proj.weight,"(64, 64)",2d-64,1024,256
|
||||
mtp.layers.*.self_attn.q_proj.weight,"(128, 128)",2d-128,512,256
|
||||
mtp.layers.*.self_attn.q_proj.weight,"(64, 64)",2d-64,2048,256
|
||||
mtp.layers.*.self_attn.v_proj.weight,"(128, 128)",2d-128,64,64
|
||||
mtp.layers.*.self_attn.v_proj.weight,"(64, 64)",2d-64,256,256
|
||||
|
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,26 @@
|
||||
mode,scale,threshold,objects,groups,total_partition_objects,unexamined_partition_objects,conservative_partition_groups,shared_groups,max_group_size,verified_shared_bytes,examined_object_bytes,representative_bytes,mapping_bytes_if_encoded,net_saved_bytes,net_saved_model_fraction,max_replacement_error
|
||||
tensor,whole,0.01,632,632,632,0,632,0,1,0,4548144832,4548144832,10112,0,0.0,0.0
|
||||
tensor,whole,0.02,632,632,632,0,632,0,1,0,4548144832,4548144832,10112,0,0.0,0.0
|
||||
tensor,whole,0.05,632,631,632,0,631,1,2,4096,4548144832,4548142784,10176,0,0.0,0.04982565601433354
|
||||
tensor,whole,0.1,632,613,632,0,613,11,6,66560,4548144832,4548101312,10816,32704,7.19062413533976e-06,0.0996284846101826
|
||||
tensor,whole,0.2,632,572,632,0,572,16,15,174592,4548144832,4548014272,11136,119424,2.6257739014763196e-05,0.1992427130434979
|
||||
block,2d-128,0.01,6160,6160,138576,132416,138576,0,1,0,201850880,201850880,98560,0,0.0,0.0
|
||||
block,2d-128,0.02,6160,6160,138576,132416,138576,0,1,0,201850880,201850880,98560,0,0.0,0.0
|
||||
block,2d-128,0.05,6160,6160,138576,132416,138576,0,1,0,201850880,201850880,98560,0,0.0,0.0
|
||||
block,2d-128,0.1,6160,6160,138576,132416,138576,0,1,0,201850880,201850880,98560,0,0.0,0.0
|
||||
block,2d-128,0.2,6160,6159,138576,132416,138575,1,2,65536,201850880,201818112,98624,0,0.0,0.1894907479098447
|
||||
block,2d-64,0.01,6656,6656,554304,547648,554304,0,1,0,54525952,54525952,106496,0,0.0,0.0
|
||||
block,2d-64,0.02,6656,6656,554304,547648,554304,0,1,0,54525952,54525952,106496,0,0.0,0.0
|
||||
block,2d-64,0.05,6656,6656,554304,547648,554304,0,1,0,54525952,54525952,106496,0,0.0,0.0
|
||||
block,2d-64,0.1,6656,6655,554304,547648,554303,1,2,16384,54525952,54517760,106560,0,0.0,0.09362662282834061
|
||||
block,2d-64,0.2,6656,6651,554304,547648,554299,3,3,65536,54525952,54484992,106688,0,0.0,0.18661711940726244
|
||||
block,3d-axis0-slice,0.01,256,256,110592,110336,110592,0,1,0,2048,2048,4096,0,0.0,0.0
|
||||
block,3d-axis0-slice,0.02,256,256,110592,110336,110592,0,1,0,2048,2048,4096,0,0.0,0.0
|
||||
block,3d-axis0-slice,0.05,256,245,110592,110336,110581,9,3,160,2048,1960,4672,0,0.0,0.04716472920859973
|
||||
block,3d-axis0-slice,0.1,256,206,110592,110336,110542,33,7,664,2048,1648,6208,0,0.0,0.09947536518147226
|
||||
block,3d-axis0-slice,0.2,256,147,110592,110336,110483,45,12,1232,2048,1176,6976,0,0.0,0.19719308219928766
|
||||
block,3d-volume,0.01,256,256,27648,27392,27648,0,1,0,8192,8192,4096,0,0.0,0.0
|
||||
block,3d-volume,0.02,256,256,27648,27392,27648,0,1,0,8192,8192,4096,0,0.0,0.0
|
||||
block,3d-volume,0.05,256,256,27648,27392,27648,0,1,0,8192,8192,4096,0,0.0,0.0
|
||||
block,3d-volume,0.1,256,256,27648,27392,27648,0,1,0,8192,8192,4096,0,0.0,0.0
|
||||
block,3d-volume,0.2,256,254,27648,27392,27646,2,2,128,8192,8128,4224,0,0.0,0.19802419391338302
|
||||
|
@@ -0,0 +1,5 @@
|
||||
ndim,tensor_count,weight_bytes,model_byte_fraction
|
||||
1,319,896704,0.00019715818935468765
|
||||
2,294,4543217664,0.9989166642264043
|
||||
3,18,884736,0.00019452678678461222
|
||||
5,1,3145728,0.000691650797456399
|
||||
|
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,325 @@
|
||||
a_layer,b_layer,role,cosine,verified,identity
|
||||
0,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
0,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
0,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
0,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.021603168261668075,True,False
|
||||
0,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016060099025396145,True,False
|
||||
0,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.011134467760118835,True,False
|
||||
0,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.02119149075528504,True,False
|
||||
0,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016503057196088678,True,False
|
||||
0,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.014928623062813407,True,False
|
||||
0,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
0,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.018257428983934995,True,False
|
||||
0,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016567710854311527,True,False
|
||||
0,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.01878152412750565,True,False
|
||||
0,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.01800800596875963,True,False
|
||||
0,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.028970941325172344,True,False
|
||||
0,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.022135507322117717,True,False
|
||||
0,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.025104063163733144,True,False
|
||||
0,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
1,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.052801535487043146,True,False
|
||||
1,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.07322831943625455,True,False
|
||||
1,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.06485177649773176,True,False
|
||||
1,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0681849224686047,True,False
|
||||
1,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.05608619803907555,True,False
|
||||
1,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.06094079473458446,True,False
|
||||
1,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0593113361588042,True,False
|
||||
1,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0669216453028539,True,False
|
||||
1,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
1,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
2,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.08891505471270902,True,False
|
||||
2,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
2,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.021603168261668075,True,False
|
||||
4,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
4,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.17566810276395187,True,False
|
||||
4,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15586663085457408,True,False
|
||||
4,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1185198625732783,True,False
|
||||
4,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12851474649025826,True,False
|
||||
4,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.07448081425577983,True,False
|
||||
4,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
4,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016060099025396145,True,False
|
||||
5,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.052801535487043146,True,False
|
||||
5,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.17566810276395187,True,False
|
||||
5,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
5,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14377798800191915,True,False
|
||||
5,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
5,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12262286697791511,True,False
|
||||
6,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.011134467760118835,True,False
|
||||
6,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.07322831943625455,True,False
|
||||
6,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15586663085457408,True,False
|
||||
6,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
6,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.21699783588937813,True,False
|
||||
6,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1333883918363074,True,False
|
||||
6,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1578017536943907,True,False
|
||||
6,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
6,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12015120428871615,True,False
|
||||
6,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1273046517923862,True,False
|
||||
6,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1311555097410448,True,False
|
||||
8,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.02119149075528504,True,False
|
||||
8,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.06485177649773176,True,False
|
||||
8,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1185198625732783,True,False
|
||||
8,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14377798800191915,True,False
|
||||
8,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
8,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15843144658604474,True,False
|
||||
8,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
8,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.10970674337391326,True,False
|
||||
8,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1424385230009979,True,False
|
||||
8,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12404030329005741,True,False
|
||||
9,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016503057196088678,True,False
|
||||
9,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0681849224686047,True,False
|
||||
9,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.21699783588937813,True,False
|
||||
9,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
9,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1803763955625086,True,False
|
||||
9,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14949388376365874,True,False
|
||||
9,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16608693423041895,True,False
|
||||
9,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
9,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12672973282980668,True,False
|
||||
10,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.014928623062813407,True,False
|
||||
10,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.05608619803907555,True,False
|
||||
10,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
10,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.18714191280792747,True,False
|
||||
10,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15298970272056192,True,False
|
||||
10,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
10,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1455075615532816,True,False
|
||||
10,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14423059933719398,True,False
|
||||
12,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1333883918363074,True,False
|
||||
12,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15843144658604474,True,False
|
||||
12,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1803763955625086,True,False
|
||||
12,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.18714191280792747,True,False
|
||||
12,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
12,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1976207505481654,True,False
|
||||
12,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.19384548810955707,True,False
|
||||
12,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
12,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.13400309235372598,True,False
|
||||
13,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.018257428983934995,True,False
|
||||
13,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.06094079473458446,True,False
|
||||
13,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12851474649025826,True,False
|
||||
13,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1976207505481654,True,False
|
||||
13,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
13,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1779032565964785,True,False
|
||||
13,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1787518876331122,True,False
|
||||
13,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
13,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16267368170342447,True,False
|
||||
13,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.184820355422847,True,False
|
||||
13,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1855431215985299,True,False
|
||||
14,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.016567710854311527,True,False
|
||||
14,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0593113361588042,True,False
|
||||
14,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.08891505471270902,True,False
|
||||
14,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.07448081425577983,True,False
|
||||
14,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14949388376365874,True,False
|
||||
14,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.19384548810955707,True,False
|
||||
14,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
14,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1319696165434511,True,False
|
||||
14,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.19933052471925228,True,False
|
||||
14,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
14,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1251036484760185,True,False
|
||||
14,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16409792240962637,True,False
|
||||
14,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.147211090153063,True,False
|
||||
16,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.01878152412750565,True,False
|
||||
16,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1779032565964785,True,False
|
||||
16,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1319696165434511,True,False
|
||||
16,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
16,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
16,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.23701407107786296,True,False
|
||||
16,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.26531637641439015,True,False
|
||||
16,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1973593532239659,True,False
|
||||
16,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14352438103427206,True,False
|
||||
17,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.01800800596875963,True,False
|
||||
17,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1578017536943907,True,False
|
||||
17,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.15298970272056192,True,False
|
||||
17,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1787518876331122,True,False
|
||||
17,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.19933052471925228,True,False
|
||||
17,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
17,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
17,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.27601649873822437,True,False
|
||||
17,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.25004925100500414,True,False
|
||||
17,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2683584311866289,True,False
|
||||
17,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.23217037996094514,True,False
|
||||
18,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.028970941325172344,True,False
|
||||
18,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.0669216453028539,True,False
|
||||
18,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
18,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.23701407107786296,True,False
|
||||
18,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.27601649873822437,True,False
|
||||
18,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
18,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.29759171873785567,True,False
|
||||
18,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.28909453533949464,True,False
|
||||
18,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.22289938280365845,True,False
|
||||
20,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.022135507322117717,True,False
|
||||
20,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12015120428871615,True,False
|
||||
20,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.10970674337391326,True,False
|
||||
20,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16608693423041895,True,False
|
||||
20,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
20,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16267368170342447,True,False
|
||||
20,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1251036484760185,True,False
|
||||
20,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.26531637641439015,True,False
|
||||
20,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.25004925100500414,True,False
|
||||
20,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.29759171873785567,True,False
|
||||
20,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
20,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2927090499229011,True,False
|
||||
20,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2916728222104041,True,False
|
||||
21,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.025104063163733144,True,False
|
||||
21,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1273046517923862,True,False
|
||||
21,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1424385230009979,True,False
|
||||
21,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1455075615532816,True,False
|
||||
21,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
21,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.184820355422847,True,False
|
||||
21,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.16409792240962637,True,False
|
||||
21,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1973593532239659,True,False
|
||||
21,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2683584311866289,True,False
|
||||
21,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.28909453533949464,True,False
|
||||
21,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2927090499229011,True,False
|
||||
21,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
21,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.27977406968857854,True,False
|
||||
22,0,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
22,1,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
22,2,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
22,4,model.language_model.layers.*.linear_attn.in_proj_a.weight,,False,False
|
||||
22,5,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12262286697791511,True,False
|
||||
22,6,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1311555097410448,True,False
|
||||
22,8,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12404030329005741,True,False
|
||||
22,9,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.12672973282980668,True,False
|
||||
22,10,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14423059933719398,True,False
|
||||
22,12,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.13400309235372598,True,False
|
||||
22,13,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.1855431215985299,True,False
|
||||
22,14,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.147211090153063,True,False
|
||||
22,16,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.14352438103427206,True,False
|
||||
22,17,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.23217037996094514,True,False
|
||||
22,18,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.22289938280365845,True,False
|
||||
22,20,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.2916728222104041,True,False
|
||||
22,21,model.language_model.layers.*.linear_attn.in_proj_a.weight,0.27977406968857854,True,False
|
||||
22,22,model.language_model.layers.*.linear_attn.in_proj_a.weight,1.0,False,True
|
||||
|
@@ -0,0 +1,633 @@
|
||||
name,shape,ndim,dtype,numel,bytes,role,layer,mean,rms,max_abs
|
||||
model.language_model.layers.0.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,0,-2.5419921875,2.8428578077590725,5.90625
|
||||
model.language_model.layers.0.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,0,0.9159862995147705,0.9217019394932431,1.0107421875
|
||||
model.language_model.layers.1.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,1,-0.766021728515625,1.562546486163045,2.75
|
||||
model.language_model.layers.1.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,1,0.8760199546813965,0.882772150804847,1.024658203125
|
||||
model.language_model.layers.10.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,10,-1.208251953125,1.4532067060572578,2.046875
|
||||
model.language_model.layers.10.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,10,1.0112952142953873,1.0144369416249537,1.16796875
|
||||
model.language_model.layers.12.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,12,-1.6552734375,1.8625297390400493,3.015625
|
||||
model.language_model.layers.12.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,12,0.9601620733737946,0.9634251107082851,1.130859375
|
||||
model.language_model.layers.13.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,13,-1.4144287109375,1.510479954237702,2.265625
|
||||
model.language_model.layers.13.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,13,0.9877877533435822,0.9899024920840337,1.11474609375
|
||||
model.language_model.layers.14.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,14,-1.09912109375,1.533638473223965,2.59375
|
||||
model.language_model.layers.14.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,14,0.9645823240280151,0.9672255993966286,1.10595703125
|
||||
model.language_model.layers.16.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,16,-1.7109375,1.7936473218924704,2.453125
|
||||
model.language_model.layers.16.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,16,0.8653912842273712,0.8692723927344954,1.040771484375
|
||||
model.language_model.layers.17.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,17,-1.768798828125,1.8086773253671764,2.40625
|
||||
model.language_model.layers.17.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,17,0.7928063869476318,0.7980471928645709,0.993194580078125
|
||||
model.language_model.layers.18.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,18,-1.69677734375,1.7212882979155513,2.234375
|
||||
model.language_model.layers.18.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,18,0.7872825860977173,0.7911254701787153,0.9977569580078125
|
||||
model.language_model.layers.2.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,2,-1.2023239135742188,1.5029171827885457,3.25
|
||||
model.language_model.layers.2.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,2,0.8907453417778015,0.9000626793352592,1.0537109375
|
||||
model.language_model.layers.20.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,20,-1.664794921875,1.7204063064264483,2.25
|
||||
model.language_model.layers.20.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,20,0.8415859341621399,0.8442340143169718,1.0244140625
|
||||
model.language_model.layers.21.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,21,-1.83447265625,1.8661585570389687,2.265625
|
||||
model.language_model.layers.21.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,21,0.9078961908817291,0.9098948839174701,1.04296875
|
||||
model.language_model.layers.22.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,22,-1.749267578125,1.814547993096487,2.15625
|
||||
model.language_model.layers.22.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,22,1.0633795857429504,1.0649380921373046,1.2373046875
|
||||
model.language_model.layers.4.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,4,-1.1160888671875,1.573153454807781,3.078125
|
||||
model.language_model.layers.4.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,4,0.8434299230575562,0.8544616726903967,1.0498046875
|
||||
model.language_model.layers.5.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,5,-1.345733642578125,1.9737311918644171,3.328125
|
||||
model.language_model.layers.5.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,5,0.8853708505630493,0.8928125777319282,1.0703125
|
||||
model.language_model.layers.6.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,6,-0.924560546875,1.1713577146067924,2.671875
|
||||
model.language_model.layers.6.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,6,0.937565041705966,0.9418680396820469,1.10302734375
|
||||
model.language_model.layers.8.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,8,-1.4276123046875,1.5802849067268956,3.265625
|
||||
model.language_model.layers.8.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,8,0.9574419558048248,0.9634701406436879,1.17578125
|
||||
model.language_model.layers.9.linear_attn.A_log,"(16,)",1,F32,16,64,model.language_model.layers.*.linear_attn.A_log,9,-1.752166748046875,1.8962492752891067,2.703125
|
||||
model.language_model.layers.9.linear_attn.norm.weight,"(128,)",1,F32,128,512,model.language_model.layers.*.linear_attn.norm.weight,9,0.97660231590271,0.9809916884875952,1.1455078125
|
||||
model.language_model.embed_tokens.weight,"(248320, 2048)",2,BF16,508559360,1017118720,model.language_model.embed_tokens.weight,-1,9.87077120664951e-06,0.015007928633636286,0.197265625
|
||||
model.language_model.layers.0.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,0,0.09555475553497672,0.15187859860523678,1.0859375
|
||||
model.language_model.layers.0.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,0,0.0006016545487265527,0.06308888776276214,1.5234375
|
||||
model.language_model.layers.0.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,0,-1.1572265625,6.370118909172119,12.3125
|
||||
model.language_model.layers.0.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,0,0.00020052339823450893,0.029087606938219022,0.2216796875
|
||||
model.language_model.layers.0.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,0,9.180475922221376e-05,0.013443365387314386,0.119140625
|
||||
model.language_model.layers.0.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,0,5.650052049559669e-06,0.014758352047841181,0.39453125
|
||||
model.language_model.layers.0.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,0,-1.2114403831961496e-06,0.013986197489437876,0.37109375
|
||||
model.language_model.layers.0.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,0,-1.682266086777684e-05,0.01201362947529297,0.388671875
|
||||
model.language_model.layers.0.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,0,-4.579636075247954e-07,0.008271090674254746,0.330078125
|
||||
model.language_model.layers.0.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,0,1.0803703818374153e-05,0.009989848530660638,0.1875
|
||||
model.language_model.layers.0.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,0,1.2639582468662918e-06,0.008215167992998686,0.11474609375
|
||||
model.language_model.layers.0.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,0,-0.0067373557249084115,0.1086428965479044,0.99609375
|
||||
model.language_model.layers.1.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,1,0.20870919525623322,0.22850867473342584,0.61328125
|
||||
model.language_model.layers.1.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,1,0.00030097006485145056,0.05669035503528881,1.5625
|
||||
model.language_model.layers.1.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,1,-1.052734375,3.073152660208929,6.65625
|
||||
model.language_model.layers.1.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,1,1.4156091936001758e-05,0.024413679152183832,0.220703125
|
||||
model.language_model.layers.1.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,1,-0.00014877098203669448,0.014451269033729877,0.15234375
|
||||
model.language_model.layers.1.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,1,-4.853525533118445e-06,0.01482576287583218,0.255859375
|
||||
model.language_model.layers.1.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,1,-4.412525976188464e-06,0.014568041830426256,0.2001953125
|
||||
model.language_model.layers.1.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,1,2.332247853726113e-06,0.012184268600894286,0.455078125
|
||||
model.language_model.layers.1.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,1,5.54108184586682e-07,0.00810784551122672,0.2734375
|
||||
model.language_model.layers.1.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,1,9.073857168912883e-06,0.010212754570926883,0.142578125
|
||||
model.language_model.layers.1.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,1,-2.3373373599722483e-06,0.008126413449240725,0.11865234375
|
||||
model.language_model.layers.1.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,1,-0.005656745750457048,0.09427294029637952,0.54296875
|
||||
model.language_model.layers.10.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,10,0.15872192569077015,0.27164572436102286,0.6171875
|
||||
model.language_model.layers.10.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,10,-0.0007379838181596673,0.06830613403695973,1.25
|
||||
model.language_model.layers.10.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,10,-2.16278076171875,3.041065340609617,7.28125
|
||||
model.language_model.layers.10.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,10,-8.714331785597551e-05,0.017147214192356434,0.11767578125
|
||||
model.language_model.layers.10.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,10,-0.00015136123823822345,0.011674008729477254,0.0859375
|
||||
model.language_model.layers.10.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,10,-6.512366322017754e-06,0.015292003408611508,0.2431640625
|
||||
model.language_model.layers.10.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,10,-8.52498674149371e-05,0.011903205521723808,0.150390625
|
||||
model.language_model.layers.10.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,10,6.316189763831519e-06,0.011903507511222908,0.30859375
|
||||
model.language_model.layers.10.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,10,-6.106724749141423e-08,0.008299914236029136,0.208984375
|
||||
model.language_model.layers.10.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,10,-2.324051282595954e-06,0.009179932401653506,0.13671875
|
||||
model.language_model.layers.10.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,10,-1.564070842671623e-06,0.008727795095731869,0.12255859375
|
||||
model.language_model.layers.10.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,10,-0.011391313164494932,0.16650153545868948,0.63671875
|
||||
model.language_model.layers.11.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,11,0.032779622822999954,0.1692437174974945,0.65625
|
||||
model.language_model.layers.11.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,11,1.4238889311427056e-06,0.008395161732559041,0.2421875
|
||||
model.language_model.layers.11.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,11,1.4327512621274453e-06,0.009103386394159245,0.1748046875
|
||||
model.language_model.layers.11.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,11,2.078183690977046e-06,0.00884748655359451,0.140625
|
||||
model.language_model.layers.11.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,11,0.01722284685820341,0.15380996823531595,0.58203125
|
||||
model.language_model.layers.11.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,11,0.6151275634765625,0.6597999706103095,1.4375
|
||||
model.language_model.layers.11.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,11,-4.683068861480871e-06,0.01174463372349867,0.1396484375
|
||||
model.language_model.layers.11.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,11,-6.131917740882786e-06,0.012295249513636653,0.240234375
|
||||
model.language_model.layers.11.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,11,0.6161422729492188,0.6511713445577767,1.1328125
|
||||
model.language_model.layers.11.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,11,-2.9925813798232885e-05,0.012163861956325002,0.2412109375
|
||||
model.language_model.layers.11.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,11,7.5102475781685385e-06,0.014888869682307635,0.1552734375
|
||||
model.language_model.layers.12.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,12,0.26889727637171745,0.3292817300709189,0.62890625
|
||||
model.language_model.layers.12.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,12,0.00032632781884700535,0.09772404322550603,1.234375
|
||||
model.language_model.layers.12.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,12,-3.496551513671875,4.058185988311173,7.46875
|
||||
model.language_model.layers.12.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,12,-0.00018116199635187513,0.020444603965792407,0.1484375
|
||||
model.language_model.layers.12.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,12,-4.0317171379200545e-05,0.010743962668614463,0.0791015625
|
||||
model.language_model.layers.12.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,12,7.781599651781679e-06,0.016665809696906838,0.26953125
|
||||
model.language_model.layers.12.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,12,-4.675414117657258e-05,0.012616807270715447,0.1826171875
|
||||
model.language_model.layers.12.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,12,1.4729152200219225e-06,0.011476801540265865,0.298828125
|
||||
model.language_model.layers.12.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,12,1.0457843859524678e-06,0.008423235529643944,0.1884765625
|
||||
model.language_model.layers.12.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,12,8.18576424821083e-06,0.009472447824563206,0.23828125
|
||||
model.language_model.layers.12.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,12,1.0104942249453304e-06,0.008791148082624213,0.11376953125
|
||||
model.language_model.layers.12.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,12,0.031112442025914788,0.12879091431054787,0.4765625
|
||||
model.language_model.layers.13.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,13,0.232861602678895,0.28740931489560884,0.6328125
|
||||
model.language_model.layers.13.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,13,-0.0012397262783755043,0.0771414825749899,0.67578125
|
||||
model.language_model.layers.13.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,13,-1.475341796875,2.39097532809558,5.375
|
||||
model.language_model.layers.13.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,13,-0.0002199913735694281,0.016203232395463157,0.1455078125
|
||||
model.language_model.layers.13.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,13,-0.00017839931808794063,0.009952179504275616,0.10009765625
|
||||
model.language_model.layers.13.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,13,2.9179273165786993e-06,0.015835845289678272,0.2158203125
|
||||
model.language_model.layers.13.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,13,5.881411352340327e-06,0.012331320856281865,0.1787109375
|
||||
model.language_model.layers.13.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,13,-4.353393460746098e-06,0.011825279279174,0.31640625
|
||||
model.language_model.layers.13.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,13,2.896835891259483e-07,0.008557289286997264,0.271484375
|
||||
model.language_model.layers.13.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,13,1.0580261440249853e-05,0.009664119005780037,0.1962890625
|
||||
model.language_model.layers.13.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,13,-1.8727934769265098e-06,0.008881907189765451,0.119140625
|
||||
model.language_model.layers.13.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,13,0.04897372436244041,0.10805549733912552,0.380859375
|
||||
model.language_model.layers.14.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,14,0.21899888664484024,0.24888835819250885,0.515625
|
||||
model.language_model.layers.14.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,14,0.00037967084426782094,0.08331865162655389,0.734375
|
||||
model.language_model.layers.14.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,14,-2.2860107421875,3.3004761684662856,7.40625
|
||||
model.language_model.layers.14.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,14,-0.0001371101245695172,0.016105199807047717,0.154296875
|
||||
model.language_model.layers.14.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,14,-9.110311782478675e-05,0.01034784269424491,0.10986328125
|
||||
model.language_model.layers.14.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,14,-2.1270679876183446e-06,0.016039110546304334,0.314453125
|
||||
model.language_model.layers.14.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,14,-8.110897876532214e-06,0.012852117255934343,0.1728515625
|
||||
model.language_model.layers.14.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,14,2.762643928013028e-06,0.011928614818943932,0.421875
|
||||
model.language_model.layers.14.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,14,-3.7334664855249535e-06,0.008537178745857807,0.279296875
|
||||
model.language_model.layers.14.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,14,7.169518554585423e-06,0.01050483670409655,0.263671875
|
||||
model.language_model.layers.14.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,14,-8.076213993245956e-07,0.009027392030954798,0.13671875
|
||||
model.language_model.layers.14.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,14,0.09636163339018822,0.11019530606933525,0.328125
|
||||
model.language_model.layers.15.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,15,0.10911215841770172,0.12466131555593847,0.486328125
|
||||
model.language_model.layers.15.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,15,-1.7299953231148084e-06,0.009031443773640731,0.1962890625
|
||||
model.language_model.layers.15.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,15,1.3207215079648569e-05,0.010797157880584657,0.2451171875
|
||||
model.language_model.layers.15.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,15,-5.836836459302668e-07,0.009495428068480694,0.1826171875
|
||||
model.language_model.layers.15.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,15,0.160281787160784,0.17411452277120051,0.427734375
|
||||
model.language_model.layers.15.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,15,0.6101093292236328,0.6606615117915355,1.390625
|
||||
model.language_model.layers.15.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,15,1.0313123176453054e-05,0.011781971940935327,0.171875
|
||||
model.language_model.layers.15.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,15,3.2290907635151855e-06,0.01443958574981722,0.201171875
|
||||
model.language_model.layers.15.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,15,0.6063613891601562,0.6516914189821662,1.078125
|
||||
model.language_model.layers.15.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,15,-2.080716491325088e-05,0.013066587205176929,0.2333984375
|
||||
model.language_model.layers.15.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,15,1.1813592924748637e-05,0.01702842388331824,0.392578125
|
||||
model.language_model.layers.16.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,16,0.343410836532712,0.3555100649844507,0.71875
|
||||
model.language_model.layers.16.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,16,-0.0010187850248257746,0.12131959888922034,1.25
|
||||
model.language_model.layers.16.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,16,-1.6535491943359375,3.4306294710677805,5.28125
|
||||
model.language_model.layers.16.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,16,-0.00040711706062879216,0.019010974147942117,0.1474609375
|
||||
model.language_model.layers.16.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,16,-0.00016220749716922,0.011994693576622582,0.203125
|
||||
model.language_model.layers.16.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,16,2.9395824504673215e-06,0.01839649067123788,0.279296875
|
||||
model.language_model.layers.16.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,16,-1.3744430889505062e-05,0.013106653141967533,0.158203125
|
||||
model.language_model.layers.16.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,16,-1.1683691683583852e-06,0.011435638427761064,0.326171875
|
||||
model.language_model.layers.16.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,16,-1.4952454272538357e-06,0.009066413695231633,0.353515625
|
||||
model.language_model.layers.16.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,16,4.156314234681659e-05,0.011800534597486789,0.2578125
|
||||
model.language_model.layers.16.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,16,-1.4486515269278531e-06,0.009458331855335966,0.11572265625
|
||||
model.language_model.layers.16.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,16,0.22678589820861816,0.24283223012121077,0.546875
|
||||
model.language_model.layers.17.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,17,0.40909794718027115,0.41420397619623966,0.8515625
|
||||
model.language_model.layers.17.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,17,-0.0007928749170484176,0.1427199131150052,0.84375
|
||||
model.language_model.layers.17.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,17,-1.5863037109375,3.145992001369952,5.8125
|
||||
model.language_model.layers.17.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,17,-0.0002446489017984277,0.019432861311383113,0.2216796875
|
||||
model.language_model.layers.17.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,17,-0.00028408583852979064,0.011455641742157624,0.11669921875
|
||||
model.language_model.layers.17.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,17,-6.122790621556258e-06,0.018850718521384917,0.361328125
|
||||
model.language_model.layers.17.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,17,1.7430077882997497e-05,0.013284800321653467,0.1337890625
|
||||
model.language_model.layers.17.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,17,-1.8784842093881593e-06,0.011016520359717747,0.26171875
|
||||
model.language_model.layers.17.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,17,1.6134364031145907e-06,0.009095506394657166,0.26953125
|
||||
model.language_model.layers.17.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,17,6.465429543157655e-05,0.012932345230835652,0.396484375
|
||||
model.language_model.layers.17.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,17,-4.222612684595094e-06,0.009451093364267178,0.140625
|
||||
model.language_model.layers.17.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,17,0.2981297969818115,0.31212707224811526,0.58203125
|
||||
model.language_model.layers.18.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,18,0.4494924545288086,0.4545845805845828,0.87109375
|
||||
model.language_model.layers.18.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,18,0.000160603896650476,0.14188962547118888,0.9453125
|
||||
model.language_model.layers.18.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,18,-1.5224609375,4.044787719211163,5.46875
|
||||
model.language_model.layers.18.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,18,-0.000354912899865667,0.018177394584053246,0.1865234375
|
||||
model.language_model.layers.18.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,18,-0.000348121350555175,0.011650516406628657,0.1474609375
|
||||
model.language_model.layers.18.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,18,-2.801285425906771e-06,0.01932307704334137,0.302734375
|
||||
model.language_model.layers.18.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,18,3.0413390159751696e-05,0.013580282672648016,0.2099609375
|
||||
model.language_model.layers.18.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,18,5.721753637381866e-07,0.010791681091114965,0.396484375
|
||||
model.language_model.layers.18.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,18,-6.59695743752906e-07,0.009220799866825103,0.376953125
|
||||
model.language_model.layers.18.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,18,8.449190781037264e-05,0.013813956126686252,0.2734375
|
||||
model.language_model.layers.18.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,18,-3.3797074411766443e-06,0.00960778958739809,0.10302734375
|
||||
model.language_model.layers.18.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,18,0.3679823362035677,0.38049220561196123,0.67578125
|
||||
model.language_model.layers.19.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,19,0.1011038941796869,0.2188507118664161,0.7421875
|
||||
model.language_model.layers.19.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,19,2.4411962869774297e-06,0.009416140657580934,0.38671875
|
||||
model.language_model.layers.19.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,19,6.661238714663713e-05,0.01359737914612424,0.349609375
|
||||
model.language_model.layers.19.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,19,-9.332812076388979e-07,0.00973094482190661,0.1240234375
|
||||
model.language_model.layers.19.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,19,0.3443268625997007,0.3526435167969441,0.546875
|
||||
model.language_model.layers.19.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,19,0.43672245740890503,0.5101940111828327,1.2421875
|
||||
model.language_model.layers.19.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,19,-2.6758214080602194e-07,0.01115193103746146,0.1806640625
|
||||
model.language_model.layers.19.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,19,-8.259318417455641e-07,0.017919884244499158,0.341796875
|
||||
model.language_model.layers.19.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,19,0.4400367736816406,0.5185481967268359,1.0625
|
||||
model.language_model.layers.19.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,19,6.793664314847653e-05,0.012547566379924864,0.15234375
|
||||
model.language_model.layers.19.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,19,-3.25356766062157e-05,0.021533393668245943,0.546875
|
||||
model.language_model.layers.2.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,2,0.2333534136414528,0.25578013108533854,0.7421875
|
||||
model.language_model.layers.2.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,2,-0.0009711920154131803,0.0699787295154593,1.203125
|
||||
model.language_model.layers.2.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,2,-1.531646728515625,2.7997863603942768,5.28125
|
||||
model.language_model.layers.2.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,2,0.00012737364886561409,0.020580835741280747,0.1904296875
|
||||
model.language_model.layers.2.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,2,-0.00021474828716350203,0.012867150783559326,0.1494140625
|
||||
model.language_model.layers.2.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,2,-7.305419878167231e-06,0.015296609635115742,0.2353515625
|
||||
model.language_model.layers.2.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,2,1.8538136823925688e-05,0.014831222973712392,0.1708984375
|
||||
model.language_model.layers.2.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,2,6.226499901958605e-06,0.012155681350565658,0.396484375
|
||||
model.language_model.layers.2.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,2,-4.951895974369495e-07,0.007964106725808892,0.3671875
|
||||
model.language_model.layers.2.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,2,6.107943125547091e-05,0.012312766730790738,0.1728515625
|
||||
model.language_model.layers.2.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,2,-2.1764107683565404e-06,0.007951716538597723,0.14453125
|
||||
model.language_model.layers.2.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,2,0.10721788182854652,0.14926779821439945,0.6171875
|
||||
model.language_model.layers.20.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,20,0.39076799154281616,0.4001752339553643,0.9609375
|
||||
model.language_model.layers.20.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,20,9.075260383421362e-05,0.1398477462038532,0.9609375
|
||||
model.language_model.layers.20.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,20,-1.0079345703125,3.4398750076762483,5.3125
|
||||
model.language_model.layers.20.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,20,-0.0004249333025541091,0.020555450330303295,0.2314453125
|
||||
model.language_model.layers.20.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,20,-0.00038497815725691,0.013473339189517963,0.1669921875
|
||||
model.language_model.layers.20.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,20,1.3699399179049887e-06,0.01875879440105595,0.3984375
|
||||
model.language_model.layers.20.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,20,1.5797673497242037e-05,0.013929090997023318,0.27734375
|
||||
model.language_model.layers.20.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,20,3.4517880942476487e-06,0.011156345981577036,0.39453125
|
||||
model.language_model.layers.20.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,20,-1.2719515250439694e-06,0.00957307701852322,0.373046875
|
||||
model.language_model.layers.20.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,20,4.00437334413904e-05,0.013427776388924359,0.265625
|
||||
model.language_model.layers.20.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,20,-3.335567268344132e-07,0.00994430727183077,0.1455078125
|
||||
model.language_model.layers.20.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,20,0.3715863972902298,0.37970192231618044,0.62109375
|
||||
model.language_model.layers.21.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,21,0.44352924823760986,0.44916903078759246,1.203125
|
||||
model.language_model.layers.21.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,21,0.0006429380024049655,0.14050966017891983,1.0234375
|
||||
model.language_model.layers.21.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,21,0.08379364013671875,3.303196151328201,5.34375
|
||||
model.language_model.layers.21.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,21,-0.00030593817302815296,0.021425148318092637,0.234375
|
||||
model.language_model.layers.21.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,21,-0.0002982664101409682,0.012089513394843168,0.11767578125
|
||||
model.language_model.layers.21.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,21,-2.9257048916112027e-06,0.018411063099048608,0.39453125
|
||||
model.language_model.layers.21.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,21,-3.6805679609420387e-05,0.01427042687438596,0.2119140625
|
||||
model.language_model.layers.21.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,21,1.0814873818025582e-05,0.011783994920447497,0.3359375
|
||||
model.language_model.layers.21.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,21,3.7257452291522553e-06,0.009740438269728707,0.283203125
|
||||
model.language_model.layers.21.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,21,4.347164541715296e-05,0.01366394178359644,0.376953125
|
||||
model.language_model.layers.21.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,21,-1.9860518428394557e-06,0.010192647161382656,0.2001953125
|
||||
model.language_model.layers.21.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,21,0.39837953448295593,0.4037610412929416,0.65234375
|
||||
model.language_model.layers.22.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,22,0.31986570358276367,0.3279013006012118,1.0234375
|
||||
model.language_model.layers.22.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,22,0.0008979494676661185,0.12143289672117986,0.98828125
|
||||
model.language_model.layers.22.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,22,0.264739990234375,3.6225085020426246,8.0625
|
||||
model.language_model.layers.22.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,22,-0.0006314610018307576,0.023313385449278027,0.25390625
|
||||
model.language_model.layers.22.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,22,-0.0002720628299357486,0.013776980313481728,0.1943359375
|
||||
model.language_model.layers.22.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,22,3.8010201965493298e-06,0.016764196896689124,0.4296875
|
||||
model.language_model.layers.22.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,22,-4.9468020722115424e-05,0.014891482667100319,0.248046875
|
||||
model.language_model.layers.22.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,22,1.3340557990845076e-06,0.012518738787194348,0.55078125
|
||||
model.language_model.layers.22.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,22,-3.2801688128341906e-06,0.0096870165213663,0.39453125
|
||||
model.language_model.layers.22.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,22,6.999503861985796e-05,0.013968549839218811,0.396484375
|
||||
model.language_model.layers.22.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,22,3.97610155118804e-06,0.010452139908499549,0.1513671875
|
||||
model.language_model.layers.22.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,22,0.38950735330581665,0.3927199440840361,0.65234375
|
||||
model.language_model.layers.23.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,23,0.16185743268579245,0.22320587027448527,0.87890625
|
||||
model.language_model.layers.23.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,23,2.200153014944121e-06,0.009198179693873125,0.9140625
|
||||
model.language_model.layers.23.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,23,5.746251235838462e-05,0.01555093104020481,0.4296875
|
||||
model.language_model.layers.23.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,23,-1.4030004445502939e-06,0.01144567089869159,0.255859375
|
||||
model.language_model.layers.23.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,23,0.2778373658657074,0.28538803537333535,0.796875
|
||||
model.language_model.layers.23.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,23,0.3784341812133789,0.45290137200158004,1.296875
|
||||
model.language_model.layers.23.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,23,-3.258974928144198e-06,0.010211249530655503,0.1923828125
|
||||
model.language_model.layers.23.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,23,4.249854031940226e-06,0.014946339830350153,0.36328125
|
||||
model.language_model.layers.23.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,23,0.3749351501464844,0.4735104392931629,0.9921875
|
||||
model.language_model.layers.23.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,23,2.8210594946158612e-05,0.013931200188097417,0.3046875
|
||||
model.language_model.layers.23.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,23,-4.7418587849623606e-06,0.0201368708703942,0.365234375
|
||||
model.language_model.layers.3.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,3,0.14525676146149635,0.1640384010422585,0.7109375
|
||||
model.language_model.layers.3.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,3,-3.666761125140459e-07,0.007925089389266608,0.322265625
|
||||
model.language_model.layers.3.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,3,-7.894271173424181e-06,0.011534900694163063,0.158203125
|
||||
model.language_model.layers.3.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,3,2.630765779604652e-06,0.008013811534961623,0.11865234375
|
||||
model.language_model.layers.3.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,3,0.046131135895848274,0.12433499188529953,0.51171875
|
||||
model.language_model.layers.3.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,3,0.4114065170288086,0.46410242264785667,1.59375
|
||||
model.language_model.layers.3.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,3,-1.2478612181981985e-05,0.012234452222983938,0.142578125
|
||||
model.language_model.layers.3.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,3,7.17851804013514e-07,0.011753633554238597,0.380859375
|
||||
model.language_model.layers.3.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,3,0.4230690002441406,0.4549413197073127,1.1796875
|
||||
model.language_model.layers.3.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,3,4.508237743732629e-05,0.016050479790290943,0.3203125
|
||||
model.language_model.layers.3.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,3,4.0868584337827585e-06,0.011898094706242741,0.1396484375
|
||||
model.language_model.layers.4.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,4,0.2650436982512474,0.28638167705127865,0.71875
|
||||
model.language_model.layers.4.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,4,3.396201100258622e-05,0.09066321443526168,0.8125
|
||||
model.language_model.layers.4.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,4,-2.729248046875,3.8747778798092267,5.96875
|
||||
model.language_model.layers.4.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,4,-0.00011186926542450237,0.02038835749963588,0.1728515625
|
||||
model.language_model.layers.4.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,4,-0.00019476671903717602,0.010351199541655327,0.1259765625
|
||||
model.language_model.layers.4.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,4,1.2304641167435136e-06,0.01636563283090178,0.29296875
|
||||
model.language_model.layers.4.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,4,2.0802682229954816e-05,0.013888945005603458,0.166015625
|
||||
model.language_model.layers.4.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,4,-2.994475370804439e-06,0.011735409435323977,0.396484375
|
||||
model.language_model.layers.4.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,4,-4.1091264927226615e-06,0.00801339942052328,0.2451171875
|
||||
model.language_model.layers.4.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,4,1.6863014756946307e-05,0.011355842511871286,0.2080078125
|
||||
model.language_model.layers.4.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,4,1.027859200892813e-06,0.008108758248932731,0.10302734375
|
||||
model.language_model.layers.4.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,4,0.0494824443012476,0.12864073061216597,0.5234375
|
||||
model.language_model.layers.5.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,5,0.22753581404685974,0.2630511444050265,0.67578125
|
||||
model.language_model.layers.5.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,5,0.00038598675951107,0.08068910936433069,1.1015625
|
||||
model.language_model.layers.5.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,5,-2.72894287109375,3.6278404173514955,6.6875
|
||||
model.language_model.layers.5.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,5,-0.0001803523750822933,0.018341404674334767,0.1455078125
|
||||
model.language_model.layers.5.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,5,-0.00022759488123824667,0.010625336231416064,0.10400390625
|
||||
model.language_model.layers.5.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,5,3.021544299360978e-06,0.01587561641265017,0.3046875
|
||||
model.language_model.layers.5.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,5,1.619939023567163e-05,0.013311542875305954,0.1533203125
|
||||
model.language_model.layers.5.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,5,7.180698611445463e-06,0.011705867874160678,0.404296875
|
||||
model.language_model.layers.5.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,5,-3.832363420119428e-06,0.008129977884391225,0.25390625
|
||||
model.language_model.layers.5.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,5,9.510474803579592e-06,0.011259738864004266,0.1318359375
|
||||
model.language_model.layers.5.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,5,4.334927638522066e-08,0.008256794109219937,0.08349609375
|
||||
model.language_model.layers.5.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,5,0.0527042793110013,0.15491478237820172,0.578125
|
||||
model.language_model.layers.6.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,6,0.18705136142671108,0.23969713140560028,0.609375
|
||||
model.language_model.layers.6.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,6,0.0003755964487671311,0.07146913561403648,0.75390625
|
||||
model.language_model.layers.6.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,6,-1.911773681640625,2.9807612149394664,6.15625
|
||||
model.language_model.layers.6.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,6,-0.0002217653484422044,0.016199257108442264,0.1337890625
|
||||
model.language_model.layers.6.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,6,-0.0002016041112824496,0.011786386085972541,0.1455078125
|
||||
model.language_model.layers.6.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,6,3.843747654041589e-06,0.015354747770723804,0.21484375
|
||||
model.language_model.layers.6.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,6,-9.215484209998404e-06,0.013138127645844377,0.193359375
|
||||
model.language_model.layers.6.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,6,4.4970565675102325e-06,0.012014377201321634,0.310546875
|
||||
model.language_model.layers.6.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,6,1.5256064373652192e-06,0.008180750870201545,0.2373046875
|
||||
model.language_model.layers.6.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,6,1.2075660691879235e-05,0.010919152343969056,0.2333984375
|
||||
model.language_model.layers.6.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,6,2.4837448766669578e-06,0.008417535304985934,0.09814453125
|
||||
model.language_model.layers.6.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,6,0.04677234560949728,0.14343473354286257,0.56640625
|
||||
model.language_model.layers.7.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,7,0.01859602495096624,0.09770610871894504,0.53515625
|
||||
model.language_model.layers.7.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,7,-2.069178868239751e-06,0.008313764288599495,0.30859375
|
||||
model.language_model.layers.7.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,7,-5.941976052590073e-06,0.010130333371973213,0.1865234375
|
||||
model.language_model.layers.7.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,7,-1.921608185482948e-06,0.008620142534023984,0.10791015625
|
||||
model.language_model.layers.7.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,7,0.03399748355150223,0.17481274831958912,0.625
|
||||
model.language_model.layers.7.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.k_norm.weight,7,0.5165120363235474,0.5423433220393242,1.09375
|
||||
model.language_model.layers.7.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.k_proj.weight,7,6.845853454978679e-06,0.011321329966324158,0.12353515625
|
||||
model.language_model.layers.7.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.self_attn.o_proj.weight,7,1.0774090243398544e-05,0.011294127412937538,0.236328125
|
||||
model.language_model.layers.7.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,model.language_model.layers.*.self_attn.q_norm.weight,7,0.5257186889648438,0.5454621661599409,1.1953125
|
||||
model.language_model.layers.7.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,model.language_model.layers.*.self_attn.q_proj.weight,7,1.3410202473564982e-06,0.013098970846866802,0.26171875
|
||||
model.language_model.layers.7.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,model.language_model.layers.*.self_attn.v_proj.weight,7,-4.638805624887499e-06,0.011705499680617187,0.1376953125
|
||||
model.language_model.layers.8.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,8,0.21140070259571075,0.29139364044388155,0.60546875
|
||||
model.language_model.layers.8.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,8,-0.0008830481529571443,0.07194979471045664,0.734375
|
||||
model.language_model.layers.8.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,8,-2.32696533203125,3.1389153700014405,5.40625
|
||||
model.language_model.layers.8.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,8,-7.89401556176017e-05,0.01640450978485818,0.1240234375
|
||||
model.language_model.layers.8.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,8,-0.00019124166351502936,0.009413019985098999,0.0830078125
|
||||
model.language_model.layers.8.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,8,-1.0459140759450092e-06,0.01570452643614291,0.2265625
|
||||
model.language_model.layers.8.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,8,6.922933192797694e-06,0.012484979769894367,0.2275390625
|
||||
model.language_model.layers.8.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,8,2.282237288962069e-07,0.011714720270195043,0.3515625
|
||||
model.language_model.layers.8.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,8,1.9946693415948355e-06,0.008243732742036089,0.29296875
|
||||
model.language_model.layers.8.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,8,-1.3840076219445603e-06,0.009590909948543426,0.15625
|
||||
model.language_model.layers.8.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,8,3.841095268237252e-06,0.00861522200030427,0.09130859375
|
||||
model.language_model.layers.8.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,8,0.010415758239105344,0.18284911199552537,0.66015625
|
||||
model.language_model.layers.9.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.input_layernorm.weight,9,0.19137806724756956,0.296521861015528,0.57421875
|
||||
model.language_model.layers.9.linear_attn.conv1d.weight,"(6144, 1, 4)",3,BF16,24576,49152,model.language_model.layers.*.linear_attn.conv1d.weight,9,-0.000410568358773844,0.06652172297471458,1.3359375
|
||||
model.language_model.layers.9.linear_attn.dt_bias,"(16,)",1,BF16,16,32,model.language_model.layers.*.linear_attn.dt_bias,9,-2.45068359375,3.1450531520506346,4.90625
|
||||
model.language_model.layers.9.linear_attn.in_proj_a.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_a.weight,9,-0.00011285979962849524,0.018485456249646506,0.251953125
|
||||
model.language_model.layers.9.linear_attn.in_proj_b.weight,"(16, 2048)",2,BF16,32768,65536,model.language_model.layers.*.linear_attn.in_proj_b.weight,9,-0.00012505355687153497,0.012157132620716115,0.10009765625
|
||||
model.language_model.layers.9.linear_attn.in_proj_qkv.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,9,-1.83865233245677e-06,0.015200480417789385,0.30078125
|
||||
model.language_model.layers.9.linear_attn.in_proj_z.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.in_proj_z.weight,9,-2.0241770911732723e-05,0.012578215645634584,0.1982421875
|
||||
model.language_model.layers.9.linear_attn.out_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,model.language_model.layers.*.linear_attn.out_proj.weight,9,-4.150400907093621e-06,0.01150305952966105,0.30078125
|
||||
model.language_model.layers.9.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.down_proj.weight,9,-2.7189171275125937e-06,0.008310183109864908,0.181640625
|
||||
model.language_model.layers.9.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.gate_proj.weight,9,-9.032502292217909e-06,0.00903253707533701,0.1669921875
|
||||
model.language_model.layers.9.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,model.language_model.layers.*.mlp.up_proj.weight,9,-5.926681719845353e-07,0.008713214518297324,0.0927734375
|
||||
model.language_model.layers.9.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.layers.*.post_attention_layernorm.weight,9,0.0009866869077086449,0.17870262924322744,0.6484375
|
||||
model.language_model.norm.weight,"(2048,)",1,BF16,2048,4096,model.language_model.norm.weight,-1,2.536298483610153,2.5765266808418485,5.375
|
||||
model.visual.blocks.0.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,0,0.0056819673627614975,0.4223296682397815,3.515625
|
||||
model.visual.blocks.0.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,0,2.8629653116164083e-05,0.013697778976293876,0.353515625
|
||||
model.visual.blocks.0.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,0,-0.0006211028861192366,0.8058543942936214,6.3125
|
||||
model.visual.blocks.0.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,0,2.288061551664981e-05,0.02266044393425182,0.640625
|
||||
model.visual.blocks.0.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,0,-2.248420387506485,2.785209948019622,9.25
|
||||
model.visual.blocks.0.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,0,-1.672273021108577e-05,0.01791097511042567,0.2890625
|
||||
model.visual.blocks.0.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,0,-0.010476455092430115,0.8661675219608709,10.875
|
||||
model.visual.blocks.0.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,0,7.985690316429744e-05,0.01681887613844911,0.26171875
|
||||
model.visual.blocks.0.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,0,-0.0007249293848872185,0.21335208117994894,3.8125
|
||||
model.visual.blocks.0.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,0,0.5173702943138778,0.9214386100936024,4.03125
|
||||
model.visual.blocks.0.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,0,0.002009127005294431,0.26724968567574725,4.90625
|
||||
model.visual.blocks.0.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,0,2.059753055276815,3.6960545327961047,16.875
|
||||
model.visual.blocks.1.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,1,0.03787066414952278,0.7577770028883627,7.6875
|
||||
model.visual.blocks.1.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,1,1.0665584427915764e-05,0.013909349158950824,0.35546875
|
||||
model.visual.blocks.1.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,1,0.005118881289187509,0.6554565198492014,6.96875
|
||||
model.visual.blocks.1.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,1,-2.2588287930780382e-05,0.022658501611919644,0.50390625
|
||||
model.visual.blocks.1.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,1,-1.8916514813899994,2.255432446216723,4.78125
|
||||
model.visual.blocks.1.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,1,-0.00015137788937700248,0.018427025670974255,0.384765625
|
||||
model.visual.blocks.1.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,1,-0.02125891763716936,0.48354598792262593,8.8125
|
||||
model.visual.blocks.1.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,1,4.1878749072722654e-05,0.016416665982819812,0.59375
|
||||
model.visual.blocks.1.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,1,-0.006031324039213359,0.20026363711287062,3.125
|
||||
model.visual.blocks.1.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,1,0.5489383637905121,0.676293812386044,2.015625
|
||||
model.visual.blocks.1.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,1,-0.0010115213226526976,0.226877326061835,2.875
|
||||
model.visual.blocks.1.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,1,0.9165963572449982,1.1850193608703092,3.484375
|
||||
model.visual.blocks.10.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,10,0.0013865428045392036,0.3155090787478211,4.96875
|
||||
model.visual.blocks.10.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,10,-4.397379537918189e-06,0.012832670932776341,0.1201171875
|
||||
model.visual.blocks.10.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,10,-0.00729733055535083,0.4923193573431798,6.5
|
||||
model.visual.blocks.10.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,10,2.8232800771503066e-06,0.01617202900396593,0.234375
|
||||
model.visual.blocks.10.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,10,-1.645782470703125,1.7657915613666026,7.8125
|
||||
model.visual.blocks.10.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,10,-5.362843342912593e-05,0.01571570549852373,0.28125
|
||||
model.visual.blocks.10.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,10,-0.002808693330734968,0.20955213395129205,2.609375
|
||||
model.visual.blocks.10.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,10,-4.729164932659735e-06,0.012869168141970897,0.62109375
|
||||
model.visual.blocks.10.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,10,0.003513883682899177,0.19258665385120421,2.375
|
||||
model.visual.blocks.10.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,10,0.9836215972900391,1.0083027671073959,1.8125
|
||||
model.visual.blocks.10.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,10,0.000604923814535141,0.2915448741855378,3.953125
|
||||
model.visual.blocks.10.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,10,1.1706929206848145,1.1883884565165725,2.40625
|
||||
model.visual.blocks.11.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,11,-0.0005531555507332087,0.27908196456952605,4.96875
|
||||
model.visual.blocks.11.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,11,-3.6013991529804112e-06,0.012820869846561344,0.0986328125
|
||||
model.visual.blocks.11.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,11,0.00846743700094521,0.48997687252328,6.90625
|
||||
model.visual.blocks.11.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,11,9.277315033732561e-06,0.015729353782592298,0.1767578125
|
||||
model.visual.blocks.11.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,11,-1.6464985609054565,1.722120238661639,6.03125
|
||||
model.visual.blocks.11.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,11,-3.589357073445487e-05,0.015698389546222692,0.328125
|
||||
model.visual.blocks.11.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,11,-0.00445837527513504,0.2281571578445295,3.03125
|
||||
model.visual.blocks.11.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,11,-8.2175139034385e-07,0.014062640336574176,0.5390625
|
||||
model.visual.blocks.11.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,11,0.0063120657578110695,0.19104643837290222,3.1875
|
||||
model.visual.blocks.11.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,11,1.0900201797485352,1.11257957339,1.890625
|
||||
model.visual.blocks.11.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,11,0.02426927164196968,0.38187088189437013,4.46875
|
||||
model.visual.blocks.11.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,11,1.3532802313566208,1.381797140286228,3.578125
|
||||
model.visual.blocks.12.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,12,-0.0026703281328082085,0.1916711651620828,2.046875
|
||||
model.visual.blocks.12.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,12,4.0955608962556234e-07,0.012925623041989713,0.1455078125
|
||||
model.visual.blocks.12.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,12,0.008885961996080974,0.5122118241036924,6.5
|
||||
model.visual.blocks.12.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,12,-9.870303735948216e-06,0.01573086232798125,0.16015625
|
||||
model.visual.blocks.12.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,12,-1.675450623035431,1.7334162851669006,3.453125
|
||||
model.visual.blocks.12.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,12,-8.740557205228275e-06,0.014830930647847022,0.408203125
|
||||
model.visual.blocks.12.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,12,-0.005204610526561737,0.26138935010758596,2.9375
|
||||
model.visual.blocks.12.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,12,1.7673791958203178e-05,0.01386013349041631,0.40234375
|
||||
model.visual.blocks.12.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,12,0.010884693590924144,0.2269366679227373,4.4375
|
||||
model.visual.blocks.12.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,12,1.1681642532348633,1.1884347634813108,2.390625
|
||||
model.visual.blocks.12.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,12,0.01296179462224245,0.4114089854061476,5.6875
|
||||
model.visual.blocks.12.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,12,1.344668447971344,1.3677649716188205,2.328125
|
||||
model.visual.blocks.13.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,13,-0.0028251297771930695,0.24074961333040454,2.34375
|
||||
model.visual.blocks.13.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,13,2.5278697873831746e-06,0.01285373134719242,0.1064453125
|
||||
model.visual.blocks.13.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,13,0.004933949753952523,0.508857160517705,7.15625
|
||||
model.visual.blocks.13.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,13,1.9118003685703044e-07,0.015042486712391608,0.1669921875
|
||||
model.visual.blocks.13.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,13,-1.7249782234430313,1.7699195917996853,3.46875
|
||||
model.visual.blocks.13.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,13,-3.6795391052404924e-05,0.014446679506643701,0.416015625
|
||||
model.visual.blocks.13.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,13,-0.0022364631295204163,0.27349516648220606,3.859375
|
||||
model.visual.blocks.13.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,13,5.478725110724669e-06,0.013285561268431308,0.341796875
|
||||
model.visual.blocks.13.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,13,0.015306847577448934,0.27487751521193093,6.1875
|
||||
model.visual.blocks.13.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,13,1.268209457397461,1.287942725109198,2.359375
|
||||
model.visual.blocks.13.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,13,-0.003963736351579428,0.42726025831730696,5.21875
|
||||
model.visual.blocks.13.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,13,1.3601853847503662,1.3767604243470128,2.015625
|
||||
model.visual.blocks.14.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,14,-0.003052172716706991,0.21088079614630725,3.375
|
||||
model.visual.blocks.14.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,14,-1.1134785495858779e-06,0.012735885919582902,0.1611328125
|
||||
model.visual.blocks.14.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,14,0.003906862887864311,0.5396591512615192,7.15625
|
||||
model.visual.blocks.14.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,14,1.169477530685766e-05,0.015290766734037807,0.154296875
|
||||
model.visual.blocks.14.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,14,-1.36743825674057,1.416465957664333,3.328125
|
||||
model.visual.blocks.14.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,14,-0.00016947123852519008,0.013272032998930029,0.1474609375
|
||||
model.visual.blocks.14.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,14,-0.0021008243784308434,0.17204831815289337,2.734375
|
||||
model.visual.blocks.14.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,14,2.1631967396136542e-06,0.013681268914449974,0.44921875
|
||||
model.visual.blocks.14.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,14,0.01591563131660223,0.312620166366634,8.25
|
||||
model.visual.blocks.14.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,14,1.3175182342529297,1.3302436689183337,2.015625
|
||||
model.visual.blocks.14.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,14,0.1486361138522625,1.1076539753453722,4.25
|
||||
model.visual.blocks.14.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,14,3.054824598133564,3.133953286674968,5.03125
|
||||
model.visual.blocks.15.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,15,-0.003838975913822651,0.17870155036895466,2.046875
|
||||
model.visual.blocks.15.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,15,2.1557483374512465e-06,0.012738003993849507,0.1748046875
|
||||
model.visual.blocks.15.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,15,-0.0070781482791062444,0.5481550569364706,8.1875
|
||||
model.visual.blocks.15.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,15,1.8805244741589473e-06,0.015028560293926872,0.138671875
|
||||
model.visual.blocks.15.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,15,-1.2592892870306969,1.3363415167024717,3.828125
|
||||
model.visual.blocks.15.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,15,-0.0001719071445260706,0.013613661762691076,0.150390625
|
||||
model.visual.blocks.15.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,15,-0.004280758090317249,0.22024041578351156,1.3203125
|
||||
model.visual.blocks.15.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,15,-1.4436593793638874e-05,0.0138859545324591,0.4375
|
||||
model.visual.blocks.15.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,15,0.01718392875045538,0.3302859808392976,8.375
|
||||
model.visual.blocks.15.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,15,1.3481388092041016,1.357684000352654,2.09375
|
||||
model.visual.blocks.15.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,15,0.21574118733406067,1.6791112424771022,4.59375
|
||||
model.visual.blocks.15.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,15,4.459657579660416,4.584192991018382,6.90625
|
||||
model.visual.blocks.16.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,16,-0.001141368702519685,0.17187718118886977,1.4765625
|
||||
model.visual.blocks.16.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,16,-9.010508974505171e-07,0.013030151340051142,0.11572265625
|
||||
model.visual.blocks.16.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,16,-0.003146370562414328,0.5747692983747378,7.875
|
||||
model.visual.blocks.16.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,16,7.913643638816451e-06,0.014997625709589514,0.19140625
|
||||
model.visual.blocks.16.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,16,-1.254018947482109,1.3324690973240458,4.0625
|
||||
model.visual.blocks.16.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,16,-0.0001544157969156773,0.01342578177786983,0.1455078125
|
||||
model.visual.blocks.16.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,16,-0.002572661265730858,0.2296740317693101,2.046875
|
||||
model.visual.blocks.16.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,16,-1.2149443716483188e-05,0.013883817502173138,0.345703125
|
||||
model.visual.blocks.16.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,16,0.022709594340994954,0.37735895627898264,9.8125
|
||||
model.visual.blocks.16.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,16,1.4540157318115234,1.4634179653266648,2.1875
|
||||
model.visual.blocks.16.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,16,0.18935611844062805,2.0355711199787074,6.0
|
||||
model.visual.blocks.16.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,16,5.341764807701111,5.473404786269683,7.90625
|
||||
model.visual.blocks.17.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,17,0.0010246038436889648,0.2094856224874406,1.5859375
|
||||
model.visual.blocks.17.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,17,-2.305236045496706e-06,0.013429564136500571,0.2001953125
|
||||
model.visual.blocks.17.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,17,0.0007098486336568991,0.5673033733273393,7.84375
|
||||
model.visual.blocks.17.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,17,-1.9605784129887263e-06,0.014858123558947355,0.1181640625
|
||||
model.visual.blocks.17.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,17,-1.2608060913626105,1.3344331944274648,4.96875
|
||||
model.visual.blocks.17.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,17,-0.00024945183689631736,0.01327430448006607,0.1513671875
|
||||
model.visual.blocks.17.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,17,-0.0037037022411823273,0.3233218477507246,3.640625
|
||||
model.visual.blocks.17.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,17,-1.5574131797709e-05,0.013947912596998567,0.2890625
|
||||
model.visual.blocks.17.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,17,0.02418094128370285,0.36228133811509267,8.8125
|
||||
model.visual.blocks.17.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,17,1.5454788208007812,1.5540872689932543,2.34375
|
||||
model.visual.blocks.17.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,17,0.3789477199316025,2.3624701748229344,7.46875
|
||||
model.visual.blocks.17.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,17,6.084241628646851,6.227228139512012,8.4375
|
||||
model.visual.blocks.18.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,18,-0.0002320706844329834,0.2934008246430531,1.1953125
|
||||
model.visual.blocks.18.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,18,-6.532082189314448e-07,0.013284318756608127,0.158203125
|
||||
model.visual.blocks.18.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,18,-0.0057055339954483015,0.5908193339357068,9.25
|
||||
model.visual.blocks.18.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,18,4.928585952967968e-06,0.014708558443394285,0.2216796875
|
||||
model.visual.blocks.18.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,18,-1.2480670548975468,1.3291303175041957,5.0625
|
||||
model.visual.blocks.18.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,18,-0.00021408203113450902,0.013495977011313135,0.150390625
|
||||
model.visual.blocks.18.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,18,-0.006308715790510178,0.3856628539897182,4.3125
|
||||
model.visual.blocks.18.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,18,-3.2339109456755444e-05,0.014627635036042322,0.318359375
|
||||
model.visual.blocks.18.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,18,0.02438737452030182,0.412171430074191,9.875
|
||||
model.visual.blocks.18.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,18,1.5765914916992188,1.582837659856145,2.3125
|
||||
model.visual.blocks.18.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,18,0.4297136813402176,3.1746328527745495,10.125
|
||||
model.visual.blocks.18.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,18,8.43129649758339,8.632484963591324,12.0625
|
||||
model.visual.blocks.19.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,19,-0.004621189087629318,0.3139398128022543,1.328125
|
||||
model.visual.blocks.19.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,19,2.4595190205589645e-06,0.013508958055272125,0.1337890625
|
||||
model.visual.blocks.19.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,19,0.005580020369961858,0.5351795263988486,8.1875
|
||||
model.visual.blocks.19.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,19,-8.630430732773734e-06,0.014762091303607939,0.1318359375
|
||||
model.visual.blocks.19.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,19,-1.2422089144820347,1.3163568037560458,4.5625
|
||||
model.visual.blocks.19.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,19,-0.00018224952321353738,0.013614127660122474,0.16796875
|
||||
model.visual.blocks.19.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,19,-0.007145557086914778,0.4602625036781678,4.1875
|
||||
model.visual.blocks.19.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,19,-4.168556236101156e-05,0.014827426301555785,0.1640625
|
||||
model.visual.blocks.19.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,19,0.024289263412356377,0.38014348627552946,6.5625
|
||||
model.visual.blocks.19.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,19,1.5708999633789062,1.5778643421156753,2.46875
|
||||
model.visual.blocks.19.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,19,0.33808582834899426,3.3086580127585967,10.4375
|
||||
model.visual.blocks.19.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,19,10.046417891979218,10.27741256232265,14.5625
|
||||
model.visual.blocks.2.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,2,0.020055142042110674,0.3886311638709507,4.28125
|
||||
model.visual.blocks.2.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,2,-2.695733576807413e-06,0.013976514875259815,0.255859375
|
||||
model.visual.blocks.2.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,2,-0.013754264956029752,0.657041102431288,8.4375
|
||||
model.visual.blocks.2.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,2,2.5094986518444293e-06,0.020591849089717593,0.302734375
|
||||
model.visual.blocks.2.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,2,-2.050982117652893,2.5296018322840883,4.78125
|
||||
model.visual.blocks.2.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,2,3.021457652678636e-06,0.01653758038173641,0.35546875
|
||||
model.visual.blocks.2.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,2,-0.01768375840038061,0.3234429143823406,5.59375
|
||||
model.visual.blocks.2.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,2,2.5083720841499735e-06,0.014604041553001833,0.58203125
|
||||
model.visual.blocks.2.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,2,-0.0018650367856025696,0.1970889226403788,3.3125
|
||||
model.visual.blocks.2.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,2,0.6477055847644806,0.705977945861401,1.734375
|
||||
model.visual.blocks.2.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,2,0.0044156142976135015,0.1629384723954203,2.28125
|
||||
model.visual.blocks.2.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,2,1.049049954948714,1.2132500636992718,3.0
|
||||
model.visual.blocks.20.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,20,-0.007338074967265129,0.3249699629246988,1.9296875
|
||||
model.visual.blocks.20.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,20,3.2072206779254486e-07,0.014146542579561038,0.142578125
|
||||
model.visual.blocks.20.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,20,-0.0014855863992124796,0.5315386037202189,7.8125
|
||||
model.visual.blocks.20.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,20,-3.534431219938868e-06,0.01480353550453723,0.134765625
|
||||
model.visual.blocks.20.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,20,-1.2434953674674034,1.323903681900149,4.15625
|
||||
model.visual.blocks.20.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,20,-6.702584891227908e-05,0.013731709166772268,0.150390625
|
||||
model.visual.blocks.20.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,20,-0.0063021189998835325,0.5291946069594037,4.625
|
||||
model.visual.blocks.20.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,20,-4.8746930603832395e-05,0.01500120652131212,0.1533203125
|
||||
model.visual.blocks.20.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,20,0.0250953808426857,0.44610627333196545,6.28125
|
||||
model.visual.blocks.20.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,20,1.6392402648925781,1.6461320049763404,2.8125
|
||||
model.visual.blocks.20.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,20,0.0293462872505188,3.1623750669665887,10.25
|
||||
model.visual.blocks.20.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,20,11.327821254730225,11.56503368296681,15.75
|
||||
model.visual.blocks.21.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,21,-0.005094312131404877,0.3524695004655374,2.953125
|
||||
model.visual.blocks.21.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,21,-1.5384484597091852e-06,0.014542213556737195,0.1357421875
|
||||
model.visual.blocks.21.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,21,0.002228708937764168,0.5125241811949673,7.375
|
||||
model.visual.blocks.21.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,21,1.0859385383884165e-05,0.014994967131387578,0.173828125
|
||||
model.visual.blocks.21.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,21,-1.2665919363498688,1.3450251529553447,3.46875
|
||||
model.visual.blocks.21.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,21,7.30989772137064e-05,0.013954792610339387,0.2578125
|
||||
model.visual.blocks.21.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,21,0.003942374140024185,0.8887231783004025,5.4375
|
||||
model.visual.blocks.21.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,21,-5.30296018196319e-05,0.015118207294661507,0.20703125
|
||||
model.visual.blocks.21.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,21,0.03733087331056595,0.4753083526711642,5.8125
|
||||
model.visual.blocks.21.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,21,1.5873088836669922,1.595838376433378,3.1875
|
||||
model.visual.blocks.21.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,21,-0.07360374182462692,3.0850272367238394,17.125
|
||||
model.visual.blocks.21.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,21,12.204975128173828,12.389441211741739,15.875
|
||||
model.visual.blocks.22.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,22,0.005250638350844383,0.5808595524391316,3.5
|
||||
model.visual.blocks.22.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,22,2.0765462032112936e-06,0.014772226332934188,0.18359375
|
||||
model.visual.blocks.22.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,22,0.014487470247937987,0.4916721195343452,6.28125
|
||||
model.visual.blocks.22.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,22,-5.31823748935641e-06,0.01523024465727359,0.166015625
|
||||
model.visual.blocks.22.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,22,-1.2670710682868958,1.3412129116547036,3.875
|
||||
model.visual.blocks.22.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,22,0.0004379041676110876,0.014189801218711325,0.23828125
|
||||
model.visual.blocks.22.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,22,0.011957859620451927,0.7787687092193024,4.3125
|
||||
model.visual.blocks.22.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,22,-2.478536209907234e-05,0.014944462334069182,0.15234375
|
||||
model.visual.blocks.22.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,22,0.022316202521324158,0.533177673147604,6.6875
|
||||
model.visual.blocks.22.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,22,1.5854969024658203,1.5964364161053683,3.34375
|
||||
model.visual.blocks.22.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,22,-0.5513232499361038,2.664964048785582,17.875
|
||||
model.visual.blocks.22.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,22,10.946296691894531,11.013918537503798,14.5
|
||||
model.visual.blocks.23.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,23,0.004559919238090515,1.1180409459483411,6.09375
|
||||
model.visual.blocks.23.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,23,-1.3123207243115331e-08,0.018186976140920567,0.244140625
|
||||
model.visual.blocks.23.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,23,-0.0045881786694129305,0.4793571911685673,6.375
|
||||
model.visual.blocks.23.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,23,-1.0563681768512486e-05,0.0164637041064667,0.1484375
|
||||
model.visual.blocks.23.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,23,-1.1823353916406631,1.2596480183914525,3.390625
|
||||
model.visual.blocks.23.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,23,0.0006163056673373288,0.014160105140531405,0.84765625
|
||||
model.visual.blocks.23.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,23,-0.007630281150341034,0.7601651155732886,9.0
|
||||
model.visual.blocks.23.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,23,2.5521093651711957e-05,0.014476899120657829,0.69921875
|
||||
model.visual.blocks.23.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,23,0.01085265725851059,0.4646660501895986,7.125
|
||||
model.visual.blocks.23.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,23,1.530099868774414,1.5427840122256975,3.28125
|
||||
model.visual.blocks.23.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,23,-0.547552265226841,2.1320240363512677,9.1875
|
||||
model.visual.blocks.23.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,23,7.73004150390625,7.794828769449906,16.625
|
||||
model.visual.blocks.3.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,3,0.0055896323174238205,0.22725533239644694,4.21875
|
||||
model.visual.blocks.3.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,3,-8.324880100274434e-06,0.014204998605310243,0.306640625
|
||||
model.visual.blocks.3.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,3,-0.010611735711184641,0.6356958380251622,9.5
|
||||
model.visual.blocks.3.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,3,3.0863008081573794e-05,0.018493728387148186,0.26171875
|
||||
model.visual.blocks.3.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,3,-2.4899610206484795,2.8378251598762203,4.8125
|
||||
model.visual.blocks.3.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,3,0.00012417413540886414,0.016658970395077077,0.37109375
|
||||
model.visual.blocks.3.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,3,-0.0060559650883078575,0.2120297601458163,4.625
|
||||
model.visual.blocks.3.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,3,1.6447852463617762e-05,0.016438568045283906,0.55078125
|
||||
model.visual.blocks.3.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,3,0.0005567893385887146,0.21667533989354099,4.25
|
||||
model.visual.blocks.3.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,3,0.7782039642333984,0.8228475783365038,1.671875
|
||||
model.visual.blocks.3.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,3,0.007098435831721872,0.19464676342919898,3.40625
|
||||
model.visual.blocks.3.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,3,0.9266856936737895,0.974532049879689,2.0625
|
||||
model.visual.blocks.4.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,4,0.006985343061387539,0.16223469691833878,1.296875
|
||||
model.visual.blocks.4.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,4,1.8378288352272065e-07,0.01442967947584689,0.298828125
|
||||
model.visual.blocks.4.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,4,0.018887083482695743,0.5794811256502446,9.1875
|
||||
model.visual.blocks.4.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,4,-7.655587233305894e-06,0.018059560768096988,0.2109375
|
||||
model.visual.blocks.4.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,4,-2.366614764323458,2.6965073178001546,4.625
|
||||
model.visual.blocks.4.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,4,0.00016276028306286966,0.016659989978346723,0.283203125
|
||||
model.visual.blocks.4.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,4,-0.004570281831547618,0.14836421255117274,1.6796875
|
||||
model.visual.blocks.4.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,4,1.3213413660478901e-05,0.01714581722997102,0.58203125
|
||||
model.visual.blocks.4.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,4,0.0012555653520394117,0.20431990398495709,3.8125
|
||||
model.visual.blocks.4.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,4,0.7098050117492676,0.7350691020181069,1.390625
|
||||
model.visual.blocks.4.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,4,0.008271024329587817,0.1808966163691234,3.484375
|
||||
model.visual.blocks.4.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,4,0.9531378536485136,0.986081198167738,2.234375
|
||||
model.visual.blocks.5.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,5,0.005363904405385256,0.15417362578206295,1.453125
|
||||
model.visual.blocks.5.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,5,1.138475018258589e-06,0.01455602676419094,0.2021484375
|
||||
model.visual.blocks.5.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,5,-0.0042238371291508274,0.5847475682637407,9.0
|
||||
model.visual.blocks.5.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,5,-5.719098949252821e-06,0.017454819217105327,0.2451171875
|
||||
model.visual.blocks.5.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,5,-2.2070916444063187,2.5344800306111703,4.6875
|
||||
model.visual.blocks.5.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,5,0.00016606873607712125,0.016482466770763107,0.1982421875
|
||||
model.visual.blocks.5.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,5,-0.004116903990507126,0.14437497298756788,1.5390625
|
||||
model.visual.blocks.5.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,5,-2.0686353935546464e-06,0.01632456835844649,0.455078125
|
||||
model.visual.blocks.5.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,5,0.0018707644194364548,0.2150864657981796,4.1875
|
||||
model.visual.blocks.5.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,5,0.7877235412597656,0.8056694065263549,1.7109375
|
||||
model.visual.blocks.5.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,5,0.005293702706694603,0.21285220841568864,4.125
|
||||
model.visual.blocks.5.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,5,0.9217546582221985,0.9412579347437405,2.3125
|
||||
model.visual.blocks.6.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,6,-0.0002876807702705264,0.18460897803677073,1.8125
|
||||
model.visual.blocks.6.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,6,-1.6337773984381743e-06,0.01408807292678129,0.1865234375
|
||||
model.visual.blocks.6.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,6,0.0054503006394952536,0.5035026330586864,7.5625
|
||||
model.visual.blocks.6.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,6,-2.637964617714187e-06,0.017114318284296055,0.185546875
|
||||
model.visual.blocks.6.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,6,-2.073364906013012,2.3344424149697662,4.59375
|
||||
model.visual.blocks.6.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,6,0.00012550517553477913,0.016654967329708652,0.259765625
|
||||
model.visual.blocks.6.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,6,-0.0050590806640684605,0.16059811363634433,2.59375
|
||||
model.visual.blocks.6.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,6,-8.25678072504353e-06,0.015881123862622196,0.59375
|
||||
model.visual.blocks.6.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,6,-0.00010237316018901765,0.20565713205216193,4.15625
|
||||
model.visual.blocks.6.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,6,0.9554309844970703,0.9669409615623171,1.75
|
||||
model.visual.blocks.6.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,6,0.003839905373752117,0.23620944229480795,3.90625
|
||||
model.visual.blocks.6.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,6,0.9070353507995605,0.9159196848385002,1.8203125
|
||||
model.visual.blocks.7.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,7,-0.0023070083698257804,0.20943125114081082,2.859375
|
||||
model.visual.blocks.7.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,7,8.261984775215225e-07,0.014035490681116675,0.119140625
|
||||
model.visual.blocks.7.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,7,0.009815204190090299,0.49869548059937374,7.65625
|
||||
model.visual.blocks.7.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,7,2.9990876174723624e-06,0.01649365033942753,0.251953125
|
||||
model.visual.blocks.7.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,7,-1.6987448390573263,1.8391509676765594,3.9375
|
||||
model.visual.blocks.7.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,7,0.00017219371261305294,0.016408897942078948,0.28125
|
||||
model.visual.blocks.7.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,7,-0.008466638159006834,0.17516442409220903,1.9375
|
||||
model.visual.blocks.7.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,7,2.340813741535941e-06,0.014372225294541803,0.255859375
|
||||
model.visual.blocks.7.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,7,0.003123911563307047,0.1572316380772548,3.375
|
||||
model.visual.blocks.7.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,7,0.9082684516906738,0.9232805356334663,1.7265625
|
||||
model.visual.blocks.7.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,7,0.009727229364216328,0.24249418742314302,3.78125
|
||||
model.visual.blocks.7.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,7,1.0042033195495605,1.019765058492056,1.5
|
||||
model.visual.blocks.8.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,8,0.0016495957970619202,0.24097282912908569,3.96875
|
||||
model.visual.blocks.8.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,8,-2.7905498484193103e-06,0.01345574124824084,0.283203125
|
||||
model.visual.blocks.8.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,8,0.008241087314672768,0.47954093823312155,7.1875
|
||||
model.visual.blocks.8.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,8,-1.5548966505463113e-05,0.01653429785702248,0.2275390625
|
||||
model.visual.blocks.8.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,8,-1.6126111019402742,1.7683807290973117,3.9375
|
||||
model.visual.blocks.8.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,8,0.0001252524039837416,0.01624719901557155,0.28515625
|
||||
model.visual.blocks.8.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,8,-0.005296873860061169,0.2216358320198793,2.46875
|
||||
model.visual.blocks.8.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,8,1.4216143319060667e-05,0.013571702866860212,0.2470703125
|
||||
model.visual.blocks.8.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,8,0.0055080801248550415,0.20896838474609777,3.859375
|
||||
model.visual.blocks.8.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,8,0.9037561416625977,0.9260690715493488,1.71875
|
||||
model.visual.blocks.8.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,8,0.003971373429521918,0.310706905909723,3.921875
|
||||
model.visual.blocks.8.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,8,1.0045695304870605,1.0178206173386362,1.5546875
|
||||
model.visual.blocks.9.attn.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.attn.proj.bias,9,0.0010921573266386986,0.22680602547304463,3.265625
|
||||
model.visual.blocks.9.attn.proj.weight,"(1024, 1024)",2,BF16,1048576,2097152,model.visual.layers.*.attn.proj.weight,9,1.4016385481707694e-06,0.01288622156493415,0.1123046875
|
||||
model.visual.blocks.9.attn.qkv.bias,"(3072,)",1,BF16,3072,6144,model.visual.layers.*.attn.qkv.bias,9,0.001626015097523729,0.4891119861662029,7.25
|
||||
model.visual.blocks.9.attn.qkv.weight,"(3072, 1024)",2,BF16,3145728,6291456,model.visual.layers.*.attn.qkv.weight,9,-4.602129067709608e-06,0.016435399387831742,0.2119140625
|
||||
model.visual.blocks.9.mlp.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.layers.*.mlp.linear_fc1.bias,9,-1.6456444784998894,1.7699452545615144,8.125
|
||||
model.visual.blocks.9.mlp.linear_fc1.weight,"(4096, 1024)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc1.weight,9,5.019114076590045e-05,0.015971322934801892,0.2734375
|
||||
model.visual.blocks.9.mlp.linear_fc2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.mlp.linear_fc2.bias,9,-0.00503852404654026,0.2206179125090425,2.28125
|
||||
model.visual.blocks.9.mlp.linear_fc2.weight,"(1024, 4096)",2,BF16,4194304,8388608,model.visual.layers.*.mlp.linear_fc2.weight,9,2.0485118040584893e-05,0.013085999668089059,0.29296875
|
||||
model.visual.blocks.9.norm1.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.bias,9,0.003136083483695984,0.18839499922526784,2.953125
|
||||
model.visual.blocks.9.norm1.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm1.weight,9,0.9992551803588867,1.0213929426253632,1.8203125
|
||||
model.visual.blocks.9.norm2.bias,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.bias,9,0.0005389824509620667,0.30238562581121287,2.90625
|
||||
model.visual.blocks.9.norm2.weight,"(1024,)",1,BF16,1024,2048,model.visual.layers.*.norm2.weight,9,1.1041545867919922,1.1208131568613,1.734375
|
||||
model.visual.merger.linear_fc1.bias,"(4096,)",1,BF16,4096,8192,model.visual.merger.linear_fc1.bias,-1,-0.0005523472136701457,0.01626457054117911,0.115234375
|
||||
model.visual.merger.linear_fc1.weight,"(4096, 4096)",2,BF16,16777216,33554432,model.visual.merger.linear_fc1.weight,-1,-0.00015279042523767844,0.011285763080892746,0.0947265625
|
||||
model.visual.merger.linear_fc2.bias,"(2048,)",1,BF16,2048,4096,model.visual.merger.linear_fc2.bias,-1,-0.00020432728342711926,0.022563638409818742,0.4921875
|
||||
model.visual.merger.linear_fc2.weight,"(2048, 4096)",2,BF16,8388608,16777216,model.visual.merger.linear_fc2.weight,-1,5.585822149016192e-07,0.009865405435281218,0.19921875
|
||||
model.visual.merger.norm.bias,"(1024,)",1,BF16,1024,2048,model.visual.merger.norm.bias,-1,0.049520475789904594,0.3246462954969724,3.453125
|
||||
model.visual.merger.norm.weight,"(1024,)",1,BF16,1024,2048,model.visual.merger.norm.weight,-1,1.4015541076660156,1.41604852572399,2.9375
|
||||
model.visual.patch_embed.proj.bias,"(1024,)",1,BF16,1024,2048,model.visual.patch_embed.proj.bias,-1,-0.009102825075387955,0.3202035696576827,3.1875
|
||||
model.visual.patch_embed.proj.weight,"(1024, 3, 2, 16, 16)",5,BF16,1572864,3145728,model.visual.patch_embed.proj.weight,-1,-0.0002164951033312712,0.007756875182799582,0.06884765625
|
||||
model.visual.pos_embed.weight,"(2304, 1024)",2,BF16,2359296,4718592,model.visual.pos_embed.weight,-1,0.003249331564613057,0.1360526766744249,7.21875
|
||||
mtp.fc.weight,"(2048, 4096)",2,BF16,8388608,16777216,mtp.fc.weight,-1,5.3096681990941023e-05,0.006383883815548282,0.3125
|
||||
mtp.layers.0.input_layernorm.weight,"(2048,)",1,BF16,2048,4096,mtp.layers.*.input_layernorm.weight,0,0.3319206237792969,0.3510988118095836,1.1328125
|
||||
mtp.layers.0.mlp.down_proj.weight,"(2048, 6144)",2,BF16,12582912,25165824,mtp.layers.*.mlp.down_proj.weight,0,-5.053560631807126e-06,0.010496220813270141,0.5
|
||||
mtp.layers.0.mlp.gate_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,mtp.layers.*.mlp.gate_proj.weight,0,-2.054409559734451e-05,0.013481584415555293,0.23046875
|
||||
mtp.layers.0.mlp.up_proj.weight,"(6144, 2048)",2,BF16,12582912,25165824,mtp.layers.*.mlp.up_proj.weight,0,9.151307207333424e-07,0.01273030821065844,0.337890625
|
||||
mtp.layers.0.post_attention_layernorm.weight,"(2048,)",1,BF16,2048,4096,mtp.layers.*.post_attention_layernorm.weight,0,0.41015276312828064,0.42291990555998127,0.78125
|
||||
mtp.layers.0.self_attn.k_norm.weight,"(256,)",1,BF16,256,512,mtp.layers.*.self_attn.k_norm.weight,0,0.7713251113891602,0.8445900427746588,2.046875
|
||||
mtp.layers.0.self_attn.k_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,mtp.layers.*.self_attn.k_proj.weight,0,1.4686407705633009e-05,0.011793984318089815,0.1669921875
|
||||
mtp.layers.0.self_attn.o_proj.weight,"(2048, 2048)",2,BF16,4194304,8388608,mtp.layers.*.self_attn.o_proj.weight,0,2.4605430018389413e-06,0.01508009745888187,0.5234375
|
||||
mtp.layers.0.self_attn.q_norm.weight,"(256,)",1,BF16,256,512,mtp.layers.*.self_attn.q_norm.weight,0,0.8072891235351562,0.8754819538932436,2.4375
|
||||
mtp.layers.0.self_attn.q_proj.weight,"(4096, 2048)",2,BF16,8388608,16777216,mtp.layers.*.self_attn.q_proj.weight,0,1.1836838601198777e-05,0.015479120981825292,0.3203125
|
||||
mtp.layers.0.self_attn.v_proj.weight,"(512, 2048)",2,BF16,1048576,2097152,mtp.layers.*.self_attn.v_proj.weight,0,3.6436836885478385e-06,0.026195977141207714,0.30859375
|
||||
mtp.norm.weight,"(2048,)",1,BF16,2048,4096,mtp.norm.weight,-1,2.8050971627235413,2.8500763286553985,6.3125
|
||||
mtp.pre_fc_norm_embedding.weight,"(2048,)",1,BF16,2048,4096,mtp.pre_fc_norm_embedding.weight,-1,-0.43043196201324463,0.4398642357123278,0.640625
|
||||
mtp.pre_fc_norm_hidden.weight,"(2048,)",1,BF16,2048,4096,mtp.pre_fc_norm_hidden.weight,-1,-0.28146980330348015,0.2999654479966,0.69921875
|
||||
|
@@ -0,0 +1,159 @@
|
||||
mode,stratum,layer_distance,pairs,median_cosine,q25,q75
|
||||
tensor,1D,1,183,0.9406331265725202,0.7235207005037987,0.9931629316616981
|
||||
tensor,1D,2,145,0.9122918066446817,0.5723343348709384,0.9897324183654799
|
||||
tensor,1D,3,121,0.901223383994159,0.4828586539344226,0.9863693486780729
|
||||
tensor,1D,4,111,0.9003575650487153,0.5367023333809763,0.9804638617660654
|
||||
tensor,1D,5,72,0.8953776736324203,0.3538804544441773,0.9680854066934634
|
||||
tensor,1D,6,57,0.8879212864097908,0.36305448910479954,0.9558740974722054
|
||||
tensor,1D,7,47,0.8776805321645917,0.4535526800238403,0.9433999957988433
|
||||
tensor,1D,8,52,0.8825518190229764,0.4824499801571124,0.9176637782905971
|
||||
tensor,1D,9,31,0.8958045595384818,0.4305895667884128,0.9193278445717825
|
||||
tensor,1D,10,17,0.8827035753940228,0.2092648777131711,0.9150312769507561
|
||||
tensor,1D,11,16,0.8221685308473708,0.4800308710016442,0.9274441239891063
|
||||
tensor,1D,12,24,0.8415123611721862,0.3169780147322381,0.9009036671689896
|
||||
tensor,1D,13,14,0.8516287869780905,0.1243725256398088,0.9508162094932566
|
||||
tensor,1D,14,14,0.6469216553724613,0.12853366420837778,0.9574883079605649
|
||||
tensor,1D,15,12,0.5833474451569504,0.046843864053728236,0.9769132059565613
|
||||
tensor,1D,16,16,0.7953515120653798,0.16121722307705078,0.8372406530303363
|
||||
tensor,1D,17,9,0.19251868016561066,0.03853257212466431,0.47599207111348585
|
||||
tensor,1D,18,9,0.04608358701275898,-0.006661812445004314,0.7982991506705737
|
||||
tensor,1D,19,10,0.09441293378335211,0.025237536550509452,0.49922331710553325
|
||||
tensor,1D,20,14,0.5022983141733379,0.05512087268438936,0.8423207884676235
|
||||
tensor,1D,21,5,0.8934819071719368,0.24010298911027775,0.9021497732991838
|
||||
tensor,1D,22,3,0.003892505911476348,0.0008970125263505391,0.4356982824466657
|
||||
tensor,1D,23,1,0.024749331545084748,0.024749331545084748,0.024749331545084748
|
||||
tensor,2D,1,19,0.1976207505481654,0.0006525511235171834,0.2716061024781503
|
||||
tensor,2D,2,31,0.1319696165434511,0.0004437785635444379,0.243080676179633
|
||||
tensor,2D,3,31,0.11611374254300347,0.0002621828568059063,0.20816418030431522
|
||||
tensor,2D,4,69,0.0019111009657841638,8.141249860595188e-05,0.022487085707574864
|
||||
tensor,2D,5,45,0.000506090620166805,-8.291910146519681e-05,0.00599106215370278
|
||||
tensor,2D,6,44,0.00041039555013748237,-0.0006400509766726694,0.006286639984231686
|
||||
tensor,2D,7,50,0.00025574624849179794,-0.00033235836165827667,0.004524560602680037
|
||||
tensor,2D,8,75,0.0002696699762064535,-0.0003334243181754402,0.0043499444214343055
|
||||
tensor,2D,9,57,0.00028671564523554733,-0.00015013298858171262,0.0022893779585961825
|
||||
tensor,2D,10,54,7.181569169932083e-05,-0.00029630089314203645,0.001144667363346839
|
||||
tensor,2D,11,48,0.00021728323517404915,-0.00018317637500131213,0.0010438034931358998
|
||||
tensor,2D,12,74,5.5818951912901686e-05,-0.0003723509751072158,0.0006649377509370046
|
||||
tensor,2D,13,56,0.00023809228660712564,-0.00020331365964920253,0.000892605117750516
|
||||
tensor,2D,14,51,0.00023447788066002293,-9.917778299137728e-05,0.000713026509065598
|
||||
tensor,2D,15,37,9.091975960188207e-05,-0.00021742221041551871,0.00033608530344073333
|
||||
tensor,2D,16,55,0.00012381177776811685,-0.00020495997834418578,0.0006928152055894044
|
||||
tensor,2D,17,40,0.00020240660120543027,-0.00026382525874345676,0.000542300816362679
|
||||
tensor,2D,18,25,5.7811223671745594e-05,-0.00022620903865658743,0.0005118670799619454
|
||||
tensor,2D,19,23,0.00039112184326041497,-0.00010049174294494889,0.0010346739140039576
|
||||
tensor,2D,20,24,0.000265491219726623,-0.0001958276840250541,0.0007077565884118048
|
||||
tensor,2D,21,16,6.914175780974358e-06,-0.00028062530736593067,0.0008637952736474721
|
||||
tensor,2D,22,13,-0.00025490323864760033,-0.0004976782457785606,0.0001580595880243879
|
||||
tensor,2D,23,6,-0.00034347924301872656,-0.0007262675393089821,-0.00011141867617238705
|
||||
tensor,3D,1,1,-0.00680822959446004,-0.00680822959446004,-0.00680822959446004
|
||||
tensor,3D,2,2,-0.012075220938999093,-0.012980137400828292,-0.011170304477169895
|
||||
tensor,3D,3,1,-0.0010720687851472769,-0.0010720687851472769,-0.0010720687851472769
|
||||
tensor,3D,4,3,-0.0054659741398542325,-0.009296679734407527,-0.004429821782672542
|
||||
tensor,3D,5,2,0.010378648653957328,0.00870465890772889,0.012052638400185764
|
||||
tensor,3D,6,4,-0.001207609704539777,-0.003950241011882311,0.004702402921592683
|
||||
tensor,3D,7,3,0.0007654251132529078,-0.011486421146842795,0.0038046823169703373
|
||||
tensor,3D,8,6,-0.0008439355677116029,-0.0037378805904456115,0.0031215275083032084
|
||||
tensor,3D,9,4,-0.0026239820437580167,-0.0036245316783073207,0.0013061750169709205
|
||||
tensor,3D,10,4,-0.001271951809912471,-0.0060886723799495855,0.0034686674864937684
|
||||
tensor,3D,11,5,-0.004550474178988823,-0.004781306789044417,0.0050607377208408836
|
||||
tensor,3D,12,6,-0.0026088290601245533,-0.007923664560857482,0.006914391737556268
|
||||
tensor,3D,13,5,0.0013332822970221264,-0.0026091070128858385,0.0020424739613098585
|
||||
tensor,3D,14,4,0.0008607282396081882,-0.002961687969598026,0.008582534185585031
|
||||
tensor,3D,15,4,0.000536649649268034,-0.006516127851339201,0.007806962616522985
|
||||
tensor,3D,16,6,0.005405527088084872,-0.008270095286873539,0.007503111340192185
|
||||
tensor,3D,17,3,0.003652404097247392,-0.003705339105968001,0.00510019946539501
|
||||
tensor,3D,18,3,-0.011706507766527128,-0.013407487905083771,-0.009676616224298069
|
||||
tensor,3D,19,2,-0.001913393756159387,-0.0052948070490088945,0.0014680195366901208
|
||||
tensor,3D,20,3,0.011951381051850295,0.008569075846974247,0.013104140256261814
|
||||
tensor,3D,21,2,-0.008922546499844752,-0.018991073095849714,0.0011459800961602113
|
||||
tensor,3D,22,1,3.711960334307905e-05,3.711960334307905e-05,3.711960334307905e-05
|
||||
block,2d-128,0,3390,0.007793596035932384,0.0008041244980700559,0.022869426809521965
|
||||
block,2d-128,1,69,0.004283737242705982,-0.0013359793502242799,0.009291223245626997
|
||||
block,2d-128,2,101,0.002140727979805649,-0.004216128113395974,0.006727795921497354
|
||||
block,2d-128,3,69,0.0012800926790249956,-0.003000592803802066,0.007263477599848168
|
||||
block,2d-128,4,263,0.001583838161828825,-0.004179970579220923,0.007362195589800033
|
||||
block,2d-128,5,78,0.0029623031901291697,-0.002068802338494363,0.007715193824410069
|
||||
block,2d-128,6,174,0.002047977773085714,-0.0025244999191454693,0.007750911362254816
|
||||
block,2d-128,7,100,0.0020361314382989775,-0.003062414263868483,0.0076813599806539925
|
||||
block,2d-128,8,321,0.0034170788377522656,-0.001459503917936015,0.008568902349863619
|
||||
block,2d-128,9,108,0.004622566491664202,-0.00030913283627803743,0.009786372572060848
|
||||
block,2d-128,10,203,0.004454188722342168,-0.0006831975509908592,0.008540574305915417
|
||||
block,2d-128,11,134,0.0039352085175531504,-0.0017064287867521906,0.008925241561231641
|
||||
block,2d-128,12,532,0.003251781487081145,-0.0012796507741402387,0.0088326642641113
|
||||
block,2d-128,13,130,0.003723166442033233,-0.0008399249287298995,0.008038314156931613
|
||||
block,2d-128,14,137,0.002342468861900797,-0.0027744015612124846,0.007983134522471498
|
||||
block,2d-128,15,122,0.0038527925678218857,-0.00014374713757119012,0.008013752754379396
|
||||
block,2d-128,16,390,0.0034266076308263316,-0.0022279939354585164,0.009309021336800772
|
||||
block,2d-128,17,150,0.0036599721357675244,-0.003146422112013574,0.009006579887796635
|
||||
block,2d-128,18,139,0.0035243681663050756,-0.0016097963567532094,0.008863143808684481
|
||||
block,2d-128,19,142,0.003298317396498646,-0.00216909646960855,0.009489362110928035
|
||||
block,2d-128,20,198,0.0023368275261753642,-0.0025869835152945437,0.0067517550852595545
|
||||
block,2d-128,21,101,0.0011502151156738772,-0.005492110246231151,0.007882221778894924
|
||||
block,2d-128,22,40,0.003846590896373531,-0.0020196368366427325,0.009238380039353778
|
||||
block,2d-128,23,24,0.0015900279251504135,-0.0034150014260014377,0.0065953378604191246
|
||||
block,2d-64,0,3407,0.020411292155791013,0.008566729110530762,0.03569139114296313
|
||||
block,2d-64,1,88,0.008415934314043649,-0.003203824302970467,0.01889094345097298
|
||||
block,2d-64,2,164,0.009675813300819843,-0.004711837470762339,0.020945070937976786
|
||||
block,2d-64,3,73,0.009454457835350863,-0.0006829409261957054,0.02170490311449347
|
||||
block,2d-64,4,338,0.006813719398160595,-0.004772048080579105,0.018045621570531245
|
||||
block,2d-64,5,75,0.011231706607558869,-0.0006943540283602765,0.022172646210020717
|
||||
block,2d-64,6,118,0.008006055417925154,-0.0003843199497181453,0.018406001412445316
|
||||
block,2d-64,7,85,0.013831026305481255,0.004535720514774543,0.023925556311973416
|
||||
block,2d-64,8,264,0.013879360999226419,0.003533830484789825,0.024982882867886597
|
||||
block,2d-64,9,93,0.01468007218218553,0.006762887964066403,0.024410504891315212
|
||||
block,2d-64,10,176,0.012299105745675366,0.0019620226161680167,0.02498214610621178
|
||||
block,2d-64,11,114,0.013304371636863828,0.003517815772814807,0.02161109096095775
|
||||
block,2d-64,12,447,0.010440781560487453,0.0002481645111614831,0.020212544190424727
|
||||
block,2d-64,13,126,0.013956002337464685,0.0028689345210515405,0.022436635938580824
|
||||
block,2d-64,14,138,0.012482215702635066,-0.0017681288556580551,0.022861117559496998
|
||||
block,2d-64,15,108,0.011999026405420592,0.0026512574736944133,0.022669337612718547
|
||||
block,2d-64,16,434,0.01204641193967287,-4.675386640052482e-05,0.021641739064571487
|
||||
block,2d-64,17,145,0.014295415151533953,0.002994286738522399,0.026167283872632573
|
||||
block,2d-64,18,142,0.012052517201990746,0.003250968375752944,0.024226648842664292
|
||||
block,2d-64,19,118,0.016783769527618606,0.006470002050503073,0.027477607514195174
|
||||
block,2d-64,20,261,0.010187027433596979,-0.0016399905463545146,0.021019963982582265
|
||||
block,2d-64,21,107,0.01346166883130134,-0.0006729272182597161,0.02168460705584324
|
||||
block,2d-64,22,58,0.018115860023517876,0.006176186637218011,0.02813714013589269
|
||||
block,2d-64,23,36,0.011484288868861606,0.0009840500965715385,0.025732391656693
|
||||
block,3d-axis0-slice,0,21,0.9979974213550225,0.9919132082454284,0.9991074607826514
|
||||
block,3d-axis0-slice,1,30,0.9964777342885875,0.9930253717757319,0.9986384781445289
|
||||
block,3d-axis0-slice,2,19,0.9944159697777617,0.9918918379626837,0.9984051014947126
|
||||
block,3d-axis0-slice,3,25,0.9952442911267027,0.9901901449619488,0.9969886550034407
|
||||
block,3d-axis0-slice,4,25,0.9974318963005511,0.9943409923291456,0.9986038889492098
|
||||
block,3d-axis0-slice,5,18,0.998267157357618,0.9952774441912474,0.9991537904234183
|
||||
block,3d-axis0-slice,6,13,0.9961877275009366,0.9923370760320499,0.9974347454424667
|
||||
block,3d-axis0-slice,7,16,0.9949061047476491,0.9910801621582696,0.9962644181928078
|
||||
block,3d-axis0-slice,8,20,0.9964421107245474,0.9897394461959558,0.9981449317504201
|
||||
block,3d-axis0-slice,9,18,0.997667747065203,0.9953788750600205,0.9993726748769715
|
||||
block,3d-axis0-slice,10,8,0.9962070328625536,0.9881559034236903,0.9987132222644496
|
||||
block,3d-axis0-slice,11,8,0.9942417984106242,0.990425139830641,0.997612429002756
|
||||
block,3d-axis0-slice,12,21,0.9972829832575825,0.9953609819889534,0.9989504871466931
|
||||
block,3d-axis0-slice,13,8,0.9990981742596038,0.9972523909618378,0.999415167966053
|
||||
block,3d-axis0-slice,14,3,0.9992770560381737,0.9946946839786991,0.9993095988827703
|
||||
block,3d-axis0-slice,15,5,0.9976076675869366,0.997220089080847,0.9982798840475241
|
||||
block,3d-axis0-slice,16,5,0.9938412885348621,0.9904227326475673,0.9975516447539853
|
||||
block,3d-axis0-slice,17,4,0.9976380431427594,0.99681231970145,0.9980051631791542
|
||||
block,3d-axis0-slice,18,3,0.990204033484286,0.9897726565262364,0.9907733664461621
|
||||
block,3d-axis0-slice,20,3,0.9978354987542325,0.9969266728454125,0.9984509007507711
|
||||
block,3d-axis0-slice,21,1,0.9903946099595192,0.9903946099595192,0.9903946099595192
|
||||
block,3d-volume,0,11,0.8808719619367277,0.8022160073788235,0.9290339633281453
|
||||
block,3d-volume,1,30,0.860270360468439,0.7843981969261361,0.9080938775304134
|
||||
block,3d-volume,2,32,0.8560786678157898,0.7873096209196705,0.9172702050933522
|
||||
block,3d-volume,3,20,0.8635443286938794,0.7739478595366118,0.9352764032739678
|
||||
block,3d-volume,4,29,0.839065051747565,0.7756163554455586,0.892317257603093
|
||||
block,3d-volume,5,20,0.802039122632485,0.7698342438427777,0.8892367804699988
|
||||
block,3d-volume,6,19,0.8649610706130815,0.8068751614830983,0.9147549633399301
|
||||
block,3d-volume,7,13,0.8338788272523221,0.8111808153950231,0.8961788833920625
|
||||
block,3d-volume,8,25,0.8210691749524753,0.7869135461084978,0.8984399780703005
|
||||
block,3d-volume,9,7,0.7921202796541179,0.7829800826876393,0.8383111208949781
|
||||
block,3d-volume,10,14,0.8712767579792928,0.808286835207875,0.9311130535641426
|
||||
block,3d-volume,11,15,0.8567301881799948,0.8420882491714214,0.8723359157908674
|
||||
block,3d-volume,12,13,0.878820052327295,0.8005055367654049,0.9003593656208086
|
||||
block,3d-volume,13,5,0.8413887031483342,0.8269826043982158,0.8663564744427157
|
||||
block,3d-volume,14,5,0.8482732480061685,0.8037558138103084,0.8543993886099361
|
||||
block,3d-volume,15,4,0.9044995313662272,0.8779254085355535,0.9110599729107347
|
||||
block,3d-volume,16,4,0.843600741507026,0.7971940369401024,0.8823409979846442
|
||||
block,3d-volume,17,5,0.7775241450511589,0.7623044324936186,0.8205122958709042
|
||||
block,3d-volume,18,1,0.7536817290838455,0.7536817290838455,0.7536817290838455
|
||||
block,3d-volume,19,1,0.8474271739113811,0.8474271739113811,0.8474271739113811
|
||||
block,3d-volume,21,1,0.8644054697085172,0.8644054697085172,0.8644054697085172
|
||||
|
@@ -0,0 +1,376 @@
|
||||
mode,ndim,method,samples,bin_left,bin_right,count,density
|
||||
tensor,1,shuffled,48,-1.0,-0.92,0,0.0
|
||||
tensor,1,shuffled,48,-0.92,-0.84,0,0.0
|
||||
tensor,1,shuffled,48,-0.84,-0.76,0,0.0
|
||||
tensor,1,shuffled,48,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,1,shuffled,48,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,1,shuffled,48,-0.6,-0.52,0,0.0
|
||||
tensor,1,shuffled,48,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,1,shuffled,48,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,1,shuffled,48,-0.36,-0.28,0,0.0
|
||||
tensor,1,shuffled,48,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,1,shuffled,48,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,1,shuffled,48,-0.12,-0.040000000000000036,1,0.2604166666666668
|
||||
tensor,1,shuffled,48,-0.040000000000000036,0.040000000000000036,12,3.1249999999999973
|
||||
tensor,1,shuffled,48,0.040000000000000036,0.1200000000000001,2,0.5208333333333329
|
||||
tensor,1,shuffled,48,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,1,shuffled,48,0.19999999999999996,0.28,0,0.0
|
||||
tensor,1,shuffled,48,0.28,0.3600000000000001,1,0.26041666666666646
|
||||
tensor,1,shuffled,48,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,1,shuffled,48,0.43999999999999995,0.52,2,0.5208333333333329
|
||||
tensor,1,shuffled,48,0.52,0.6000000000000001,2,0.5208333333333329
|
||||
tensor,1,shuffled,48,0.6000000000000001,0.6799999999999999,2,0.5208333333333344
|
||||
tensor,1,shuffled,48,0.6799999999999999,0.76,0,0.0
|
||||
tensor,1,shuffled,48,0.76,0.8400000000000001,0,0.0
|
||||
tensor,1,shuffled,48,0.8400000000000001,0.9199999999999999,6,1.5625000000000029
|
||||
tensor,1,shuffled,48,0.9199999999999999,1.0,20,5.208333333333329
|
||||
tensor,1,gaussian_norm_matched,48,-1.0,-0.92,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.92,-0.84,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.84,-0.76,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.6,-0.52,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.36,-0.28,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,-0.12,-0.040000000000000036,8,2.0833333333333344
|
||||
tensor,1,gaussian_norm_matched,48,-0.040000000000000036,0.040000000000000036,31,8.072916666666659
|
||||
tensor,1,gaussian_norm_matched,48,0.040000000000000036,0.1200000000000001,5,1.3020833333333321
|
||||
tensor,1,gaussian_norm_matched,48,0.1200000000000001,0.19999999999999996,1,0.2604166666666672
|
||||
tensor,1,gaussian_norm_matched,48,0.19999999999999996,0.28,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.28,0.3600000000000001,2,0.5208333333333329
|
||||
tensor,1,gaussian_norm_matched,48,0.3600000000000001,0.43999999999999995,1,0.2604166666666672
|
||||
tensor,1,gaussian_norm_matched,48,0.43999999999999995,0.52,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.52,0.6000000000000001,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.6799999999999999,0.76,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.76,0.8400000000000001,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,1,gaussian_norm_matched,48,0.9199999999999999,1.0,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-1.0,-0.92,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.92,-0.84,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.84,-0.76,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.6,-0.52,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.36,-0.28,1,0.2777777777777779
|
||||
tensor,1,random_actual_same_role,45,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,1,random_actual_same_role,45,-0.12,-0.040000000000000036,2,0.5555555555555558
|
||||
tensor,1,random_actual_same_role,45,-0.040000000000000036,0.040000000000000036,5,1.3888888888888877
|
||||
tensor,1,random_actual_same_role,45,0.040000000000000036,0.1200000000000001,3,0.8333333333333326
|
||||
tensor,1,random_actual_same_role,45,0.1200000000000001,0.19999999999999996,1,0.2777777777777783
|
||||
tensor,1,random_actual_same_role,45,0.19999999999999996,0.28,1,0.2777777777777775
|
||||
tensor,1,random_actual_same_role,45,0.28,0.3600000000000001,1,0.2777777777777775
|
||||
tensor,1,random_actual_same_role,45,0.3600000000000001,0.43999999999999995,2,0.5555555555555566
|
||||
tensor,1,random_actual_same_role,45,0.43999999999999995,0.52,1,0.2777777777777775
|
||||
tensor,1,random_actual_same_role,45,0.52,0.6000000000000001,2,0.555555555555555
|
||||
tensor,1,random_actual_same_role,45,0.6000000000000001,0.6799999999999999,1,0.2777777777777783
|
||||
tensor,1,random_actual_same_role,45,0.6799999999999999,0.76,2,0.555555555555555
|
||||
tensor,1,random_actual_same_role,45,0.76,0.8400000000000001,2,0.555555555555555
|
||||
tensor,1,random_actual_same_role,45,0.8400000000000001,0.9199999999999999,8,2.2222222222222263
|
||||
tensor,1,random_actual_same_role,45,0.9199999999999999,1.0,13,3.611111111111108
|
||||
tensor,2,shuffled,48,-1.0,-0.92,0,0.0
|
||||
tensor,2,shuffled,48,-0.92,-0.84,0,0.0
|
||||
tensor,2,shuffled,48,-0.84,-0.76,0,0.0
|
||||
tensor,2,shuffled,48,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,2,shuffled,48,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,2,shuffled,48,-0.6,-0.52,0,0.0
|
||||
tensor,2,shuffled,48,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,2,shuffled,48,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,2,shuffled,48,-0.36,-0.28,0,0.0
|
||||
tensor,2,shuffled,48,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,2,shuffled,48,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,2,shuffled,48,-0.12,-0.040000000000000036,1,0.2604166666666668
|
||||
tensor,2,shuffled,48,-0.040000000000000036,0.040000000000000036,42,10.937499999999991
|
||||
tensor,2,shuffled,48,0.040000000000000036,0.1200000000000001,5,1.3020833333333321
|
||||
tensor,2,shuffled,48,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,2,shuffled,48,0.19999999999999996,0.28,0,0.0
|
||||
tensor,2,shuffled,48,0.28,0.3600000000000001,0,0.0
|
||||
tensor,2,shuffled,48,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,2,shuffled,48,0.43999999999999995,0.52,0,0.0
|
||||
tensor,2,shuffled,48,0.52,0.6000000000000001,0,0.0
|
||||
tensor,2,shuffled,48,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,2,shuffled,48,0.6799999999999999,0.76,0,0.0
|
||||
tensor,2,shuffled,48,0.76,0.8400000000000001,0,0.0
|
||||
tensor,2,shuffled,48,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,2,shuffled,48,0.9199999999999999,1.0,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-1.0,-0.92,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.92,-0.84,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.84,-0.76,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.6,-0.52,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.36,-0.28,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,-0.12,-0.040000000000000036,1,0.2604166666666668
|
||||
tensor,2,gaussian_norm_matched,48,-0.040000000000000036,0.040000000000000036,47,12.239583333333323
|
||||
tensor,2,gaussian_norm_matched,48,0.040000000000000036,0.1200000000000001,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.19999999999999996,0.28,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.28,0.3600000000000001,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.43999999999999995,0.52,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.52,0.6000000000000001,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.6799999999999999,0.76,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.76,0.8400000000000001,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,2,gaussian_norm_matched,48,0.9199999999999999,1.0,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-1.0,-0.92,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.92,-0.84,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.84,-0.76,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.6,-0.52,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.36,-0.28,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,2,random_actual_same_role,48,-0.12,-0.040000000000000036,2,0.5208333333333336
|
||||
tensor,2,random_actual_same_role,48,-0.040000000000000036,0.040000000000000036,39,10.156249999999991
|
||||
tensor,2,random_actual_same_role,48,0.040000000000000036,0.1200000000000001,5,1.3020833333333321
|
||||
tensor,2,random_actual_same_role,48,0.1200000000000001,0.19999999999999996,1,0.2604166666666672
|
||||
tensor,2,random_actual_same_role,48,0.19999999999999996,0.28,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.28,0.3600000000000001,1,0.26041666666666646
|
||||
tensor,2,random_actual_same_role,48,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.43999999999999995,0.52,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.52,0.6000000000000001,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.6799999999999999,0.76,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.76,0.8400000000000001,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,2,random_actual_same_role,48,0.9199999999999999,1.0,0,0.0
|
||||
tensor,3,shuffled,4,-1.0,-0.92,0,0.0
|
||||
tensor,3,shuffled,4,-0.92,-0.84,0,0.0
|
||||
tensor,3,shuffled,4,-0.84,-0.76,0,0.0
|
||||
tensor,3,shuffled,4,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,3,shuffled,4,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,3,shuffled,4,-0.6,-0.52,0,0.0
|
||||
tensor,3,shuffled,4,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,3,shuffled,4,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,3,shuffled,4,-0.36,-0.28,0,0.0
|
||||
tensor,3,shuffled,4,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,3,shuffled,4,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,3,shuffled,4,-0.12,-0.040000000000000036,0,0.0
|
||||
tensor,3,shuffled,4,-0.040000000000000036,0.040000000000000036,3,9.374999999999991
|
||||
tensor,3,shuffled,4,0.040000000000000036,0.1200000000000001,1,3.1249999999999973
|
||||
tensor,3,shuffled,4,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,3,shuffled,4,0.19999999999999996,0.28,0,0.0
|
||||
tensor,3,shuffled,4,0.28,0.3600000000000001,0,0.0
|
||||
tensor,3,shuffled,4,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,3,shuffled,4,0.43999999999999995,0.52,0,0.0
|
||||
tensor,3,shuffled,4,0.52,0.6000000000000001,0,0.0
|
||||
tensor,3,shuffled,4,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,3,shuffled,4,0.6799999999999999,0.76,0,0.0
|
||||
tensor,3,shuffled,4,0.76,0.8400000000000001,0,0.0
|
||||
tensor,3,shuffled,4,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,3,shuffled,4,0.9199999999999999,1.0,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-1.0,-0.92,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.92,-0.84,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.84,-0.76,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.6,-0.52,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.36,-0.28,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.12,-0.040000000000000036,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,-0.040000000000000036,0.040000000000000036,4,12.49999999999999
|
||||
tensor,3,gaussian_norm_matched,4,0.040000000000000036,0.1200000000000001,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.19999999999999996,0.28,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.28,0.3600000000000001,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.43999999999999995,0.52,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.52,0.6000000000000001,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.6799999999999999,0.76,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.76,0.8400000000000001,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,3,gaussian_norm_matched,4,0.9199999999999999,1.0,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-1.0,-0.92,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.92,-0.84,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.84,-0.76,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.76,-0.6799999999999999,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.6799999999999999,-0.6,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.6,-0.52,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.52,-0.43999999999999995,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.43999999999999995,-0.36,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.36,-0.28,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.28,-0.19999999999999996,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.19999999999999996,-0.12,0,0.0
|
||||
tensor,3,random_actual_same_role,3,-0.12,-0.040000000000000036,1,4.166666666666669
|
||||
tensor,3,random_actual_same_role,3,-0.040000000000000036,0.040000000000000036,1,4.166666666666663
|
||||
tensor,3,random_actual_same_role,3,0.040000000000000036,0.1200000000000001,1,4.166666666666663
|
||||
tensor,3,random_actual_same_role,3,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.19999999999999996,0.28,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.28,0.3600000000000001,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.43999999999999995,0.52,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.52,0.6000000000000001,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.6799999999999999,0.76,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.76,0.8400000000000001,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
tensor,3,random_actual_same_role,3,0.9199999999999999,1.0,0,0.0
|
||||
block,2,shuffled,96,-1.0,-0.92,0,0.0
|
||||
block,2,shuffled,96,-0.92,-0.84,0,0.0
|
||||
block,2,shuffled,96,-0.84,-0.76,0,0.0
|
||||
block,2,shuffled,96,-0.76,-0.6799999999999999,0,0.0
|
||||
block,2,shuffled,96,-0.6799999999999999,-0.6,0,0.0
|
||||
block,2,shuffled,96,-0.6,-0.52,0,0.0
|
||||
block,2,shuffled,96,-0.52,-0.43999999999999995,0,0.0
|
||||
block,2,shuffled,96,-0.43999999999999995,-0.36,0,0.0
|
||||
block,2,shuffled,96,-0.36,-0.28,0,0.0
|
||||
block,2,shuffled,96,-0.28,-0.19999999999999996,0,0.0
|
||||
block,2,shuffled,96,-0.19999999999999996,-0.12,0,0.0
|
||||
block,2,shuffled,96,-0.12,-0.040000000000000036,2,0.2604166666666668
|
||||
block,2,shuffled,96,-0.040000000000000036,0.040000000000000036,89,11.588541666666657
|
||||
block,2,shuffled,96,0.040000000000000036,0.1200000000000001,5,0.6510416666666661
|
||||
block,2,shuffled,96,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,2,shuffled,96,0.19999999999999996,0.28,0,0.0
|
||||
block,2,shuffled,96,0.28,0.3600000000000001,0,0.0
|
||||
block,2,shuffled,96,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,2,shuffled,96,0.43999999999999995,0.52,0,0.0
|
||||
block,2,shuffled,96,0.52,0.6000000000000001,0,0.0
|
||||
block,2,shuffled,96,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
block,2,shuffled,96,0.6799999999999999,0.76,0,0.0
|
||||
block,2,shuffled,96,0.76,0.8400000000000001,0,0.0
|
||||
block,2,shuffled,96,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,2,shuffled,96,0.9199999999999999,1.0,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-1.0,-0.92,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.92,-0.84,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.84,-0.76,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.76,-0.6799999999999999,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.6799999999999999,-0.6,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.6,-0.52,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.52,-0.43999999999999995,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.43999999999999995,-0.36,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.36,-0.28,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.28,-0.19999999999999996,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.19999999999999996,-0.12,0,0.0
|
||||
block,2,gaussian_norm_matched,96,-0.12,-0.040000000000000036,3,0.3906250000000002
|
||||
block,2,gaussian_norm_matched,96,-0.040000000000000036,0.040000000000000036,91,11.848958333333323
|
||||
block,2,gaussian_norm_matched,96,0.040000000000000036,0.1200000000000001,2,0.26041666666666646
|
||||
block,2,gaussian_norm_matched,96,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.19999999999999996,0.28,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.28,0.3600000000000001,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.43999999999999995,0.52,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.52,0.6000000000000001,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.6799999999999999,0.76,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.76,0.8400000000000001,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,2,gaussian_norm_matched,96,0.9199999999999999,1.0,0,0.0
|
||||
block,2,random_actual_same_role,156,-1.0,-0.92,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.92,-0.84,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.84,-0.76,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.76,-0.6799999999999999,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.6799999999999999,-0.6,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.6,-0.52,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.52,-0.43999999999999995,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.43999999999999995,-0.36,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.36,-0.28,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.28,-0.19999999999999996,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.19999999999999996,-0.12,0,0.0
|
||||
block,2,random_actual_same_role,156,-0.12,-0.040000000000000036,3,0.2403846153846155
|
||||
block,2,random_actual_same_role,156,-0.040000000000000036,0.040000000000000036,144,11.538461538461528
|
||||
block,2,random_actual_same_role,156,0.040000000000000036,0.1200000000000001,9,0.7211538461538455
|
||||
block,2,random_actual_same_role,156,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,2,random_actual_same_role,156,0.19999999999999996,0.28,0,0.0
|
||||
block,2,random_actual_same_role,156,0.28,0.3600000000000001,0,0.0
|
||||
block,2,random_actual_same_role,156,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,2,random_actual_same_role,156,0.43999999999999995,0.52,0,0.0
|
||||
block,2,random_actual_same_role,156,0.52,0.6000000000000001,0,0.0
|
||||
block,2,random_actual_same_role,156,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
block,2,random_actual_same_role,156,0.6799999999999999,0.76,0,0.0
|
||||
block,2,random_actual_same_role,156,0.76,0.8400000000000001,0,0.0
|
||||
block,2,random_actual_same_role,156,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,2,random_actual_same_role,156,0.9199999999999999,1.0,0,0.0
|
||||
block,3,shuffled,4,-1.0,-0.92,0,0.0
|
||||
block,3,shuffled,4,-0.92,-0.84,0,0.0
|
||||
block,3,shuffled,4,-0.84,-0.76,0,0.0
|
||||
block,3,shuffled,4,-0.76,-0.6799999999999999,0,0.0
|
||||
block,3,shuffled,4,-0.6799999999999999,-0.6,0,0.0
|
||||
block,3,shuffled,4,-0.6,-0.52,0,0.0
|
||||
block,3,shuffled,4,-0.52,-0.43999999999999995,0,0.0
|
||||
block,3,shuffled,4,-0.43999999999999995,-0.36,1,3.1250000000000018
|
||||
block,3,shuffled,4,-0.36,-0.28,0,0.0
|
||||
block,3,shuffled,4,-0.28,-0.19999999999999996,0,0.0
|
||||
block,3,shuffled,4,-0.19999999999999996,-0.12,0,0.0
|
||||
block,3,shuffled,4,-0.12,-0.040000000000000036,0,0.0
|
||||
block,3,shuffled,4,-0.040000000000000036,0.040000000000000036,2,6.249999999999995
|
||||
block,3,shuffled,4,0.040000000000000036,0.1200000000000001,0,0.0
|
||||
block,3,shuffled,4,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,3,shuffled,4,0.19999999999999996,0.28,1,3.1249999999999973
|
||||
block,3,shuffled,4,0.28,0.3600000000000001,0,0.0
|
||||
block,3,shuffled,4,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,3,shuffled,4,0.43999999999999995,0.52,0,0.0
|
||||
block,3,shuffled,4,0.52,0.6000000000000001,0,0.0
|
||||
block,3,shuffled,4,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
block,3,shuffled,4,0.6799999999999999,0.76,0,0.0
|
||||
block,3,shuffled,4,0.76,0.8400000000000001,0,0.0
|
||||
block,3,shuffled,4,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,3,shuffled,4,0.9199999999999999,1.0,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-1.0,-0.92,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.92,-0.84,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.84,-0.76,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.76,-0.6799999999999999,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.6799999999999999,-0.6,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.6,-0.52,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.52,-0.43999999999999995,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.43999999999999995,-0.36,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.36,-0.28,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.28,-0.19999999999999996,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.19999999999999996,-0.12,1,3.1250000000000018
|
||||
block,3,gaussian_norm_matched,4,-0.12,-0.040000000000000036,0,0.0
|
||||
block,3,gaussian_norm_matched,4,-0.040000000000000036,0.040000000000000036,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.040000000000000036,0.1200000000000001,1,3.1249999999999973
|
||||
block,3,gaussian_norm_matched,4,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.19999999999999996,0.28,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.28,0.3600000000000001,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.43999999999999995,0.52,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.52,0.6000000000000001,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.6000000000000001,0.6799999999999999,1,3.1250000000000058
|
||||
block,3,gaussian_norm_matched,4,0.6799999999999999,0.76,1,3.1249999999999973
|
||||
block,3,gaussian_norm_matched,4,0.76,0.8400000000000001,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,3,gaussian_norm_matched,4,0.9199999999999999,1.0,0,0.0
|
||||
block,3,random_actual_same_role,6,-1.0,-0.92,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.92,-0.84,1,2.0833333333333317
|
||||
block,3,random_actual_same_role,6,-0.84,-0.76,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.76,-0.6799999999999999,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.6799999999999999,-0.6,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.6,-0.52,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.52,-0.43999999999999995,1,2.0833333333333317
|
||||
block,3,random_actual_same_role,6,-0.43999999999999995,-0.36,1,2.0833333333333344
|
||||
block,3,random_actual_same_role,6,-0.36,-0.28,1,2.0833333333333344
|
||||
block,3,random_actual_same_role,6,-0.28,-0.19999999999999996,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.19999999999999996,-0.12,1,2.0833333333333344
|
||||
block,3,random_actual_same_role,6,-0.12,-0.040000000000000036,0,0.0
|
||||
block,3,random_actual_same_role,6,-0.040000000000000036,0.040000000000000036,1,2.0833333333333317
|
||||
block,3,random_actual_same_role,6,0.040000000000000036,0.1200000000000001,0,0.0
|
||||
block,3,random_actual_same_role,6,0.1200000000000001,0.19999999999999996,0,0.0
|
||||
block,3,random_actual_same_role,6,0.19999999999999996,0.28,0,0.0
|
||||
block,3,random_actual_same_role,6,0.28,0.3600000000000001,0,0.0
|
||||
block,3,random_actual_same_role,6,0.3600000000000001,0.43999999999999995,0,0.0
|
||||
block,3,random_actual_same_role,6,0.43999999999999995,0.52,0,0.0
|
||||
block,3,random_actual_same_role,6,0.52,0.6000000000000001,0,0.0
|
||||
block,3,random_actual_same_role,6,0.6000000000000001,0.6799999999999999,0,0.0
|
||||
block,3,random_actual_same_role,6,0.6799999999999999,0.76,0,0.0
|
||||
block,3,random_actual_same_role,6,0.76,0.8400000000000001,0,0.0
|
||||
block,3,random_actual_same_role,6,0.8400000000000001,0.9199999999999999,0,0.0
|
||||
block,3,random_actual_same_role,6,0.9199999999999999,1.0,0,0.0
|
||||
|
@@ -0,0 +1,87 @@
|
||||
mode,role,scale,shape,ndim,dtype,verified_pairs,minimum_direct_error,median_direct_error,maximum_cosine,minimum_scaled_error,pairs_under_5pct_direct_error,pairs_under_20pct_direct_error
|
||||
tensor,model.language_model.layers.*.input_layernorm.weight,whole,"(2048,)",1,BF16,74,0.13912108191960726,0.3618018896863233,0.9970857492150444,0.07628898159153481,0,10
|
||||
tensor,model.language_model.layers.*.linear_attn.A_log,whole,"(16,)",1,F32,74,0.27903195486574556,0.5226947550912504,0.9627591388053268,0.2703605752450334,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.conv1d.weight,whole,"(6144, 1, 4)",3,BF16,74,1.098915802713169,1.6166702679005591,0.02807999077877503,0.9995776806560605,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.dt_bias,whole,"(16,)",1,BF16,74,0.5885114617058197,1.0044528421940808,0.8092128713956994,0.5875155561919422,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.in_proj_a.weight,whole,"(16, 2048)",2,BF16,74,1.1349952189534511,1.2168184642463737,0.29759171873785567,0.9546932328966456,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.in_proj_b.weight,whole,"(16, 2048)",2,BF16,74,1.0858901378940746,1.2225962811652291,0.3211769991079106,0.9470191842006356,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,whole,"(6144, 2048)",2,BF16,74,1.2723349817001153,1.4635296420527903,0.0010135353358410714,0.9999989240512659,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.in_proj_z.weight,whole,"(2048, 2048)",2,BF16,74,1.2910830542163632,1.4537858403559945,0.013531813033101601,0.9999084408264775,0,0
|
||||
tensor,model.language_model.layers.*.linear_attn.norm.weight,whole,"(128,)",1,F32,74,0.0996284846101826,0.14453921179048346,0.9962083215944217,0.08699988497707883,0,74
|
||||
tensor,model.language_model.layers.*.linear_attn.out_proj.weight,whole,"(2048, 2048)",2,BF16,74,1.3358019613674468,1.4033556107959,0.0016585943863867917,0.9999986245313849,0,0
|
||||
tensor,model.language_model.layers.*.mlp.down_proj.weight,whole,"(2048, 6144)",2,BF16,74,1.289324616789422,1.3421967541150206,0.0006628169766189873,0.9999997803368037,0,0
|
||||
tensor,model.language_model.layers.*.mlp.gate_proj.weight,whole,"(6144, 2048)",2,BF16,74,1.156201781439196,1.6720196078871572,0.009922418979925494,0.9999507715890752,0,0
|
||||
tensor,model.language_model.layers.*.mlp.up_proj.weight,whole,"(6144, 2048)",2,BF16,73,1.2210613405090442,1.3293668966359742,0.0008737601640509451,0.9999996182715151,0,0
|
||||
tensor,model.language_model.layers.*.post_attention_layernorm.weight,whole,"(2048,)",1,BF16,73,0.08389205016305105,0.33550848021824864,0.9982024397574774,0.05993237240606879,0,15
|
||||
tensor,model.language_model.layers.*.self_attn.k_norm.weight,whole,"(256,)",1,BF16,15,0.4220354869409763,0.5730767172719882,0.9109326955126066,0.41255499542017027,0,0
|
||||
tensor,model.language_model.layers.*.self_attn.k_proj.weight,whole,"(512, 2048)",2,BF16,15,1.3216500796717134,1.390928677608635,0.0029925337418067574,0.9999915875965523,0,0
|
||||
tensor,model.language_model.layers.*.self_attn.o_proj.weight,whole,"(2048, 2048)",2,BF16,15,1.1817960007584458,1.3582179033168547,0.000661789759001042,0.9999997810171335,0,0
|
||||
tensor,model.language_model.layers.*.self_attn.q_norm.weight,whole,"(256,)",1,BF16,15,0.39404064792499355,0.543936200795208,0.9230509542708148,0.38467770382456284,0,0
|
||||
tensor,model.language_model.layers.*.self_attn.q_proj.weight,whole,"(4096, 2048)",2,BF16,15,1.2822624900273465,1.452325507475276,0.031103970688887258,0.9995161544504345,0,0
|
||||
tensor,model.language_model.layers.*.self_attn.v_proj.weight,whole,"(512, 2048)",2,BF16,15,1.1381535663510483,1.279855729883491,0.004274933965921905,0.9999908624280459,0,0
|
||||
tensor,model.visual.layers.*.attn.proj.bias,whole,"(1024,)",1,BF16,73,0.6680979684279816,1.0326843804327857,0.7647843141032159,0.644286390436484,0,0
|
||||
tensor,model.visual.layers.*.attn.proj.weight,whole,"(1024, 1024)",2,BF16,73,1.2263109992320418,1.494030495278536,0.0022893779585961825,0.9999973793708475,0,0
|
||||
tensor,model.visual.layers.*.attn.qkv.bias,whole,"(3072,)",1,BF16,73,1.1508235097225006,1.2895358606896887,0.07880074887800716,0.9968903861389502,0,0
|
||||
tensor,model.visual.layers.*.attn.qkv.weight,whole,"(3072, 1024)",2,BF16,73,1.1919409580791358,1.3528352612266674,0.0012420225434440872,0.9999972512712101,0,0
|
||||
tensor,model.visual.layers.*.mlp.linear_fc1.bias,whole,"(4096,)",1,BF16,73,0.3423255120414103,0.4506608457966791,0.942832226875403,0.33326774816229754,0,0
|
||||
tensor,model.visual.layers.*.mlp.linear_fc1.weight,whole,"(4096, 1024)",2,BF16,73,1.2318420108577686,1.5641645589759754,0.0024998220990413433,0.999996875439855,0,0
|
||||
tensor,model.visual.layers.*.mlp.linear_fc2.bias,whole,"(1024,)",1,BF16,73,0.6856625742029931,1.008978107126644,0.7922147396405718,0.6102424160087705,0,0
|
||||
tensor,model.visual.layers.*.mlp.linear_fc2.weight,whole,"(1024, 4096)",2,BF16,73,1.2583800420385622,1.3660455322298402,0.0019111009657841638,0.9999962185103182,0,0
|
||||
tensor,model.visual.layers.*.norm1.bias,whole,"(1024,)",1,BF16,73,0.25115943947656993,0.690883854809181,0.9691259474486883,0.24656621419343377,0,0
|
||||
tensor,model.visual.layers.*.norm1.weight,whole,"(1024,)",1,BF16,73,0.04823731166106643,0.14504506306481724,0.9992280358422151,0.03928526933736397,1,58
|
||||
tensor,model.visual.layers.*.norm2.bias,whole,"(1024,)",1,BF16,73,0.46245716111033913,0.9605989201912757,0.9255664605094013,0.3785851650290843,0,0
|
||||
tensor,model.visual.layers.*.norm2.weight,whole,"(1024,)",1,BF16,73,0.08109603491499054,0.25584969812385294,0.998097572177166,0.06165416785625297,0,29
|
||||
block,model.language_model.embed_tokens.weight,2d-128,"(128, 128)",2,BF16,274,1.0971431655919834,1.194996165634402,0.38371384285373505,0.9234520490000654,0,0
|
||||
block,model.language_model.embed_tokens.weight,2d-64,"(64, 64)",2,BF16,274,0.7867192520997638,1.2251136172712975,0.6417520096030089,0.7669122232501576,0,0
|
||||
block,model.language_model.layers.*.linear_attn.conv1d.weight,3d-axis0-slice,"(1, 1, 4)",3,BF16,274,0.030015470714371044,0.14107363259145156,0.9999764693972976,0.006860076655215805,12,264
|
||||
block,model.language_model.layers.*.linear_attn.conv1d.weight,3d-volume,"(4, 1, 4)",3,BF16,274,0.20434866638862353,0.6511326882375805,0.9939530725821022,0.10980568976423064,0,0
|
||||
block,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,2d-128,"(128, 128)",2,BF16,274,1.0859194381904604,1.1672768187290392,0.022506491289619568,0.9997466968436706,0,0
|
||||
block,model.language_model.layers.*.linear_attn.in_proj_qkv.weight,2d-64,"(64, 64)",2,BF16,274,1.090431605185219,1.8770856997459073,0.049580432466001016,0.9987701340731432,0,0
|
||||
block,model.language_model.layers.*.linear_attn.in_proj_z.weight,2d-128,"(128, 128)",2,BF16,274,1.1104948912862496,1.8773864848571475,0.023475850982642838,0.9997244042338073,0,0
|
||||
block,model.language_model.layers.*.linear_attn.in_proj_z.weight,2d-64,"(64, 64)",2,BF16,274,1.0991661244633573,1.8514486050124992,0.043485528874022565,0.9988418359874629,0,0
|
||||
block,model.language_model.layers.*.linear_attn.out_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.1951881759983194,1.6594942134980784,0.020468545616965008,0.9997904973744881,0,0
|
||||
block,model.language_model.layers.*.linear_attn.out_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.2072392784651196,1.5582677535715024,0.04454056700388926,0.9990075764932776,0,0
|
||||
block,model.language_model.layers.*.mlp.down_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.2582617322579883,1.3090099938602142,0.023571230885306684,0.9997221599397262,0,0
|
||||
block,model.language_model.layers.*.mlp.down_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.245852433773206,1.2940629124466638,0.055858158466128856,0.9984387142598052,0,0
|
||||
block,model.language_model.layers.*.mlp.gate_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.1461492372241606,1.1966255074601801,0.021946362169951123,0.9997591495892925,0,0
|
||||
block,model.language_model.layers.*.mlp.gate_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.1257085731592742,1.1760753399312671,0.05964405801998305,0.9982197084524573,0,0
|
||||
block,model.language_model.layers.*.mlp.up_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.1938896537465822,1.5743736046705104,0.024746719402858743,0.9996937530457997,0,0
|
||||
block,model.language_model.layers.*.mlp.up_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.1982869032487864,1.2816774375554039,0.05539464318026474,0.9984645379315839,0,0
|
||||
block,model.language_model.layers.*.self_attn.k_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.1432871543196577,1.195761062340063,0.029770641663691994,0.9995567562149396,0,0
|
||||
block,model.language_model.layers.*.self_attn.k_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.089843800559886,2.0769322472882257,0.047714166736051784,0.9988610305206046,0,0
|
||||
block,model.language_model.layers.*.self_attn.o_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.0759865369389314,1.9552667922072795,0.3717312413966341,0.9283403924044872,0,0
|
||||
block,model.language_model.layers.*.self_attn.o_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.090185636068711,1.5183831236960614,0.4006143023357977,0.9162467903157981,0,0
|
||||
block,model.language_model.layers.*.self_attn.q_proj.weight,2d-128,"(128, 128)",2,BF16,274,0.681062120292795,1.6110791655011307,0.766589729840059,0.6421372019309778,0,0
|
||||
block,model.language_model.layers.*.self_attn.q_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.0731967952952182,1.132586437118316,0.14446200523831285,0.9895103481230126,0,0
|
||||
block,model.language_model.layers.*.self_attn.v_proj.weight,2d-128,"(128, 128)",2,BF16,274,1.0703772237415003,2.3064512918392657,0.025583225359911956,0.9996726957260481,0,0
|
||||
block,model.language_model.layers.*.self_attn.v_proj.weight,2d-64,"(64, 64)",2,BF16,274,1.0615939539051586,2.242144803814884,0.05979641283088267,0.9982105935185014,0,0
|
||||
block,model.visual.layers.*.attn.proj.weight,2d-128,"(128, 128)",2,BF16,274,1.0964898212068823,1.24917667593092,0.030377666073744123,0.9995384922072347,0,0
|
||||
block,model.visual.layers.*.attn.proj.weight,2d-64,"(64, 64)",2,BF16,274,1.0929319776557436,1.2036709366228333,0.04190888825600163,0.9991214366057541,0,0
|
||||
block,model.visual.layers.*.attn.qkv.weight,2d-128,"(128, 128)",2,BF16,274,1.0525674864684573,1.1472629166116397,0.3497290628302071,0.9368508860068955,0,0
|
||||
block,model.visual.layers.*.attn.qkv.weight,2d-64,"(64, 64)",2,BF16,274,1.0276302940676463,2.153982940996772,0.05039306408132999,0.998729462413368,0,0
|
||||
block,model.visual.layers.*.mlp.linear_fc1.weight,2d-128,"(128, 128)",2,BF16,274,1.1784037671219423,1.2610017357650904,0.05446205204482453,0.998515841079683,0,0
|
||||
block,model.visual.layers.*.mlp.linear_fc1.weight,2d-64,"(64, 64)",2,BF16,274,1.1930932764469049,1.5635216450228222,0.05742888587637769,0.9983495996227965,0,0
|
||||
block,model.visual.layers.*.mlp.linear_fc2.weight,2d-128,"(128, 128)",2,BF16,274,1.1976986200156163,1.5864288735783176,0.028190387759862247,0.9996025720444844,0,0
|
||||
block,model.visual.layers.*.mlp.linear_fc2.weight,2d-64,"(64, 64)",2,BF16,274,1.0973576240937362,1.2821071329134202,0.052752794427996734,0.9986076019538593,0,0
|
||||
block,model.visual.merger.linear_fc1.weight,2d-128,"(128, 128)",2,BF16,274,0.5690482377069183,1.3881733201164015,0.8394467875146913,0.5434418928747256,0,0
|
||||
block,model.visual.merger.linear_fc1.weight,2d-64,"(64, 64)",2,BF16,274,0.5704160177838491,1.383715288781834,0.8382741355593843,0.5452490015141402,0,0
|
||||
block,model.visual.merger.linear_fc2.weight,2d-128,"(128, 128)",2,BF16,274,1.361602880922912,1.4076521005397016,0.022716837303296537,0.9997419393538193,0,0
|
||||
block,model.visual.merger.linear_fc2.weight,2d-64,"(64, 64)",2,BF16,274,1.3172335252228955,1.3885238301845173,0.054916767376716406,0.9984909356929044,0,0
|
||||
block,model.visual.pos_embed.weight,2d-128,"(128, 128)",2,BF16,273,0.1894907479098447,0.8368814065208369,0.9820068756666321,0.18884516446935973,0,1
|
||||
block,model.visual.pos_embed.weight,2d-64,"(64, 64)",2,BF16,273,0.09494430310075026,0.5761056230742875,0.9960502566825586,0.08879125048454295,0,7
|
||||
block,mtp.fc.weight,2d-128,"(128, 128)",2,BF16,273,0.39795158187541674,1.3009198418830403,0.9219877695675962,0.3872189984592301,0,0
|
||||
block,mtp.fc.weight,2d-64,"(64, 64)",2,BF16,273,0.2839409442692466,1.2187888824373985,0.9598132531637373,0.2806394823456665,0,0
|
||||
block,mtp.layers.*.mlp.down_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.310074009535121,1.4215054026060296,0.027554671685217264,0.9996202979473355,0,0
|
||||
block,mtp.layers.*.mlp.down_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.249807578942172,1.470099001092512,0.044434762604782845,0.9990122881487777,0,0
|
||||
block,mtp.layers.*.mlp.gate_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.3014802290213505,1.361681996563559,0.03373996398972023,0.9994306453326176,0,0
|
||||
block,mtp.layers.*.mlp.gate_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.2564681140844673,1.4625957375534147,0.0542156946795996,0.9985292476689946,0,0
|
||||
block,mtp.layers.*.mlp.up_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.3314815136128675,1.4083678884650157,0.029758300533378034,0.9995571237049762,0,0
|
||||
block,mtp.layers.*.mlp.up_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.2939027995891361,1.3558402418466908,0.06246493086010471,0.9980471594131423,0,0
|
||||
block,mtp.layers.*.self_attn.k_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.2211071346906675,1.6568460478464606,0.02773941669941069,0.9996151883405816,0,0
|
||||
block,mtp.layers.*.self_attn.k_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.100749972037444,2.321267616896819,0.06391398313313451,0.9979554112083653,0,0
|
||||
block,mtp.layers.*.self_attn.o_proj.weight,2d-128,"(128, 128)",2,BF16,273,0.8821045427467973,1.206170586612079,0.5767316801239848,0.8169336381502268,0,0
|
||||
block,mtp.layers.*.self_attn.o_proj.weight,2d-64,"(64, 64)",2,BF16,273,0.9425522849964715,1.2696601113099386,0.5275664580383422,0.849513762309286,0,0
|
||||
block,mtp.layers.*.self_attn.q_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.0237777819118037,2.8371550549613556,0.4898664655238783,0.8717974798972196,0,0
|
||||
block,mtp.layers.*.self_attn.q_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.008280496287895,2.524395456205866,0.5430402363306829,0.8397066760041365,0,0
|
||||
block,mtp.layers.*.self_attn.v_proj.weight,2d-128,"(128, 128)",2,BF16,273,1.3534677309181036,1.402389437951931,0.024838060691519125,0.9996914877806474,0,0
|
||||
block,mtp.layers.*.self_attn.v_proj.weight,2d-64,"(64, 64)",2,BF16,273,1.269834274085324,1.3381727215333372,0.05798552649229015,0.9983174238274177,0,0
|
||||
|
@@ -0,0 +1,30 @@
|
||||
{
|
||||
"model": "Qwen/Qwen3.5-2B",
|
||||
"revision": "15852e8c16360a2fea060d615a32b45270f8a8fc",
|
||||
"tensor_count": 632,
|
||||
"parameter_count": 2274069824,
|
||||
"model_weight_bytes": 4548144832,
|
||||
"ndim_counts": {
|
||||
"1": 319,
|
||||
"2": 294,
|
||||
"3": 18,
|
||||
"5": 1
|
||||
},
|
||||
"elapsed_seconds": 6661.129063449334,
|
||||
"analysis_budget_seconds": 6600,
|
||||
"peak_rss_kib": 206180,
|
||||
"python": "3.14.4",
|
||||
"numpy": "2.5.4",
|
||||
"seed": 20261109,
|
||||
"limitations": [
|
||||
"No inference or functional equivalence measured",
|
||||
"No exhaustive pair search",
|
||||
"Block search uses per-role/shape/scale reservoirs of 256",
|
||||
"Different block scales overlap; savings must not be added",
|
||||
"Block savings count only verified disjoint sampled blocks, no extrapolation",
|
||||
"Spectral probes use aligned bounded subvolumes, not whole tensors",
|
||||
"Shared encoding is theoretical, not a runnable compressed checkpoint",
|
||||
"No cross-role grouping, neuron permutation alignment, or training"
|
||||
],
|
||||
"completed": true
|
||||
}
|
||||
@@ -0,0 +1,10 @@
|
||||
{
|
||||
"eligible_pairs_in_screened_objects": 6159,
|
||||
"candidate_pairs_before_cap": 4724,
|
||||
"selected_pairs": 2000,
|
||||
"objects_screened": 632,
|
||||
"pairs_verified": 2000,
|
||||
"timed_out": false,
|
||||
"screen_sample_elements": 2048,
|
||||
"candidate_neighbors_per_metric": 10
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,6 @@
|
||||
{
|
||||
"sampled_objects": 13328,
|
||||
"raw_block_bytes": 256387072,
|
||||
"fingerprint_float32_bytes": 105009152,
|
||||
"status": "metadata-only estimate for future SOP; no cache allocated in this run"
|
||||
}
|
||||
@@ -0,0 +1,83 @@
|
||||
{
|
||||
"REPORT.md": "363fb5d44d2298482a2cc926bc8ef8e3541551184a058194cdbfdc9083710725",
|
||||
"analysis/block_coverage.json": "4e0545ecb252e46047d6a6e05171dc0e0c0baa5554fe192847d9d2bb92bca1cf",
|
||||
"analysis/block_nulls.json": "5d661965efb61544f35c1897cd534300a7b7724d275f7faf90dfb76b1ac25d00",
|
||||
"analysis/block_objects.json": "277b115ca4b9cd1be77338d77dda2809741236e8232fba2ae87855de125ce0e4",
|
||||
"analysis/block_pairs.csv": "734e647ec472e555c47ad007f373b99860cbf8aa78a10ad300dba017bb749b67",
|
||||
"analysis/block_partitions.json": "670a6889d35c561aa59c254c6b230889b040e89a693f740a7463d0f88c0f58c1",
|
||||
"analysis/block_random_pairs.json": "5aba9411dfae0aa3854a6d2353bf93e3771ecc621c1ffe33fb33664aea03b2cf",
|
||||
"analysis/block_sampling.csv": "59e76e3746508bf6457aa8ee7cfb9314c272f80328173585a9293c537d161939",
|
||||
"analysis/block_structure.json": "6f4a71455bba4194691384ac641e035512992889288b06551b376a9af05b99f2",
|
||||
"analysis/compression_curve.csv": "7d114aa45efbf65878d277d45289c72a14d22d035a5ef383c84ab6fbfb4aae57",
|
||||
"analysis/dimensional_bytes.csv": "14743c7ff10fe25d43a06827c3672acfefe390a3b51e2c07445a770cf8e71145",
|
||||
"analysis/groups_block_2d-128_0.01.json": "2efb8330d7259bae21e7c756f65f4d782ab1cd83e452d8a08fc1dc44beb9a3d6",
|
||||
"analysis/groups_block_2d-128_0.02.json": "2efb8330d7259bae21e7c756f65f4d782ab1cd83e452d8a08fc1dc44beb9a3d6",
|
||||
"analysis/groups_block_2d-128_0.05.json": "2efb8330d7259bae21e7c756f65f4d782ab1cd83e452d8a08fc1dc44beb9a3d6",
|
||||
"analysis/groups_block_2d-128_0.10.json": "2efb8330d7259bae21e7c756f65f4d782ab1cd83e452d8a08fc1dc44beb9a3d6",
|
||||
"analysis/groups_block_2d-128_0.20.json": "4abe6cb3ad2f14f0ed1008427aa77e773caccaaa96d3ad64c4c500a9895ecd20",
|
||||
"analysis/groups_block_2d-64_0.01.json": "9dbd00eaf141ab60a83ee9ed14c05d3f52c6c7cea6501d7f3418c42cbef325f3",
|
||||
"analysis/groups_block_2d-64_0.02.json": "9dbd00eaf141ab60a83ee9ed14c05d3f52c6c7cea6501d7f3418c42cbef325f3",
|
||||
"analysis/groups_block_2d-64_0.05.json": "9dbd00eaf141ab60a83ee9ed14c05d3f52c6c7cea6501d7f3418c42cbef325f3",
|
||||
"analysis/groups_block_2d-64_0.10.json": "599ba97d39c14702f4c23cb1e164dbddf891f5df6a053e15c7bfd7166b4aa055",
|
||||
"analysis/groups_block_2d-64_0.20.json": "a765328c620659b359b51bb6a1a06851602fed7fa386bcd32234c61a6585229f",
|
||||
"analysis/groups_block_3d-axis0-slice_0.01.json": "396edfd2d1648b89048ec886963958e3824731b9345b7f9dc6c6422bbb77f84e",
|
||||
"analysis/groups_block_3d-axis0-slice_0.02.json": "396edfd2d1648b89048ec886963958e3824731b9345b7f9dc6c6422bbb77f84e",
|
||||
"analysis/groups_block_3d-axis0-slice_0.05.json": "8a1bf98871d2597e9f0599fae7bcd61346570469b776448429027a52906e09e6",
|
||||
"analysis/groups_block_3d-axis0-slice_0.10.json": "f5d13d0ce0816bcf4f800d8963fc0cface6049dcc73a69d6e291cae5e6379ed8",
|
||||
"analysis/groups_block_3d-axis0-slice_0.20.json": "47257dcdc94cb1bacc8e7322d3c7dc8a178afc1518d9797c508c7a0a78f77bc4",
|
||||
"analysis/groups_block_3d-volume_0.01.json": "2bed9bf147d216f008e4048c58722d2c9e89c0c97baa6c8973e589d31f6457fa",
|
||||
"analysis/groups_block_3d-volume_0.02.json": "2bed9bf147d216f008e4048c58722d2c9e89c0c97baa6c8973e589d31f6457fa",
|
||||
"analysis/groups_block_3d-volume_0.05.json": "2bed9bf147d216f008e4048c58722d2c9e89c0c97baa6c8973e589d31f6457fa",
|
||||
"analysis/groups_block_3d-volume_0.10.json": "2bed9bf147d216f008e4048c58722d2c9e89c0c97baa6c8973e589d31f6457fa",
|
||||
"analysis/groups_block_3d-volume_0.20.json": "4f7427dcf1c8587af13bc461e5c77f89be3bd024ccdb0eb6d3134bcde1d98d18",
|
||||
"analysis/groups_tensor_whole_0.01.json": "cd7261fae777a13e20117423b9a949c8402d114fed636c14ccacf410e6c596c3",
|
||||
"analysis/groups_tensor_whole_0.02.json": "cd7261fae777a13e20117423b9a949c8402d114fed636c14ccacf410e6c596c3",
|
||||
"analysis/groups_tensor_whole_0.05.json": "d230d7a15a996ca7f2e7c62301b9bb6643ac38b0fce3dc28c6301865cd998349",
|
||||
"analysis/groups_tensor_whole_0.10.json": "9c089be52e48eff72a6ba3e450a2e6994d76846a1a65f7bd0d44bcb3bced9987",
|
||||
"analysis/groups_tensor_whole_0.20.json": "c0517c58a9a8ab8018aa6133e00455db01beaefb5fa8e2bfa0052f6491cb1c85",
|
||||
"analysis/heatmap_pairs.csv": "ecfd7d37c33c134fcf1b59f6db0e5b001e7c04bb7fb9caca98cbc2df8499e456",
|
||||
"analysis/inventory.csv": "a4f530d785cd981f698eb87824874caab422517d30f3967f6b6b695a7c90410d",
|
||||
"analysis/layer_distance_stratified.csv": "a285f77fadf09875662060888cbdf0c92790cc969c30fe52cc176e7e2c1e7224",
|
||||
"analysis/null_histograms.csv": "3079258c40f055e020b1d96376ca4b7dff07c392c013278a3eaf97cfd04533a8",
|
||||
"analysis/role_summary.csv": "bc43ff33945a6909ca3a8e36648d382e9257c662ca65e4b22a5316e3009410fc",
|
||||
"analysis/summary.json": "438ce5950cd96885874990c6e27f52b941ebf37cea22ac5b1dc8d391129aca43",
|
||||
"analysis/tensor_coverage.json": "24796f378e34c818236865535404f5d310014b1565b1adf13f0971f1df0d10c1",
|
||||
"analysis/tensor_nulls.json": "801ba89d0262d21b571b3cfd29d2147217e23058ff880ab5cf8c5b6acff482d7",
|
||||
"analysis/tensor_objects.json": "e0c628ca0091a945b3bef7277141449f8aee2eff65225fb990bef926f36bf1a8",
|
||||
"analysis/tensor_pairs.csv": "e49e2981fb54e747d3e8018ec1f658419aa4aeccb1b895c91533e9e6fe471a00",
|
||||
"analysis/tensor_random_pairs.json": "a8d529fb0e887c55173652de176461ef2b3e6c1c26c1a30725f278fce4460d47",
|
||||
"analysis/tensor_structure.json": "f7cb1b8c1961ca56a8c42c5723b57af513e481c724f29fa1866033273dd42c6c",
|
||||
"analysis-run.txt": "3a6ddce89d0671de549208633a690978aa949372ae74f4981348bb24b0699e05",
|
||||
"cache-plan.json": "6374df74161df928ed343d6b4e91480aeaa40a3953ec189042f1fc88cd112948",
|
||||
"code_checksums.txt": "26be13ef79b544677312b13b6a1368fed28bbb8e333d0c372c6c51b2402587d0",
|
||||
"download-progress.txt": "01d95fa94e9e0ddbf9d2a78bb2fa2fa7a04571248f77fb978312f805fdd3df00",
|
||||
"environment.txt": "0c008fa401973b3947d0dcd208832189d9c3903cbc38fad524f507507c69065f",
|
||||
"execution-environment.json": "c2b5b0a080a51342441119735f7d0d9306841ddb903974d3254938810ec574e9",
|
||||
"figures/block_structure.pdf": "00aa70e155953a53cfb346ba3f82b9162a19cd7cfb26c105f802638e6cf71247",
|
||||
"figures/block_structure.png": "fce689e0216462a0fd9056038d56769834e1423efc6c7708777f35ae7fa58f11",
|
||||
"figures/block_structure.svg": "3082a77c4955ad2a6f3a76779a9a35da23eb21a316e56a77d3a24ecd00e1bbc9",
|
||||
"figures/candidate_distributions.pdf": "e48b803eb960e885cb772ecfa2dd117f9919417a6f35d14f1f4be630af0ccf4f",
|
||||
"figures/candidate_distributions.png": "b6268182faaf296ac9764007e5cd5a376cb9031a1190d6be07594c021f2705cd",
|
||||
"figures/candidate_distributions.svg": "68fba9fb299040c4c8bb293cac72d0bc3bbef84020736120f6e0daf07c2c4066",
|
||||
"figures/compression_frontier.pdf": "1fc04547ea39f8c16e816eb5316695892f966925b339c449903d00d00b97a22e",
|
||||
"figures/compression_frontier.png": "3fbe57dea35b70a880e45fb1d0a9a4e39c222695b47659883d809a32b558b002",
|
||||
"figures/compression_frontier.svg": "9659408c957af29c9126e3b0d945306b253dfd6c87e64c6a014d5fd7e4230126",
|
||||
"figures/layer_distance.pdf": "b8587f8ac424b36abc50c668767f5bd76a22a3eb301d2c808e08c74bfe9ef823",
|
||||
"figures/layer_distance.png": "d044431d18e9e07cc87c920709a755450edbb5a33a48edddecd89227159b11c4",
|
||||
"figures/layer_distance.svg": "5df5c364fc7c7fdab18fa0356b8b3aafc8cb7e72d34170becd719c98896ad33d",
|
||||
"figures/layer_heatmap.pdf": "cd66493a56dcf39779429496a3266f9883677d2f6d8e6d2b9008f9455b497285",
|
||||
"figures/layer_heatmap.png": "d3c0365e43591392053dcc35eb6e3ffdf46dad63a17a82818884be30d18e1aa9",
|
||||
"figures/layer_heatmap.svg": "96b1684b184f46abbd366885eefc6ce35b545b8e3f18028ca473f5308fd75ed0",
|
||||
"figures/null_controls.pdf": "0263314e9ee9e97efcab13a1bf13e0131ebfe3dd762923762b32591bf7a4f41a",
|
||||
"figures/null_controls.png": "c6e0a2c49a224755dacb99eb9b8aa4114b5ae6cb81cbbc34bc3364167a5ee8f3",
|
||||
"figures/null_controls.svg": "69c81590f1dd80ebc0f8c3d2b3e7f68e25e8f047202d995386d63da336e32883",
|
||||
"manifest.json": "91d1499838f33edf37740cdb153bd230cf72e70c3bae010f05148b3f807a4f1d",
|
||||
"model_config.json": "ed1c1723241f23f7f4e23430759cbd7dcfb4103cbdfe052bfe7626b57c2615b4",
|
||||
"numerical-qc.json": "86b5169282532735cabbc01f564fa5051b24551e472f3fd28d97e0f5b9041cf4",
|
||||
"plot-run.txt": "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
|
||||
"resource-observation.txt": "e23842f9b58b44f5d6895e2737107772e4bb7cbc8998b7db5e49727e848e12e9",
|
||||
"source_metadata.json": "cf804d729756175621903d11df43b1a459ea6356ed3bd72d7d41197f468d217e",
|
||||
"supervisor-run.txt": "e437f12a397a353a1c0df9e70700201c1daf4d548e0c8d5f51740b463238fcbc",
|
||||
"test-results.txt": "d1492070356dccaa22bb4e4dceef7b0528e313312166796612b1c36ff6cf65f9",
|
||||
"tests-during-io-timeout.txt": "aacfbae843953ac28bb510f75651ebcd50645c144b442d833f92703819cdf8c4"
|
||||
}
|
||||
@@ -0,0 +1,6 @@
|
||||
{
|
||||
"deleted_unix": 1794167040.2914257,
|
||||
"deleted_bytes": 4548351685,
|
||||
"deleted_directory": "weights",
|
||||
"reason": "User-authorized delete-after-analysis; archived results validated"
|
||||
}
|
||||
@@ -0,0 +1,10 @@
|
||||
e5c82881f90f07330bf23c6783f3c7df4c949dccc267f20b95a2426c3e4957d4 pilot/__init__.py
|
||||
2715316acd6e7c89daf0501c3bb0ceeaddcc6e163cf34cab1bfd501003549ba7 pilot/analyze.py
|
||||
4dd30fe71f7041d2f34560e8460c0bfd4830d6894fc0afff4789f0d7a451b47a pilot/audit.py
|
||||
c542d952c47c1d0c59e873763f7b5698ac3475cff2b387b1c170184aa2f4d3d2 pilot/download.py
|
||||
95e8e05024c0afde2e96b684bdd068cdacead320af7063ed61acf374fe2e1830 pilot/plot.py
|
||||
4644772f82d4267556607316ab5603520a4ad7df7fd53f2ce905d23ac7a1ac04 pilot/run.py
|
||||
2e620849501b6bc98a59cbf9c39495326b0c85530e18e6166aa9edb68ac2423e pilot/tensors.py
|
||||
e683fedbe5c34678818f9ec74154f9d2e608dc63b16a581be40db5d2d909b156 pilot/validate.py
|
||||
af9c9ea6a882ecdb9816a50416e4e777ff38447e9caf3b8e4c630ce2f227d324 requirements.txt
|
||||
1d465c58ef0ff09f7216eaa6d3559b300d9e968009cc891ad8bd0090c5eeed98 tests/test_pilot.py
|
||||
@@ -0,0 +1,67 @@
|
||||
Downloading README.md, expected 62814 bytes
|
||||
Downloading config.json, expected 2908 bytes
|
||||
Downloading model.safetensors-00001-of-00001.safetensors, expected 4548221488 bytes
|
||||
Weight download 465567744/4548221488 bytes (10.2%)
|
||||
Weight download 532676608/4548221488 bytes (11.7%)
|
||||
Weight download 599785472/4548221488 bytes (13.2%)
|
||||
Weight download 666894336/4548221488 bytes (14.7%)
|
||||
Weight download 734003200/4548221488 bytes (16.1%)
|
||||
Weight download 801112064/4548221488 bytes (17.6%)
|
||||
Weight download 868220928/4548221488 bytes (19.1%)
|
||||
Weight download 935329792/4548221488 bytes (20.6%)
|
||||
Weight download 1002438656/4548221488 bytes (22.0%)
|
||||
Weight download 1069547520/4548221488 bytes (23.5%)
|
||||
Weight download 1136656384/4548221488 bytes (25.0%)
|
||||
Weight download 1203765248/4548221488 bytes (26.5%)
|
||||
Weight download 1270874112/4548221488 bytes (27.9%)
|
||||
Weight download 1337982976/4548221488 bytes (29.4%)
|
||||
Weight download 1405091840/4548221488 bytes (30.9%)
|
||||
Weight download 1472200704/4548221488 bytes (32.4%)
|
||||
Weight download 1539309568/4548221488 bytes (33.8%)
|
||||
Weight download 1606418432/4548221488 bytes (35.3%)
|
||||
Weight download 1673527296/4548221488 bytes (36.8%)
|
||||
Weight download 1740636160/4548221488 bytes (38.3%)
|
||||
Weight download 1807745024/4548221488 bytes (39.7%)
|
||||
Weight download 1874853888/4548221488 bytes (41.2%)
|
||||
Weight download 1941962752/4548221488 bytes (42.7%)
|
||||
Weight download 2009071616/4548221488 bytes (44.2%)
|
||||
Weight download 2076180480/4548221488 bytes (45.6%)
|
||||
Weight download 2143289344/4548221488 bytes (47.1%)
|
||||
Weight download 2210398208/4548221488 bytes (48.6%)
|
||||
Weight download 2277507072/4548221488 bytes (50.1%)
|
||||
Weight download 2344615936/4548221488 bytes (51.6%)
|
||||
Weight download 2411724800/4548221488 bytes (53.0%)
|
||||
Weight download 2478833664/4548221488 bytes (54.5%)
|
||||
Weight download 2545942528/4548221488 bytes (56.0%)
|
||||
Weight download 2613051392/4548221488 bytes (57.5%)
|
||||
Weight download 2680160256/4548221488 bytes (58.9%)
|
||||
Weight download 2747269120/4548221488 bytes (60.4%)
|
||||
Weight download 2814377984/4548221488 bytes (61.9%)
|
||||
Weight download 2881486848/4548221488 bytes (63.4%)
|
||||
Weight download 2948595712/4548221488 bytes (64.8%)
|
||||
Weight download 3015704576/4548221488 bytes (66.3%)
|
||||
Weight download 3082813440/4548221488 bytes (67.8%)
|
||||
Weight download 3149922304/4548221488 bytes (69.3%)
|
||||
Weight download 3217031168/4548221488 bytes (70.7%)
|
||||
Weight download 3284140032/4548221488 bytes (72.2%)
|
||||
Weight download 3351248896/4548221488 bytes (73.7%)
|
||||
Weight download 3418357760/4548221488 bytes (75.2%)
|
||||
Weight download 3485466624/4548221488 bytes (76.6%)
|
||||
Weight download 3552575488/4548221488 bytes (78.1%)
|
||||
Weight download 3619684352/4548221488 bytes (79.6%)
|
||||
Weight download 3686793216/4548221488 bytes (81.1%)
|
||||
Weight download 3753902080/4548221488 bytes (82.5%)
|
||||
Weight download 3821010944/4548221488 bytes (84.0%)
|
||||
Weight download 3888119808/4548221488 bytes (85.5%)
|
||||
Weight download 3955228672/4548221488 bytes (87.0%)
|
||||
Weight download 4022337536/4548221488 bytes (88.4%)
|
||||
Weight download 4089446400/4548221488 bytes (89.9%)
|
||||
Weight download 4156555264/4548221488 bytes (91.4%)
|
||||
Weight download 4223664128/4548221488 bytes (92.9%)
|
||||
Weight download 4290772992/4548221488 bytes (94.3%)
|
||||
Weight download 4357881856/4548221488 bytes (95.8%)
|
||||
Weight download 4414003760/4548221488 bytes (97.0%)
|
||||
Weight download 4481112624/4548221488 bytes (98.5%)
|
||||
Weight download 4548221488/4548221488 bytes (100.0%)
|
||||
Downloading model.safetensors.index.json, expected 64460 bytes
|
||||
{"revision": "15852e8c16360a2fea060d615a32b45270f8a8fc", "bytes": 4548351670}
|
||||
@@ -0,0 +1,11 @@
|
||||
contourpy==1.4.0
|
||||
cycler==0.12.1
|
||||
fonttools==4.68.0
|
||||
kiwisolver==1.5.1
|
||||
matplotlib==3.11.2
|
||||
numpy==2.5.4
|
||||
packaging==26.3
|
||||
pillow==13.0.0
|
||||
pyparsing==3.3.3
|
||||
python-dateutil==4.0.0
|
||||
tzdata==2026.6
|
||||
@@ -0,0 +1,18 @@
|
||||
{
|
||||
"recorded_unix": 1794167980.891606,
|
||||
"python": "3.14.4",
|
||||
"platform": "Linux-7.0.0-31-generic-x86_64-with-glibc2.43",
|
||||
"machine": "x86_64",
|
||||
"logical_cpus": 2,
|
||||
"memory_snapshot": {
|
||||
"MemTotal_KiB": 1777468,
|
||||
"MemAvailable_KiB": 786108,
|
||||
"SwapTotal_KiB": 4194300,
|
||||
"SwapFree_KiB": 1831728
|
||||
},
|
||||
"blas_threads": 1,
|
||||
"assertions_enabled": true,
|
||||
"analysis_io_priority": "best-effort 7, applied during block verification",
|
||||
"analysis_code_commit": "6121016",
|
||||
"note": "Shared host; see resource-observation.txt. Memory snapshot is post-analysis."
|
||||
}
|
||||
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 124 KiB |
@@ -0,0 +1,520 @@
|
||||
<?xml version="1.0" encoding="utf-8" standalone="no"?>
|
||||
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"
|
||||
"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
|
||||
<svg xmlns:xlink="http://www.w3.org/1999/xlink" width="782.653531pt" height="280.599375pt" viewBox="0 0 782.653531 280.599375" xmlns="http://www.w3.org/2000/svg" version="1.1">
|
||||
<metadata>
|
||||
<rdf:RDF xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:cc="http://creativecommons.org/ns#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
|
||||
<cc:Work>
|
||||
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
|
||||
<dc:date>2026-11-09T03:59:26.496031</dc:date>
|
||||
<dc:format>image/svg+xml</dc:format>
|
||||
<dc:creator>
|
||||
<cc:Agent>
|
||||
<dc:title>Matplotlib v3.11.2, https://matplotlib.org/</dc:title>
|
||||
</cc:Agent>
|
||||
</dc:creator>
|
||||
</cc:Work>
|
||||
</rdf:RDF>
|
||||
</metadata>
|
||||
<defs>
|
||||
<style type="text/css">*{stroke-linejoin: round; stroke-linecap: butt}</style>
|
||||
</defs>
|
||||
<g id="figure_1">
|
||||
<g id="patch_1">
|
||||
<path d="M 0 280.599375
|
||||
L 782.653531 280.599375
|
||||
L 782.653531 0
|
||||
L 0 0
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="axes_1">
|
||||
<g id="patch_2">
|
||||
<path d="M 44.103906 242.398594
|
||||
L 380.921687 242.398594
|
||||
L 380.921687 22.318125
|
||||
L 44.103906 22.318125
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="PathCollection_1">
|
||||
<defs>
|
||||
<path id="md3003eeb70" d="M 0 2.12132
|
||||
C 0.562581 2.12132 1.102195 1.897805 1.5 1.5
|
||||
C 1.897805 1.102195 2.12132 0.562581 2.12132 0
|
||||
C 2.12132 -0.562581 1.897805 -1.102195 1.5 -1.5
|
||||
C 1.102195 -1.897805 0.562581 -2.12132 0 -2.12132
|
||||
C -0.562581 -2.12132 -1.102195 -1.897805 -1.5 -1.5
|
||||
C -1.897805 -1.102195 -2.12132 -0.562581 -2.12132 0
|
||||
C -2.12132 0.562581 -1.897805 1.102195 -1.5 1.5
|
||||
C -1.102195 1.897805 -0.562581 2.12132 0 2.12132
|
||||
z
|
||||
" style="stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p69fa569ed2)">
|
||||
<use xlink:href="#md3003eeb70" x="77.66076" y="151.024635" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="365.611788" y="135.69158" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="131.43602" y="194.38461" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="72.110842" y="215.931271" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="114.321766" y="188.977693" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="108.153667" y="185.291856" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="70.917488" y="183.878303" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="73.662124" y="192.718159" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="59.413805" y="208.644234" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="71.45921" y="201.943057" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="72.762737" y="178.485883" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="67.931206" y="211.698412" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="71.140257" y="186.437172" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="67.285551" y="173.105978" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="84.350762" y="207.211942" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="116.227649" y="174.716091" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="65.430375" y="172.365879" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="68.530652" y="151.39984" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="64.857563" y="32.321783" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="61.122482" y="202.682848" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="PathCollection_2">
|
||||
<defs>
|
||||
<path id="m72270fb06d" d="M 0 2.12132
|
||||
C 0.562581 2.12132 1.102195 1.897805 1.5 1.5
|
||||
C 1.897805 1.102195 2.12132 0.562581 2.12132 0
|
||||
C 2.12132 -0.562581 1.897805 -1.102195 1.5 -1.5
|
||||
C 1.102195 -1.897805 0.562581 -2.12132 0 -2.12132
|
||||
C -0.562581 -2.12132 -1.102195 -1.897805 -1.5 -1.5
|
||||
C -1.897805 -1.102195 -2.12132 -0.562581 -2.12132 0
|
||||
C -2.12132 0.562581 -1.897805 1.102195 -1.5 1.5
|
||||
C -1.102195 1.897805 -0.562581 2.12132 0 2.12132
|
||||
z
|
||||
" style="stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p69fa569ed2)">
|
||||
<use xlink:href="#m72270fb06d" x="77.66076" y="128.668363" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="365.611788" y="129.499784" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="131.43602" y="206.625629" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="72.110842" y="194.491831" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="114.321766" y="199.207014" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="108.153667" y="232.394936" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="70.917488" y="198.326897" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="73.662124" y="171.060058" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="59.413805" y="189.64184" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="71.45921" y="211.180771" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="72.762737" y="210.934243" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="67.931206" y="193.481749" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="71.140257" y="188.904511" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="67.285551" y="186.997829" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="84.350762" y="169.216329" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="116.227649" y="195.742622" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="65.430375" y="173.358431" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="68.530652" y="139.179176" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="64.857563" y="38.513348" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="61.122482" y="210.93641" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_1">
|
||||
<g id="xtick_1">
|
||||
<g id="line2d_1">
|
||||
<defs>
|
||||
<path id="m0dd6f91fe9" d="M 0 0
|
||||
L 0 3.5
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="44.805569" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_1">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="44.805569" y="256.99625" transform="rotate(-0 44.805569 256.99625)">0.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_2">
|
||||
<g id="line2d_2">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="111.623891" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_2">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="111.623891" y="256.99625" transform="rotate(-0 111.623891 256.99625)">0.2</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_3">
|
||||
<g id="line2d_3">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="178.442214" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_3">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="178.442214" y="256.99625" transform="rotate(-0 178.442214 256.99625)">0.4</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_4">
|
||||
<g id="line2d_4">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="245.260536" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_4">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="245.260536" y="256.99625" transform="rotate(-0 245.260536 256.99625)">0.6</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_5">
|
||||
<g id="line2d_5">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="312.078858" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_5">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="312.078858" y="256.99625" transform="rotate(-0 312.078858 256.99625)">0.8</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_6">
|
||||
<g id="line2d_6">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="378.89718" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_6">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="378.89718" y="256.99625" transform="rotate(-0 378.89718 256.99625)">1.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_7">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="212.512797" y="270.997031" transform="rotate(-0 212.512797 270.997031)">Normalized spectrum distance</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_2">
|
||||
<g id="ytick_1">
|
||||
<g id="line2d_7">
|
||||
<defs>
|
||||
<path id="ma0bb63b275" d="M 0 0
|
||||
L -3.5 0
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="44.103906" y="206.168603" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_8">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="209.967431" transform="rotate(-0 37.103906 209.967431)">0.1</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_2">
|
||||
<g id="line2d_8">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="44.103906" y="169.300808" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_9">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="173.099636" transform="rotate(-0 37.103906 173.099636)">0.2</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_3">
|
||||
<g id="line2d_9">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="44.103906" y="132.433012" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_10">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="136.23184" transform="rotate(-0 37.103906 136.23184)">0.3</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_4">
|
||||
<g id="line2d_10">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="44.103906" y="95.565216" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_11">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="99.364044" transform="rotate(-0 37.103906 99.364044)">0.4</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_5">
|
||||
<g id="line2d_11">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="44.103906" y="58.697421" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_12">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="62.496249" transform="rotate(-0 37.103906 62.496249)">0.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_13">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="14.798438" y="132.358359" transform="rotate(-90 14.798438 132.358359)">Leading left subspace overlap (k <= 4)</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_3">
|
||||
<path d="M 44.103906 242.398594
|
||||
L 44.103906 22.318125
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_4">
|
||||
<path d="M 44.103906 242.398594
|
||||
L 380.921687 242.398594
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="text_14">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="212.512797" y="16.318125" transform="rotate(-0 212.512797 16.318125)">2D blocks: bounded mode-unfolding probes</text>
|
||||
</g>
|
||||
<g id="legend_1">
|
||||
<g id="patch_5">
|
||||
<path d="M 319.400437 52.719375
|
||||
L 375.321687 52.719375
|
||||
Q 376.921687 52.719375 376.921687 51.119375
|
||||
L 376.921687 27.918125
|
||||
Q 376.921687 26.318125 375.321687 26.318125
|
||||
L 319.400437 26.318125
|
||||
Q 317.800437 26.318125 317.800437 27.918125
|
||||
L 317.800437 51.119375
|
||||
Q 317.800437 52.719375 319.400437 52.719375
|
||||
z
|
||||
" style="fill: #ffffff; opacity: 0.8; stroke: #cccccc; stroke-linejoin: miter"/>
|
||||
</g>
|
||||
<g id="PathCollection_3">
|
||||
<g>
|
||||
<use xlink:href="#md3003eeb70" x="329.000437" y="33.496875" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_15">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="343.400437" y="35.596875" transform="rotate(-0 343.400437 35.596875)">mode 0</text>
|
||||
</g>
|
||||
<g id="PathCollection_4">
|
||||
<g>
|
||||
<use xlink:href="#m72270fb06d" x="329.000437" y="45.4975" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_16">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="343.400437" y="47.5975" transform="rotate(-0 343.400437 47.5975)">mode 1</text>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="axes_2">
|
||||
<g id="patch_6">
|
||||
<path d="M 435.547049 242.398594
|
||||
L 772.36483 242.398594
|
||||
L 772.36483 22.318125
|
||||
L 435.547049 22.318125
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="PathCollection_5">
|
||||
<g clip-path="url(#pa6ff4b5385)">
|
||||
<use xlink:href="#md3003eeb70" x="499.351613" y="188.218445" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="543.451078" y="159.314531" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="450.856948" y="162.947028" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="539.560188" y="43.541889" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="524.351448" y="137.712444" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="514.45854" y="40.61397" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="757.054931" y="100.401598" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="523.457859" y="181.668886" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="561.744084" y="158.310407" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#md3003eeb70" x="455.743599" y="232.394936" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="PathCollection_6">
|
||||
<g clip-path="url(#pa6ff4b5385)">
|
||||
<use xlink:href="#m72270fb06d" x="499.351613" y="61.870396" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="543.451078" y="32.321783" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="450.856948" y="124.45232" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="539.560188" y="146.971873" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="524.351448" y="81.464671" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="514.45854" y="34.77306" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="757.054931" y="48.906255" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="523.457859" y="194.352189" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="561.744084" y="93.570571" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
<use xlink:href="#m72270fb06d" x="455.743599" y="190.123775" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_3">
|
||||
<g id="xtick_7">
|
||||
<g id="line2d_12">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="444.750019" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_17">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="444.750019" y="256.99625" transform="rotate(-0 444.750019 256.99625)">0.00</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_8">
|
||||
<g id="line2d_13">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="498.011803" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_18">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="498.011803" y="256.99625" transform="rotate(-0 498.011803 256.99625)">0.02</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_9">
|
||||
<g id="line2d_14">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="551.273586" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_19">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="551.273586" y="256.99625" transform="rotate(-0 551.273586 256.99625)">0.04</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_10">
|
||||
<g id="line2d_15">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="604.535369" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_20">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="604.535369" y="256.99625" transform="rotate(-0 604.535369 256.99625)">0.06</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_11">
|
||||
<g id="line2d_16">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="657.797152" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_21">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="657.797152" y="256.99625" transform="rotate(-0 657.797152 256.99625)">0.08</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_12">
|
||||
<g id="line2d_17">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="711.058936" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_22">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="711.058936" y="256.99625" transform="rotate(-0 711.058936 256.99625)">0.10</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_13">
|
||||
<g id="line2d_18">
|
||||
<g>
|
||||
<use xlink:href="#m0dd6f91fe9" x="764.320719" y="242.398594" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_23">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="764.320719" y="256.99625" transform="rotate(-0 764.320719 256.99625)">0.12</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_24">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="603.955939" y="270.997031" transform="rotate(-0 603.955939 270.997031)">Normalized spectrum distance</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_4">
|
||||
<g id="ytick_6">
|
||||
<g id="line2d_19">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="231.788538" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_25">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="235.587367" transform="rotate(-0 428.547049 235.587367)">0.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_7">
|
||||
<g id="line2d_20">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="191.132101" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_26">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="194.930929" transform="rotate(-0 428.547049 194.930929)">0.6</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_8">
|
||||
<g id="line2d_21">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="150.475663" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_27">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="154.274491" transform="rotate(-0 428.547049 154.274491)">0.7</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_9">
|
||||
<g id="line2d_22">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="109.819225" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_28">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="113.618053" transform="rotate(-0 428.547049 113.618053)">0.8</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_10">
|
||||
<g id="line2d_23">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="69.162788" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_29">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="72.961616" transform="rotate(-0 428.547049 72.961616)">0.9</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_11">
|
||||
<g id="line2d_24">
|
||||
<g>
|
||||
<use xlink:href="#ma0bb63b275" x="435.547049" y="28.50635" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_30">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="428.547049" y="32.305178" transform="rotate(-0 428.547049 32.305178)">1.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_31">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="406.24158" y="132.358359" transform="rotate(-90 406.24158 132.358359)">Leading left subspace overlap (k <= 4)</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_7">
|
||||
<path d="M 435.547049 242.398594
|
||||
L 435.547049 22.318125
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_8">
|
||||
<path d="M 435.547049 242.398594
|
||||
L 772.36483 242.398594
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="text_32">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="603.955939" y="16.318125" transform="rotate(-0 603.955939 16.318125)">3D blocks: bounded mode-unfolding probes</text>
|
||||
</g>
|
||||
<g id="legend_2">
|
||||
<g id="patch_9">
|
||||
<path d="M 710.84358 52.719375
|
||||
L 766.76483 52.719375
|
||||
Q 768.36483 52.719375 768.36483 51.119375
|
||||
L 768.36483 27.918125
|
||||
Q 768.36483 26.318125 766.76483 26.318125
|
||||
L 710.84358 26.318125
|
||||
Q 709.24358 26.318125 709.24358 27.918125
|
||||
L 709.24358 51.119375
|
||||
Q 709.24358 52.719375 710.84358 52.719375
|
||||
z
|
||||
" style="fill: #ffffff; opacity: 0.8; stroke: #cccccc; stroke-linejoin: miter"/>
|
||||
</g>
|
||||
<g id="PathCollection_7">
|
||||
<g>
|
||||
<use xlink:href="#md3003eeb70" x="720.44358" y="33.496875" style="fill: #1f77b4; fill-opacity: 0.65; stroke: #1f77b4; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_33">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="734.84358" y="35.596875" transform="rotate(-0 734.84358 35.596875)">mode 0</text>
|
||||
</g>
|
||||
<g id="PathCollection_8">
|
||||
<g>
|
||||
<use xlink:href="#m72270fb06d" x="720.44358" y="45.4975" style="fill: #ff7f0e; fill-opacity: 0.65; stroke: #ff7f0e; stroke-opacity: 0.65"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_34">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="734.84358" y="47.5975" transform="rotate(-0 734.84358 47.5975)">mode 2</text>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id="p69fa569ed2">
|
||||
<rect x="44.103906" y="22.318125" width="336.817781" height="220.080469"/>
|
||||
</clipPath>
|
||||
<clipPath id="pa6ff4b5385">
|
||||
<rect x="435.547049" y="22.318125" width="336.817781" height="220.080469"/>
|
||||
</clipPath>
|
||||
</defs>
|
||||
</svg>
|
||||
|
After Width: | Height: | Size: 31 KiB |
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 161 KiB |
File diff suppressed because it is too large
Load Diff
|
After Width: | Height: | Size: 68 KiB |
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 150 KiB |
@@ -0,0 +1,673 @@
|
||||
<?xml version="1.0" encoding="utf-8" standalone="no"?>
|
||||
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"
|
||||
"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
|
||||
<svg xmlns:xlink="http://www.w3.org/1999/xlink" width="783.589062pt" height="285.638437pt" viewBox="0 0 783.589062 285.638437" xmlns="http://www.w3.org/2000/svg" version="1.1">
|
||||
<metadata>
|
||||
<rdf:RDF xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:cc="http://creativecommons.org/ns#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
|
||||
<cc:Work>
|
||||
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
|
||||
<dc:date>2026-11-09T03:58:43.824497</dc:date>
|
||||
<dc:format>image/svg+xml</dc:format>
|
||||
<dc:creator>
|
||||
<cc:Agent>
|
||||
<dc:title>Matplotlib v3.11.2, https://matplotlib.org/</dc:title>
|
||||
</cc:Agent>
|
||||
</dc:creator>
|
||||
</cc:Work>
|
||||
</rdf:RDF>
|
||||
</metadata>
|
||||
<defs>
|
||||
<style type="text/css">*{stroke-linejoin: round; stroke-linecap: butt}</style>
|
||||
</defs>
|
||||
<g id="figure_1">
|
||||
<g id="patch_1">
|
||||
<path d="M 0 285.638437
|
||||
L 783.589062 285.638437
|
||||
L 783.589062 0
|
||||
L 0 0
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="axes_1">
|
||||
<g id="patch_2">
|
||||
<path d="M 63.191406 247.437656
|
||||
L 397.189062 247.437656
|
||||
L 397.189062 35.04
|
||||
L 63.191406 35.04
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="matplotlib.axis_1">
|
||||
<g id="xtick_1">
|
||||
<g id="line2d_1">
|
||||
<defs>
|
||||
<path id="m81e100d430" d="M 0 0
|
||||
L 0 3.5
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="102.344242" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_1">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="102.344242" y="262.035312" transform="rotate(-0 102.344242 262.035312)">2.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_2">
|
||||
<g id="line2d_2">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="142.296114" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_2">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="142.296114" y="262.035312" transform="rotate(-0 142.296114 262.035312)">5.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_3">
|
||||
<g id="line2d_3">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="182.247987" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_3">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="182.247987" y="262.035312" transform="rotate(-0 182.247987 262.035312)">7.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_4">
|
||||
<g id="line2d_4">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="222.19986" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_4">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="222.19986" y="262.035312" transform="rotate(-0 222.19986 262.035312)">10.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_5">
|
||||
<g id="line2d_5">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="262.151733" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_5">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="262.151733" y="262.035312" transform="rotate(-0 262.151733 262.035312)">12.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_6">
|
||||
<g id="line2d_6">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="302.103605" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_6">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="302.103605" y="262.035312" transform="rotate(-0 302.103605 262.035312)">15.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_7">
|
||||
<g id="line2d_7">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="342.055478" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_7">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="342.055478" y="262.035312" transform="rotate(-0 342.055478 262.035312)">17.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_8">
|
||||
<g id="line2d_8">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="382.007351" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_8">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="382.007351" y="262.035312" transform="rotate(-0 382.007351 262.035312)">20.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_9">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="230.190234" y="276.036094" transform="rotate(-0 230.190234 276.036094)">Maximum relative weight error (%)</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_2">
|
||||
<g id="ytick_1">
|
||||
<g id="line2d_9">
|
||||
<defs>
|
||||
<path id="m664b6f87e3" d="M 0 0
|
||||
L -3.5 0
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="237.783217" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_10">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="241.582045" transform="rotate(-0 56.191406 241.582045)">0.0000</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_2">
|
||||
<g id="line2d_10">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="201.015242" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_11">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="204.81407" transform="rotate(-0 56.191406 204.81407)">0.0005</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_3">
|
||||
<g id="line2d_11">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="164.247267" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_12">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="168.046095" transform="rotate(-0 56.191406 168.046095)">0.0010</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_4">
|
||||
<g id="line2d_12">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="127.479292" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_13">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="131.27812" transform="rotate(-0 56.191406 131.27812)">0.0015</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_5">
|
||||
<g id="line2d_13">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="90.711317" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_14">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="94.510145" transform="rotate(-0 56.191406 94.510145)">0.0020</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_6">
|
||||
<g id="line2d_14">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="63.191406" y="53.943342" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_15">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="56.191406" y="57.74217" transform="rotate(-0 56.191406 57.74217)">0.0025</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_16">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="14.798438" y="141.238828" transform="rotate(-90 14.798438 141.238828)">Verified net savings / model bytes (%)</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_15">
|
||||
<path d="M 78.373118 237.783217
|
||||
L 94.353867 237.783217
|
||||
L 142.296114 237.783217
|
||||
L 222.19986 237.783217
|
||||
L 382.007351 237.783217
|
||||
" clip-path="url(#pcc885c0aa3)" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m3e1248f28c" d="M 0 3
|
||||
C 0.795609 3 1.55874 2.683901 2.12132 2.12132
|
||||
C 2.683901 1.55874 3 0.795609 3 0
|
||||
C 3 -0.795609 2.683901 -1.55874 2.12132 -2.12132
|
||||
C 1.55874 -2.683901 0.795609 -3 0 -3
|
||||
C -0.795609 -3 -1.55874 -2.683901 -2.12132 -2.12132
|
||||
C -2.683901 -1.55874 -3 -0.795609 -3 0
|
||||
C -3 0.795609 -2.683901 1.55874 -2.12132 2.12132
|
||||
C -1.55874 2.683901 -0.795609 3 0 3
|
||||
z
|
||||
" style="stroke: #1f77b4"/>
|
||||
</defs>
|
||||
<g clip-path="url(#pcc885c0aa3)">
|
||||
<use xlink:href="#m3e1248f28c" x="78.373118" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="94.353867" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="142.296114" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="222.19986" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="382.007351" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_16">
|
||||
<path d="M 78.373118 237.783217
|
||||
L 94.353867 237.783217
|
||||
L 142.296114 237.783217
|
||||
L 222.19986 237.783217
|
||||
L 382.007351 237.783217
|
||||
" clip-path="url(#pcc885c0aa3)" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m2065f0e306" d="M 0 3
|
||||
C 0.795609 3 1.55874 2.683901 2.12132 2.12132
|
||||
C 2.683901 1.55874 3 0.795609 3 0
|
||||
C 3 -0.795609 2.683901 -1.55874 2.12132 -2.12132
|
||||
C 1.55874 -2.683901 0.795609 -3 0 -3
|
||||
C -0.795609 -3 -1.55874 -2.683901 -2.12132 -2.12132
|
||||
C -2.683901 -1.55874 -3 -0.795609 -3 0
|
||||
C -3 0.795609 -2.683901 1.55874 -2.12132 2.12132
|
||||
C -1.55874 2.683901 -0.795609 3 0 3
|
||||
z
|
||||
" style="stroke: #ff7f0e"/>
|
||||
</defs>
|
||||
<g clip-path="url(#pcc885c0aa3)">
|
||||
<use xlink:href="#m2065f0e306" x="78.373118" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="94.353867" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="142.296114" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="222.19986" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="382.007351" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_17">
|
||||
<path d="M 78.373118 237.783217
|
||||
L 94.353867 237.783217
|
||||
L 142.296114 237.783217
|
||||
L 222.19986 237.783217
|
||||
L 382.007351 237.783217
|
||||
" clip-path="url(#pcc885c0aa3)" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m2bc7cbfb87" d="M 0 3
|
||||
C 0.795609 3 1.55874 2.683901 2.12132 2.12132
|
||||
C 2.683901 1.55874 3 0.795609 3 0
|
||||
C 3 -0.795609 2.683901 -1.55874 2.12132 -2.12132
|
||||
C 1.55874 -2.683901 0.795609 -3 0 -3
|
||||
C -0.795609 -3 -1.55874 -2.683901 -2.12132 -2.12132
|
||||
C -2.683901 -1.55874 -3 -0.795609 -3 0
|
||||
C -3 0.795609 -2.683901 1.55874 -2.12132 2.12132
|
||||
C -1.55874 2.683901 -0.795609 3 0 3
|
||||
z
|
||||
" style="stroke: #2ca02c"/>
|
||||
</defs>
|
||||
<g clip-path="url(#pcc885c0aa3)">
|
||||
<use xlink:href="#m2bc7cbfb87" x="78.373118" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="94.353867" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="142.296114" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="222.19986" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="382.007351" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_18">
|
||||
<path d="M 78.373118 237.783217
|
||||
L 94.353867 237.783217
|
||||
L 142.296114 237.783217
|
||||
L 222.19986 237.783217
|
||||
L 382.007351 237.783217
|
||||
" clip-path="url(#pcc885c0aa3)" style="fill: none; stroke: #d62728; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m4ba93d814b" d="M 0 3
|
||||
C 0.795609 3 1.55874 2.683901 2.12132 2.12132
|
||||
C 2.683901 1.55874 3 0.795609 3 0
|
||||
C 3 -0.795609 2.683901 -1.55874 2.12132 -2.12132
|
||||
C 1.55874 -2.683901 0.795609 -3 0 -3
|
||||
C -0.795609 -3 -1.55874 -2.683901 -2.12132 -2.12132
|
||||
C -2.683901 -1.55874 -3 -0.795609 -3 0
|
||||
C -3 0.795609 -2.683901 1.55874 -2.12132 2.12132
|
||||
C -1.55874 2.683901 -0.795609 3 0 3
|
||||
z
|
||||
" style="stroke: #d62728"/>
|
||||
</defs>
|
||||
<g clip-path="url(#pcc885c0aa3)">
|
||||
<use xlink:href="#m4ba93d814b" x="78.373118" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="94.353867" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="142.296114" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="222.19986" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="382.007351" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_19">
|
||||
<path d="M 78.373118 237.783217
|
||||
L 94.353867 237.783217
|
||||
L 142.296114 237.783217
|
||||
L 222.19986 184.90628
|
||||
L 382.007351 44.694439
|
||||
" clip-path="url(#pcc885c0aa3)" style="fill: none; stroke: #9467bd; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m564b1aa843" d="M 0 3
|
||||
C 0.795609 3 1.55874 2.683901 2.12132 2.12132
|
||||
C 2.683901 1.55874 3 0.795609 3 0
|
||||
C 3 -0.795609 2.683901 -1.55874 2.12132 -2.12132
|
||||
C 1.55874 -2.683901 0.795609 -3 0 -3
|
||||
C -0.795609 -3 -1.55874 -2.683901 -2.12132 -2.12132
|
||||
C -2.683901 -1.55874 -3 -0.795609 -3 0
|
||||
C -3 0.795609 -2.683901 1.55874 -2.12132 2.12132
|
||||
C -1.55874 2.683901 -0.795609 3 0 3
|
||||
z
|
||||
" style="stroke: #9467bd"/>
|
||||
</defs>
|
||||
<g clip-path="url(#pcc885c0aa3)">
|
||||
<use xlink:href="#m564b1aa843" x="78.373118" y="237.783217" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="94.353867" y="237.783217" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="142.296114" y="237.783217" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="222.19986" y="184.90628" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="382.007351" y="44.694439" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_3">
|
||||
<path d="M 63.191406 247.437656
|
||||
L 63.191406 35.04
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_4">
|
||||
<path d="M 63.191406 247.437656
|
||||
L 397.189062 247.437656
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="legend_1">
|
||||
<g id="patch_5">
|
||||
<path d="M 68.791406 101.443125
|
||||
L 149.383906 101.443125
|
||||
Q 150.983906 101.443125 150.983906 99.843125
|
||||
L 150.983906 40.64
|
||||
Q 150.983906 39.04 149.383906 39.04
|
||||
L 68.791406 39.04
|
||||
Q 67.191406 39.04 67.191406 40.64
|
||||
L 67.191406 99.843125
|
||||
Q 67.191406 101.443125 68.791406 101.443125
|
||||
z
|
||||
" style="fill: #ffffff; opacity: 0.8; stroke: #cccccc; stroke-linejoin: miter"/>
|
||||
</g>
|
||||
<g id="line2d_20">
|
||||
<path d="M 70.391406 45.51875
|
||||
L 78.391406 45.51875
|
||||
L 86.391406 45.51875
|
||||
" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m3e1248f28c" x="78.391406" y="45.51875" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_17">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="92.791406" y="48.31875" transform="rotate(-0 92.791406 48.31875)">2d-128</text>
|
||||
</g>
|
||||
<g id="line2d_21">
|
||||
<path d="M 70.391406 57.519375
|
||||
L 78.391406 57.519375
|
||||
L 86.391406 57.519375
|
||||
" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m2065f0e306" x="78.391406" y="57.519375" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_18">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="92.791406" y="60.319375" transform="rotate(-0 92.791406 60.319375)">2d-64</text>
|
||||
</g>
|
||||
<g id="line2d_22">
|
||||
<path d="M 70.391406 69.52
|
||||
L 78.391406 69.52
|
||||
L 86.391406 69.52
|
||||
" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m2bc7cbfb87" x="78.391406" y="69.52" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_19">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="92.791406" y="72.32" transform="rotate(-0 92.791406 72.32)">3d-axis0-slice</text>
|
||||
</g>
|
||||
<g id="line2d_23">
|
||||
<path d="M 70.391406 81.520625
|
||||
L 78.391406 81.520625
|
||||
L 86.391406 81.520625
|
||||
" style="fill: none; stroke: #d62728; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m4ba93d814b" x="78.391406" y="81.520625" style="fill: #d62728; stroke: #d62728"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_20">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="92.791406" y="84.320625" transform="rotate(-0 92.791406 84.320625)">3d-volume</text>
|
||||
</g>
|
||||
<g id="line2d_24">
|
||||
<path d="M 70.391406 93.52125
|
||||
L 78.391406 93.52125
|
||||
L 86.391406 93.52125
|
||||
" style="fill: none; stroke: #9467bd; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m564b1aa843" x="78.391406" y="93.52125" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_21">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="92.791406" y="96.32125" transform="rotate(-0 92.791406 96.32125)">whole</text>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="axes_2">
|
||||
<g id="patch_6">
|
||||
<path d="M 442.391406 247.437656
|
||||
L 776.389062 247.437656
|
||||
L 776.389062 35.04
|
||||
L 442.391406 35.04
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="matplotlib.axis_3">
|
||||
<g id="xtick_9">
|
||||
<g id="line2d_25">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="481.544242" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_22">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="481.544242" y="262.035312" transform="rotate(-0 481.544242 262.035312)">2.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_10">
|
||||
<g id="line2d_26">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="521.496114" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_23">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="521.496114" y="262.035312" transform="rotate(-0 521.496114 262.035312)">5.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_11">
|
||||
<g id="line2d_27">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="561.447987" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_24">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="561.447987" y="262.035312" transform="rotate(-0 561.447987 262.035312)">7.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_12">
|
||||
<g id="line2d_28">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="601.39986" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_25">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="601.39986" y="262.035312" transform="rotate(-0 601.39986 262.035312)">10.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_13">
|
||||
<g id="line2d_29">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="641.351733" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_26">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="641.351733" y="262.035312" transform="rotate(-0 641.351733 262.035312)">12.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_14">
|
||||
<g id="line2d_30">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="681.303605" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_27">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="681.303605" y="262.035312" transform="rotate(-0 681.303605 262.035312)">15.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_15">
|
||||
<g id="line2d_31">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="721.255478" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_28">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="721.255478" y="262.035312" transform="rotate(-0 721.255478 262.035312)">17.5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_16">
|
||||
<g id="line2d_32">
|
||||
<g>
|
||||
<use xlink:href="#m81e100d430" x="761.207351" y="247.437656" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_29">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="761.207351" y="262.035312" transform="rotate(-0 761.207351 262.035312)">20.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_30">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="609.390234" y="276.036094" transform="rotate(-0 609.390234 276.036094)">Maximum relative weight error (%)</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_4">
|
||||
<g id="ytick_7">
|
||||
<g id="line2d_33">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="442.391406" y="237.783217" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_31">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="435.391406" y="241.582045" transform="rotate(-0 435.391406 241.582045)">0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_8">
|
||||
<g id="line2d_34">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="442.391406" y="194.8746" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_32">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="435.391406" y="198.673428" transform="rotate(-0 435.391406 198.673428)">10</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_9">
|
||||
<g id="line2d_35">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="442.391406" y="151.965982" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_33">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="435.391406" y="155.764811" transform="rotate(-0 435.391406 155.764811)">20</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_10">
|
||||
<g id="line2d_36">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="442.391406" y="109.057365" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_34">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="435.391406" y="112.856193" transform="rotate(-0 435.391406 112.856193)">30</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_11">
|
||||
<g id="line2d_37">
|
||||
<g>
|
||||
<use xlink:href="#m664b6f87e3" x="442.391406" y="66.148748" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_35">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="435.391406" y="69.947576" transform="rotate(-0 435.391406 69.947576)">40</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_36">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="416.264062" y="141.238828" transform="rotate(-90 416.264062 141.238828)">Non-singleton sharing groups</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_38">
|
||||
<path d="M 457.573118 237.783217
|
||||
L 473.553867 237.783217
|
||||
L 521.496114 237.783217
|
||||
L 601.39986 237.783217
|
||||
L 761.207351 233.492356
|
||||
" clip-path="url(#pc8f73d1103)" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#pc8f73d1103)">
|
||||
<use xlink:href="#m3e1248f28c" x="457.573118" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="473.553867" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="521.496114" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="601.39986" y="237.783217" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m3e1248f28c" x="761.207351" y="233.492356" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_39">
|
||||
<path d="M 457.573118 237.783217
|
||||
L 473.553867 237.783217
|
||||
L 521.496114 237.783217
|
||||
L 601.39986 233.492356
|
||||
L 761.207351 224.910632
|
||||
" clip-path="url(#pc8f73d1103)" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#pc8f73d1103)">
|
||||
<use xlink:href="#m2065f0e306" x="457.573118" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="473.553867" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="521.496114" y="237.783217" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="601.39986" y="233.492356" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#m2065f0e306" x="761.207351" y="224.910632" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_40">
|
||||
<path d="M 457.573118 237.783217
|
||||
L 473.553867 237.783217
|
||||
L 521.496114 199.165462
|
||||
L 601.39986 96.18478
|
||||
L 761.207351 44.694439
|
||||
" clip-path="url(#pc8f73d1103)" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#pc8f73d1103)">
|
||||
<use xlink:href="#m2bc7cbfb87" x="457.573118" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="473.553867" y="237.783217" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="521.496114" y="199.165462" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="601.39986" y="96.18478" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m2bc7cbfb87" x="761.207351" y="44.694439" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_41">
|
||||
<path d="M 457.573118 237.783217
|
||||
L 473.553867 237.783217
|
||||
L 521.496114 237.783217
|
||||
L 601.39986 237.783217
|
||||
L 761.207351 229.201494
|
||||
" clip-path="url(#pc8f73d1103)" style="fill: none; stroke: #d62728; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#pc8f73d1103)">
|
||||
<use xlink:href="#m4ba93d814b" x="457.573118" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="473.553867" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="521.496114" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="601.39986" y="237.783217" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#m4ba93d814b" x="761.207351" y="229.201494" style="fill: #d62728; stroke: #d62728"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_42">
|
||||
<path d="M 457.573118 237.783217
|
||||
L 473.553867 237.783217
|
||||
L 521.496114 233.492356
|
||||
L 601.39986 190.583738
|
||||
L 761.207351 169.129429
|
||||
" clip-path="url(#pc8f73d1103)" style="fill: none; stroke: #9467bd; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#pc8f73d1103)">
|
||||
<use xlink:href="#m564b1aa843" x="457.573118" y="237.783217" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="473.553867" y="237.783217" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="521.496114" y="233.492356" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="601.39986" y="190.583738" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
<use xlink:href="#m564b1aa843" x="761.207351" y="169.129429" style="fill: #9467bd; stroke: #9467bd"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_7">
|
||||
<path d="M 442.391406 247.437656
|
||||
L 442.391406 35.04
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_8">
|
||||
<path d="M 442.391406 247.437656
|
||||
L 776.389062 247.437656
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_37">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="391.189062" y="16.318125" transform="rotate(-0 391.189062 16.318125)">Conservative verified sharing; block scales are separate alternatives</text>
|
||||
</g>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id="pcc885c0aa3">
|
||||
<rect x="63.191406" y="35.04" width="333.997656" height="212.397656"/>
|
||||
</clipPath>
|
||||
<clipPath id="pc8f73d1103">
|
||||
<rect x="442.391406" y="35.04" width="333.997656" height="212.397656"/>
|
||||
</clipPath>
|
||||
</defs>
|
||||
</svg>
|
||||
|
After Width: | Height: | Size: 33 KiB |
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 137 KiB |
@@ -0,0 +1,812 @@
|
||||
<?xml version="1.0" encoding="utf-8" standalone="no"?>
|
||||
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN"
|
||||
"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
|
||||
<svg xmlns:xlink="http://www.w3.org/1999/xlink" width="856.301563pt" height="267.38058pt" viewBox="0 0 856.301563 267.38058" xmlns="http://www.w3.org/2000/svg" version="1.1">
|
||||
<metadata>
|
||||
<rdf:RDF xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:cc="http://creativecommons.org/ns#" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
|
||||
<cc:Work>
|
||||
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
|
||||
<dc:date>2026-11-09T03:59:06.813796</dc:date>
|
||||
<dc:format>image/svg+xml</dc:format>
|
||||
<dc:creator>
|
||||
<cc:Agent>
|
||||
<dc:title>Matplotlib v3.11.2, https://matplotlib.org/</dc:title>
|
||||
</cc:Agent>
|
||||
</dc:creator>
|
||||
</cc:Work>
|
||||
</rdf:RDF>
|
||||
</metadata>
|
||||
<defs>
|
||||
<style type="text/css">*{stroke-linejoin: round; stroke-linecap: butt}</style>
|
||||
</defs>
|
||||
<g id="figure_1">
|
||||
<g id="patch_1">
|
||||
<path d="M 0 267.38058
|
||||
L 856.301563 267.38058
|
||||
L 856.301563 0
|
||||
L 0 0
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="axes_1">
|
||||
<g id="patch_2">
|
||||
<path d="M 44.103906 196.19145
|
||||
L 432.351562 196.19145
|
||||
L 432.351562 50.157188
|
||||
L 44.103906 50.157188
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="matplotlib.axis_1">
|
||||
<g id="xtick_1">
|
||||
<g id="line2d_1">
|
||||
<defs>
|
||||
<path id="m20431a40b1" d="M 0 0
|
||||
L 0 3.5
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="45.708235" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_1">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="45.708235" y="210.789106" transform="rotate(-0 45.708235 210.789106)">0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_2">
|
||||
<g id="line2d_2">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="125.924693" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_2">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="125.924693" y="210.789106" transform="rotate(-0 125.924693 210.789106)">5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_3">
|
||||
<g id="line2d_3">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="206.141151" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_3">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="206.141151" y="210.789106" transform="rotate(-0 206.141151 210.789106)">10</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_4">
|
||||
<g id="line2d_4">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="286.357609" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_4">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="286.357609" y="210.789106" transform="rotate(-0 286.357609 210.789106)">15</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_5">
|
||||
<g id="line2d_5">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="366.574067" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_5">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="366.574067" y="210.789106" transform="rotate(-0 366.574067 210.789106)">20</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_6">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="238.227734" y="224.789887" transform="rotate(-0 238.227734 224.789887)">Absolute layer distance</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_2">
|
||||
<g id="ytick_1">
|
||||
<g id="line2d_6">
|
||||
<defs>
|
||||
<path id="m65c4260567" d="M 0 0
|
||||
L -3.5 0
|
||||
" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</defs>
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="183.492818" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_7">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="187.291647" transform="rotate(-0 37.103906 187.291647)">0.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_2">
|
||||
<g id="line2d_7">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="158.095555" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_8">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="161.894384" transform="rotate(-0 37.103906 161.894384)">0.2</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_3">
|
||||
<g id="line2d_8">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="132.698292" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_9">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="136.497121" transform="rotate(-0 37.103906 136.497121)">0.4</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_4">
|
||||
<g id="line2d_9">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="107.301029" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_10">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="111.099857" transform="rotate(-0 37.103906 111.099857)">0.6</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_5">
|
||||
<g id="line2d_10">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="81.903766" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_11">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="85.702594" transform="rotate(-0 37.103906 85.702594)">0.8</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_6">
|
||||
<g id="line2d_11">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="44.103906" y="56.506503" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_12">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: end" x="37.103906" y="60.305331" transform="rotate(-0 37.103906 60.305331)">1.0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_13">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="14.798438" y="123.174319" transform="rotate(-90 14.798438 123.174319)">Median exact cosine</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_12">
|
||||
<path d="M 61.751527 64.045284
|
||||
L 77.794819 67.644244
|
||||
L 93.83811 69.049782
|
||||
L 109.881402 69.159729
|
||||
L 125.924693 69.792107
|
||||
L 141.967985 70.738966
|
||||
L 158.011276 72.039402
|
||||
L 174.054568 71.420815
|
||||
L 190.09786 69.737898
|
||||
L 206.141151 71.401544
|
||||
L 222.184443 79.088666
|
||||
L 238.227734 76.632265
|
||||
L 254.271026 75.347617
|
||||
L 270.314318 101.342621
|
||||
L 286.357609 109.415676
|
||||
L 302.400901 82.494061
|
||||
L 318.444192 159.045581
|
||||
L 334.487484 177.640834
|
||||
L 350.530775 171.503668
|
||||
L 366.574067 119.707806
|
||||
L 382.617359 70.032843
|
||||
L 398.66065 182.998523
|
||||
L 414.703942 180.349992
|
||||
" clip-path="url(#p9cb1e34570)" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m077c66be99" d="M 0 1.5
|
||||
C 0.397805 1.5 0.77937 1.341951 1.06066 1.06066
|
||||
C 1.341951 0.77937 1.5 0.397805 1.5 0
|
||||
C 1.5 -0.397805 1.341951 -0.77937 1.06066 -1.06066
|
||||
C 0.77937 -1.341951 0.397805 -1.5 0 -1.5
|
||||
C -0.397805 -1.5 -0.77937 -1.341951 -1.06066 -1.06066
|
||||
C -1.341951 -0.77937 -1.5 -0.397805 -1.5 0
|
||||
C -1.5 0.397805 -1.341951 0.77937 -1.06066 1.06066
|
||||
C -0.77937 1.341951 -0.397805 1.5 0 1.5
|
||||
z
|
||||
" style="stroke: #1f77b4"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p9cb1e34570)">
|
||||
<use xlink:href="#m077c66be99" x="61.751527" y="64.045284" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="77.794819" y="67.644244" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="93.83811" y="69.049782" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="109.881402" y="69.159729" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="125.924693" y="69.792107" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="141.967985" y="70.738966" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="158.011276" y="72.039402" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="174.054568" y="71.420815" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="190.09786" y="69.737898" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="206.141151" y="71.401544" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="222.184443" y="79.088666" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="238.227734" y="76.632265" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="254.271026" y="75.347617" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="270.314318" y="101.342621" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="286.357609" y="109.415676" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="302.400901" y="82.494061" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="318.444192" y="159.045581" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="334.487484" y="177.640834" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="350.530775" y="171.503668" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="366.574067" y="119.707806" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="382.617359" y="70.032843" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="398.66065" y="182.998523" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="414.703942" y="180.349992" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_13">
|
||||
<path d="M 61.751527 158.397688
|
||||
L 77.794819 166.734483
|
||||
L 93.83811 168.747962
|
||||
L 109.881402 183.250135
|
||||
L 125.924693 183.428552
|
||||
L 141.967985 183.440704
|
||||
L 158.011276 183.460342
|
||||
L 174.054568 183.458574
|
||||
L 190.09786 183.45641
|
||||
L 206.141151 183.483699
|
||||
L 222.184443 183.465226
|
||||
L 238.227734 183.48573
|
||||
L 254.271026 183.462584
|
||||
L 270.314318 183.463043
|
||||
L 286.357609 183.481273
|
||||
L 302.400901 183.477096
|
||||
L 318.444192 183.467116
|
||||
L 334.487484 183.485477
|
||||
L 350.530775 183.443151
|
||||
L 366.574067 183.459105
|
||||
L 382.617359 183.49194
|
||||
L 398.66065 183.525188
|
||||
L 414.703942 183.536436
|
||||
" clip-path="url(#p9cb1e34570)" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="me63c40fafa" d="M 0 1.5
|
||||
C 0.397805 1.5 0.77937 1.341951 1.06066 1.06066
|
||||
C 1.341951 0.77937 1.5 0.397805 1.5 0
|
||||
C 1.5 -0.397805 1.341951 -0.77937 1.06066 -1.06066
|
||||
C 0.77937 -1.341951 0.397805 -1.5 0 -1.5
|
||||
C -0.397805 -1.5 -0.77937 -1.341951 -1.06066 -1.06066
|
||||
C -1.341951 -0.77937 -1.5 -0.397805 -1.5 0
|
||||
C -1.5 0.397805 -1.341951 0.77937 -1.06066 1.06066
|
||||
C -0.77937 1.341951 -0.397805 1.5 0 1.5
|
||||
z
|
||||
" style="stroke: #ff7f0e"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p9cb1e34570)">
|
||||
<use xlink:href="#me63c40fafa" x="61.751527" y="158.397688" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="77.794819" y="166.734483" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="93.83811" y="168.747962" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="109.881402" y="183.250135" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="125.924693" y="183.428552" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="141.967985" y="183.440704" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="158.011276" y="183.460342" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="174.054568" y="183.458574" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="190.09786" y="183.45641" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="206.141151" y="183.483699" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="222.184443" y="183.465226" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="238.227734" y="183.48573" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="254.271026" y="183.462584" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="270.314318" y="183.463043" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="286.357609" y="183.481273" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="302.400901" y="183.477096" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="318.444192" y="183.467116" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="334.487484" y="183.485477" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="350.530775" y="183.443151" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="366.574067" y="183.459105" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="382.617359" y="183.49194" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="398.66065" y="183.525188" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="414.703942" y="183.536436" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_14">
|
||||
<path d="M 61.751527 184.35737
|
||||
L 77.794819 185.026206
|
||||
L 93.83811 183.628957
|
||||
L 109.881402 184.186922
|
||||
L 125.924693 182.174872
|
||||
L 141.967985 183.646168
|
||||
L 158.011276 183.39562
|
||||
L 174.054568 183.599987
|
||||
L 190.09786 183.826028
|
||||
L 206.141151 183.654339
|
||||
L 222.184443 184.070666
|
||||
L 238.227734 183.824104
|
||||
L 254.271026 183.32351
|
||||
L 270.314318 183.383518
|
||||
L 286.357609 183.424671
|
||||
L 302.400901 182.806391
|
||||
L 318.444192 183.029013
|
||||
L 334.487484 184.979385
|
||||
L 350.530775 183.735793
|
||||
L 366.574067 181.975157
|
||||
L 382.617359 184.62586
|
||||
L 398.66065 183.488105
|
||||
" clip-path="url(#p9cb1e34570)" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="m415848a3f7" d="M 0 1.5
|
||||
C 0.397805 1.5 0.77937 1.341951 1.06066 1.06066
|
||||
C 1.341951 0.77937 1.5 0.397805 1.5 0
|
||||
C 1.5 -0.397805 1.341951 -0.77937 1.06066 -1.06066
|
||||
C 0.77937 -1.341951 0.397805 -1.5 0 -1.5
|
||||
C -0.397805 -1.5 -0.77937 -1.341951 -1.06066 -1.06066
|
||||
C -1.341951 -0.77937 -1.5 -0.397805 -1.5 0
|
||||
C -1.5 0.397805 -1.341951 0.77937 -1.06066 1.06066
|
||||
C -0.77937 1.341951 -0.397805 1.5 0 1.5
|
||||
z
|
||||
" style="stroke: #2ca02c"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p9cb1e34570)">
|
||||
<use xlink:href="#m415848a3f7" x="61.751527" y="184.35737" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="77.794819" y="185.026206" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="93.83811" y="183.628957" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="109.881402" y="184.186922" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="125.924693" y="182.174872" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="141.967985" y="183.646168" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="158.011276" y="183.39562" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="174.054568" y="183.599987" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="190.09786" y="183.826028" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="206.141151" y="183.654339" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="222.184443" y="184.070666" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="238.227734" y="183.824104" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="254.271026" y="183.32351" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="270.314318" y="183.383518" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="286.357609" y="183.424671" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="302.400901" y="182.806391" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="318.444192" y="183.029013" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="334.487484" y="184.979385" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="350.530775" y="183.735793" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="366.574067" y="181.975157" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="382.617359" y="184.62586" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="398.66065" y="183.488105" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_3">
|
||||
<path d="M 44.103906 196.19145
|
||||
L 44.103906 50.157188
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_4">
|
||||
<path d="M 44.103906 196.19145
|
||||
L 432.351562 196.19145
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="text_14">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="238.227734" y="44.157188" transform="rotate(-0 238.227734 44.157188)">tensor: dimension / scale strata</text>
|
||||
</g>
|
||||
<g id="legend_1">
|
||||
<g id="line2d_15">
|
||||
<path d="M 196.577734 240.25808
|
||||
L 204.577734 240.25808
|
||||
L 212.577734 240.25808
|
||||
" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m077c66be99" x="204.577734" y="240.25808" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_15">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="218.977734" y="243.05808" transform="rotate(-0 218.977734 243.05808)">1D</text>
|
||||
</g>
|
||||
<g id="line2d_16">
|
||||
<path d="M 196.577734 252.258705
|
||||
L 204.577734 252.258705
|
||||
L 212.577734 252.258705
|
||||
" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#me63c40fafa" x="204.577734" y="252.258705" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_16">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="218.977734" y="255.058705" transform="rotate(-0 218.977734 255.058705)">2D</text>
|
||||
</g>
|
||||
<g id="line2d_17">
|
||||
<path d="M 246.227734 240.25808
|
||||
L 254.227734 240.25808
|
||||
L 262.227734 240.25808
|
||||
" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m415848a3f7" x="254.227734" y="240.25808" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_17">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="268.627734" y="243.05808" transform="rotate(-0 268.627734 243.05808)">3D</text>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="axes_2">
|
||||
<g id="patch_5">
|
||||
<path d="M 460.853906 196.19145
|
||||
L 849.101563 196.19145
|
||||
L 849.101563 50.157188
|
||||
L 460.853906 50.157188
|
||||
z
|
||||
" style="fill: #ffffff"/>
|
||||
</g>
|
||||
<g id="matplotlib.axis_3">
|
||||
<g id="xtick_6">
|
||||
<g id="line2d_18">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="478.501527" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_18">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="478.501527" y="210.789106" transform="rotate(-0 478.501527 210.789106)">0</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_7">
|
||||
<g id="line2d_19">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="555.230313" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_19">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="555.230313" y="210.789106" transform="rotate(-0 555.230313 210.789106)">5</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_8">
|
||||
<g id="line2d_20">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="631.959099" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_20">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="631.959099" y="210.789106" transform="rotate(-0 631.959099 210.789106)">10</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_9">
|
||||
<g id="line2d_21">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="708.687884" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_21">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="708.687884" y="210.789106" transform="rotate(-0 708.687884 210.789106)">15</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="xtick_10">
|
||||
<g id="line2d_22">
|
||||
<g>
|
||||
<use xlink:href="#m20431a40b1" x="785.41667" y="196.19145" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_22">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="785.41667" y="210.789106" transform="rotate(-0 785.41667 210.789106)">20</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_23">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="654.977734" y="224.789887" transform="rotate(-0 654.977734 224.789887)">Absolute layer distance</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="matplotlib.axis_4">
|
||||
<g id="ytick_7">
|
||||
<g id="line2d_23">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="183.492818" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_8">
|
||||
<g id="line2d_24">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="158.095555" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_9">
|
||||
<g id="line2d_25">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="132.698292" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_10">
|
||||
<g id="line2d_26">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="107.301029" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_11">
|
||||
<g id="line2d_27">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="81.903766" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="ytick_12">
|
||||
<g id="line2d_28">
|
||||
<g>
|
||||
<use xlink:href="#m65c4260567" x="460.853906" y="56.506503" style="stroke: #000000; stroke-width: 0.8"/>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_24">
|
||||
<text style="font-size: 10px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="450.951563" y="123.174319" transform="rotate(-90 450.951563 123.174319)">Median exact cosine</text>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_29">
|
||||
<path d="M 478.501527 182.503138
|
||||
L 493.847284 182.948842
|
||||
L 509.193041 183.220975
|
||||
L 524.538798 183.330264
|
||||
L 539.884556 183.291693
|
||||
L 555.230313 183.116647
|
||||
L 570.57607 183.232753
|
||||
L 585.921827 183.234258
|
||||
L 601.267584 183.058896
|
||||
L 616.613341 182.905816
|
||||
L 631.959099 182.927197
|
||||
L 647.304856 182.993101
|
||||
L 662.650613 183.079887
|
||||
L 677.99637 183.020027
|
||||
L 693.342127 183.195357
|
||||
L 708.687884 183.003567
|
||||
L 724.033642 183.057686
|
||||
L 739.379399 183.028052
|
||||
L 754.725156 183.045272
|
||||
L 770.070913 183.073977
|
||||
L 785.41667 183.196073
|
||||
L 800.762427 183.346757
|
||||
L 816.108185 183.004354
|
||||
L 831.453942 183.290907
|
||||
" clip-path="url(#p1d52707d08)" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#p1d52707d08)">
|
||||
<use xlink:href="#m077c66be99" x="478.501527" y="182.503138" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="493.847284" y="182.948842" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="509.193041" y="183.220975" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="524.538798" y="183.330264" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="539.884556" y="183.291693" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="555.230313" y="183.116647" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="570.57607" y="183.232753" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="585.921827" y="183.234258" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="601.267584" y="183.058896" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="616.613341" y="182.905816" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="631.959099" y="182.927197" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="647.304856" y="182.993101" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="662.650613" y="183.079887" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="677.99637" y="183.020027" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="693.342127" y="183.195357" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="708.687884" y="183.003567" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="724.033642" y="183.057686" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="739.379399" y="183.028052" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="754.725156" y="183.045272" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="770.070913" y="183.073977" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="785.41667" y="183.196073" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="800.762427" y="183.346757" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="816.108185" y="183.004354" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
<use xlink:href="#m077c66be99" x="831.453942" y="183.290907" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_30">
|
||||
<path d="M 478.501527 180.900864
|
||||
L 493.847284 182.42411
|
||||
L 509.193041 182.264123
|
||||
L 524.538798 182.292232
|
||||
L 539.884556 182.627569
|
||||
L 555.230313 182.066545
|
||||
L 570.57607 182.476159
|
||||
L 585.921827 181.736467
|
||||
L 601.267584 181.73033
|
||||
L 616.613341 181.62865
|
||||
L 631.959099 181.931
|
||||
L 647.304856 181.803345
|
||||
L 662.650613 182.166982
|
||||
L 677.99637 181.720597
|
||||
L 693.342127 181.907748
|
||||
L 708.687884 181.969106
|
||||
L 724.033642 181.963089
|
||||
L 739.379399 181.677496
|
||||
L 754.725156 181.962314
|
||||
L 770.070913 181.361509
|
||||
L 785.41667 182.199205
|
||||
L 800.762427 181.783371
|
||||
L 816.108185 181.192352
|
||||
L 831.453942 182.034471
|
||||
" clip-path="url(#p1d52707d08)" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#p1d52707d08)">
|
||||
<use xlink:href="#me63c40fafa" x="478.501527" y="180.900864" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="493.847284" y="182.42411" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="509.193041" y="182.264123" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="524.538798" y="182.292232" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="539.884556" y="182.627569" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="555.230313" y="182.066545" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="570.57607" y="182.476159" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="585.921827" y="181.736467" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="601.267584" y="181.73033" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="616.613341" y="181.62865" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="631.959099" y="181.931" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="647.304856" y="181.803345" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="662.650613" y="182.166982" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="677.99637" y="181.720597" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="693.342127" y="181.907748" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="708.687884" y="181.969106" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="724.033642" y="181.963089" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="739.379399" y="181.677496" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="754.725156" y="181.962314" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="770.070913" y="181.361509" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="785.41667" y="182.199205" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="800.762427" y="181.783371" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="816.108185" y="181.192352" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
<use xlink:href="#me63c40fafa" x="831.453942" y="182.034471" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_31">
|
||||
<path d="M 478.501527 56.760803
|
||||
L 493.847284 56.953783
|
||||
L 509.193041 57.215599
|
||||
L 524.538798 57.110413
|
||||
L 539.884556 56.832617
|
||||
L 555.230313 56.726551
|
||||
L 570.57607 56.99061
|
||||
L 585.921827 57.153358
|
||||
L 601.267584 56.958307
|
||||
L 616.613341 56.802667
|
||||
L 631.959099 56.988158
|
||||
L 647.304856 57.237716
|
||||
L 662.650613 56.851527
|
||||
L 677.99637 56.621023
|
||||
L 693.342127 56.598307
|
||||
L 708.687884 56.810297
|
||||
L 724.033642 57.288575
|
||||
L 739.379399 56.806439
|
||||
L 754.725156 57.750457
|
||||
L 785.41667 56.781365
|
||||
L 800.762427 57.726256
|
||||
" clip-path="url(#p1d52707d08)" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g clip-path="url(#p1d52707d08)">
|
||||
<use xlink:href="#m415848a3f7" x="478.501527" y="56.760803" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="493.847284" y="56.953783" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="509.193041" y="57.215599" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="524.538798" y="57.110413" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="539.884556" y="56.832617" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="555.230313" y="56.726551" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="570.57607" y="56.99061" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="585.921827" y="57.153358" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="601.267584" y="56.958307" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="616.613341" y="56.802667" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="631.959099" y="56.988158" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="647.304856" y="57.237716" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="662.650613" y="56.851527" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="677.99637" y="56.621023" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="693.342127" y="56.598307" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="708.687884" y="56.810297" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="724.033642" y="57.288575" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="739.379399" y="56.806439" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="754.725156" y="57.750457" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="785.41667" y="56.781365" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
<use xlink:href="#m415848a3f7" x="800.762427" y="57.726256" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="line2d_32">
|
||||
<path d="M 478.501527 71.634134
|
||||
L 493.847284 74.250255
|
||||
L 509.193041 74.782543
|
||||
L 524.538798 73.834506
|
||||
L 539.884556 76.943039
|
||||
L 555.230313 81.644826
|
||||
L 570.57607 73.654599
|
||||
L 585.921827 77.601619
|
||||
L 601.267584 79.228269
|
||||
L 616.613341 82.904383
|
||||
L 631.959099 72.852593
|
||||
L 647.304856 74.699809
|
||||
L 662.650613 71.894698
|
||||
L 677.99637 76.647967
|
||||
L 693.342127 75.773724
|
||||
L 708.687884 68.633756
|
||||
L 724.033642 76.367069
|
||||
L 739.379399 84.757892
|
||||
L 754.725156 87.785553
|
||||
L 770.070913 75.881164
|
||||
L 800.762427 73.725153
|
||||
" clip-path="url(#p1d52707d08)" style="fill: none; stroke: #d62728; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<defs>
|
||||
<path id="me5bc3986e1" d="M 0 1.5
|
||||
C 0.397805 1.5 0.77937 1.341951 1.06066 1.06066
|
||||
C 1.341951 0.77937 1.5 0.397805 1.5 0
|
||||
C 1.5 -0.397805 1.341951 -0.77937 1.06066 -1.06066
|
||||
C 0.77937 -1.341951 0.397805 -1.5 0 -1.5
|
||||
C -0.397805 -1.5 -0.77937 -1.341951 -1.06066 -1.06066
|
||||
C -1.341951 -0.77937 -1.5 -0.397805 -1.5 0
|
||||
C -1.5 0.397805 -1.341951 0.77937 -1.06066 1.06066
|
||||
C -0.77937 1.341951 -0.397805 1.5 0 1.5
|
||||
z
|
||||
" style="stroke: #d62728"/>
|
||||
</defs>
|
||||
<g clip-path="url(#p1d52707d08)">
|
||||
<use xlink:href="#me5bc3986e1" x="478.501527" y="71.634134" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="493.847284" y="74.250255" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="509.193041" y="74.782543" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="524.538798" y="73.834506" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="539.884556" y="76.943039" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="555.230313" y="81.644826" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="570.57607" y="73.654599" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="585.921827" y="77.601619" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="601.267584" y="79.228269" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="616.613341" y="82.904383" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="631.959099" y="72.852593" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="647.304856" y="74.699809" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="662.650613" y="71.894698" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="677.99637" y="76.647967" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="693.342127" y="75.773724" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="708.687884" y="68.633756" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="724.033642" y="76.367069" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="739.379399" y="84.757892" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="754.725156" y="87.785553" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="770.070913" y="75.881164" style="fill: #d62728; stroke: #d62728"/>
|
||||
<use xlink:href="#me5bc3986e1" x="800.762427" y="73.725153" style="fill: #d62728; stroke: #d62728"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="patch_6">
|
||||
<path d="M 460.853906 196.19145
|
||||
L 460.853906 50.157188
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="patch_7">
|
||||
<path d="M 460.853906 196.19145
|
||||
L 849.101563 196.19145
|
||||
" style="fill: none; stroke: #000000; stroke-width: 0.8; stroke-linejoin: miter; stroke-linecap: square"/>
|
||||
</g>
|
||||
<g id="text_25">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="654.977734" y="44.157188" transform="rotate(-0 654.977734 44.157188)">block: dimension / scale strata</text>
|
||||
</g>
|
||||
<g id="legend_2">
|
||||
<g id="line2d_33">
|
||||
<path d="M 582.918984 240.25808
|
||||
L 590.918984 240.25808
|
||||
L 598.918984 240.25808
|
||||
" style="fill: none; stroke: #1f77b4; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m077c66be99" x="590.918984" y="240.25808" style="fill: #1f77b4; stroke: #1f77b4"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_26">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="605.318984" y="243.05808" transform="rotate(-0 605.318984 243.05808)">2d-128</text>
|
||||
</g>
|
||||
<g id="line2d_34">
|
||||
<path d="M 582.918984 252.258705
|
||||
L 590.918984 252.258705
|
||||
L 598.918984 252.258705
|
||||
" style="fill: none; stroke: #ff7f0e; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#me63c40fafa" x="590.918984" y="252.258705" style="fill: #ff7f0e; stroke: #ff7f0e"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_27">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="605.318984" y="255.058705" transform="rotate(-0 605.318984 255.058705)">2d-64</text>
|
||||
</g>
|
||||
<g id="line2d_35">
|
||||
<path d="M 649.643984 240.25808
|
||||
L 657.643984 240.25808
|
||||
L 665.643984 240.25808
|
||||
" style="fill: none; stroke: #2ca02c; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#m415848a3f7" x="657.643984" y="240.25808" style="fill: #2ca02c; stroke: #2ca02c"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_28">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="672.043984" y="243.05808" transform="rotate(-0 672.043984 243.05808)">3d-axis0-slice</text>
|
||||
</g>
|
||||
<g id="line2d_36">
|
||||
<path d="M 649.643984 252.258705
|
||||
L 657.643984 252.258705
|
||||
L 665.643984 252.258705
|
||||
" style="fill: none; stroke: #d62728; stroke-width: 1.5; stroke-linecap: square"/>
|
||||
<g>
|
||||
<use xlink:href="#me5bc3986e1" x="657.643984" y="252.258705" style="fill: #d62728; stroke: #d62728"/>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_29">
|
||||
<text style="font-size: 8px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: start" x="672.043984" y="255.058705" transform="rotate(-0 672.043984 255.058705)">3d-volume</text>
|
||||
</g>
|
||||
</g>
|
||||
</g>
|
||||
<g id="text_30">
|
||||
<text style="font-size: 12px; font-family: 'DejaVu Sans', 'Bitstream Vera Sans', 'Computer Modern Sans Serif', 'Lucida Grande', 'Verdana', 'Geneva', 'Lucid', 'Arial', 'Helvetica', 'Avant Garde', sans-serif; text-anchor: middle" x="427.901563" y="16.318125" transform="rotate(-0 427.901563 16.318125)">Screened candidates: module mixture and selection bias still apply</text>
|
||||
</g>
|
||||
</g>
|
||||
<defs>
|
||||
<clipPath id="p9cb1e34570">
|
||||
<rect x="44.103906" y="50.157188" width="388.247656" height="146.034262"/>
|
||||
</clipPath>
|
||||
<clipPath id="p1d52707d08">
|
||||
<rect x="460.853906" y="50.157188" width="388.247656" height="146.034262"/>
|
||||
</clipPath>
|
||||
</defs>
|
||||
</svg>
|
||||
|
After Width: | Height: | Size: 43 KiB |
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 106 KiB |
File diff suppressed because one or more lines are too long
|
After Width: | Height: | Size: 74 KiB |
Binary file not shown.
Binary file not shown.
|
After Width: | Height: | Size: 286 KiB |
File diff suppressed because it is too large
Load Diff
|
After Width: | Height: | Size: 93 KiB |
@@ -0,0 +1,34 @@
|
||||
{
|
||||
"repo": "Qwen/Qwen3.5-2B",
|
||||
"revision": "15852e8c16360a2fea060d615a32b45270f8a8fc",
|
||||
"expected_bytes": 4548351670,
|
||||
"started_unix": 1794156349.8216345,
|
||||
"files": [
|
||||
{
|
||||
"name": "README.md",
|
||||
"bytes": 62814,
|
||||
"sha256": "c0e83a849c776e6fa843d011f023132d942a0f0140d903205bf1c363adad2275",
|
||||
"upstream_sha256": null
|
||||
},
|
||||
{
|
||||
"name": "config.json",
|
||||
"bytes": 2908,
|
||||
"sha256": "ed1c1723241f23f7f4e23430759cbd7dcfb4103cbdfe052bfe7626b57c2615b4",
|
||||
"upstream_sha256": null
|
||||
},
|
||||
{
|
||||
"name": "model.safetensors-00001-of-00001.safetensors",
|
||||
"bytes": 4548221488,
|
||||
"sha256": "aa33250c4fc64891ddfaba3a314fd9542ea371843c387178b425fbcc5ed680b1",
|
||||
"upstream_sha256": "aa33250c4fc64891ddfaba3a314fd9542ea371843c387178b425fbcc5ed680b1"
|
||||
},
|
||||
{
|
||||
"name": "model.safetensors.index.json",
|
||||
"bytes": 64460,
|
||||
"sha256": "aca8afed9da75b0f050b408d270766fd77627f1af401e240f61c3b47d0db02f9",
|
||||
"upstream_sha256": null
|
||||
}
|
||||
],
|
||||
"license": "apache-2.0",
|
||||
"completed_unix": 1794159787.5429547
|
||||
}
|
||||
@@ -0,0 +1,96 @@
|
||||
{
|
||||
"architectures": [
|
||||
"Qwen3_5ForConditionalGeneration"
|
||||
],
|
||||
"image_token_id": 248056,
|
||||
"model_type": "qwen3_5",
|
||||
"text_config": {
|
||||
"attention_bias": false,
|
||||
"attention_dropout": 0.0,
|
||||
"attn_output_gate": true,
|
||||
"dtype": "bfloat16",
|
||||
"eos_token_id": 248044,
|
||||
"full_attention_interval": 4,
|
||||
"head_dim": 256,
|
||||
"hidden_act": "silu",
|
||||
"hidden_size": 2048,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 6144,
|
||||
"layer_types": [
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"linear_attention",
|
||||
"full_attention"
|
||||
],
|
||||
"linear_conv_kernel_dim": 4,
|
||||
"linear_key_head_dim": 128,
|
||||
"linear_num_key_heads": 16,
|
||||
"linear_num_value_heads": 16,
|
||||
"linear_value_head_dim": 128,
|
||||
"max_position_embeddings": 262144,
|
||||
"mlp_only_layers": [],
|
||||
"model_type": "qwen3_5_text",
|
||||
"mtp_num_hidden_layers": 1,
|
||||
"mtp_use_dedicated_embeddings": false,
|
||||
"num_attention_heads": 8,
|
||||
"num_hidden_layers": 24,
|
||||
"num_key_value_heads": 2,
|
||||
"rms_norm_eps": 1e-06,
|
||||
"tie_word_embeddings": true,
|
||||
"use_cache": true,
|
||||
"vocab_size": 248320,
|
||||
"mamba_ssm_dtype": "float32",
|
||||
"rope_parameters": {
|
||||
"mrope_interleaved": true,
|
||||
"mrope_section": [
|
||||
11,
|
||||
11,
|
||||
10
|
||||
],
|
||||
"rope_type": "default",
|
||||
"rope_theta": 10000000,
|
||||
"partial_rotary_factor": 0.25
|
||||
}
|
||||
},
|
||||
"tie_word_embeddings": true,
|
||||
"transformers_version": "4.57.0.dev0",
|
||||
"video_token_id": 248057,
|
||||
"vision_config": {
|
||||
"deepstack_visual_indexes": [],
|
||||
"depth": 24,
|
||||
"hidden_act": "gelu_pytorch_tanh",
|
||||
"hidden_size": 1024,
|
||||
"in_channels": 3,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 4096,
|
||||
"model_type": "qwen3_5",
|
||||
"num_heads": 16,
|
||||
"num_position_embeddings": 2304,
|
||||
"out_hidden_size": 2048,
|
||||
"patch_size": 16,
|
||||
"spatial_merge_size": 2,
|
||||
"temporal_patch_size": 2
|
||||
},
|
||||
"vision_end_token_id": 248054,
|
||||
"vision_start_token_id": 248053
|
||||
}
|
||||
@@ -0,0 +1,19 @@
|
||||
{
|
||||
"passed": true,
|
||||
"checked_unix": 1794167980.8863053,
|
||||
"numerical_tables": {
|
||||
"tensor": {
|
||||
"rows": 2000,
|
||||
"finite_metrics": true,
|
||||
"coverage_matches": true
|
||||
},
|
||||
"block": {
|
||||
"rows": 14778,
|
||||
"finite_metrics": true,
|
||||
"coverage_matches": true
|
||||
}
|
||||
},
|
||||
"final_tests": 6,
|
||||
"final_test_seconds": 77.193,
|
||||
"earlier_parallel_test_timeout_seconds": 300
|
||||
}
|
||||
@@ -0,0 +1,5 @@
|
||||
procs -----------memory---------- ---swap-- -----io---- -system-- -------cpu------- -----timestamp-----
|
||||
r b swpd free buff cache si so bi bo in cs us sy id wa st gu CST
|
||||
0 18 2802460 145460 6576 1115112 143 162 2885 442 2946 19 7 4 86 2 0 0 2026-11-09 02:21:08
|
||||
0 18 2799748 101732 6080 1160132 4920 2256 104468 2468 6759 7057 10 15 0 75 0 0 2026-11-09 02:21:09
|
||||
0 22 2803528 89596 5284 1169020 1768 5308 114616 5364 6557 6835 3 12 0 85 0 0 2026-11-09 02:21:10
|
||||
File diff suppressed because one or more lines are too long
@@ -0,0 +1,6 @@
|
||||
{
|
||||
"phase": "archived_weights_deleted",
|
||||
"updated_unix": 1794167040.9455411,
|
||||
"supervisor_started_unix": 1794156953.8715413,
|
||||
"weights_present": false
|
||||
}
|
||||
@@ -0,0 +1,5 @@
|
||||
waiting_for_download
|
||||
analyzing
|
||||
plotting
|
||||
validating
|
||||
results_ready_weights_retained_until_archive_copy
|
||||
@@ -0,0 +1,17 @@
|
||||
test_3d_reader_and_exact_block (test_pilot.PilotTests.test_3d_reader_and_exact_block) ... ok
|
||||
test_bfloat16 (test_pilot.PilotTests.test_bfloat16) ... ok
|
||||
test_directional_replacement_error_and_scaling (test_pilot.PilotTests.test_directional_replacement_error_and_scaling) ... ok
|
||||
test_end_to_end_synthetic_groups_and_savings (test_pilot.PilotTests.test_end_to_end_synthetic_groups_and_savings) ... Inventory: 4 tensors, 33792 bytes
|
||||
tensor: sampling 4 objects
|
||||
block: sampling 22 objects
|
||||
block: verified 100/109
|
||||
{"model": "synthetic", "revision": "test", "tensor_count": 4, "parameter_count": 8448, "model_weight_bytes": 33792, "ndim_counts": {"2": 2, "3": 2}, "elapsed_seconds": 0.33063589688390493, "analysis_budget_seconds": 60, "peak_rss_kib": 50188, "python": "3.14.4", "numpy": "2.5.4", "seed": 20261109, "limitations": ["No inference or functional equivalence measured", "No exhaustive pair search", "Block search uses per-role/shape/scale reservoirs of 256", "Different block scales overlap; savings must not be added", "Block savings count only verified disjoint sampled blocks, no extrapolation", "Spectral probes use aligned bounded subvolumes, not whole tensors", "Shared encoding is theoretical, not a runnable compressed checkpoint", "No cross-role grouping, neuron permutation alignment, or training"], "completed": true}
|
||||
ok
|
||||
test_no_transitive_chaining (test_pilot.PilotTests.test_no_transitive_chaining) ... ok
|
||||
test_ranged_download_preserves_prefix (test_pilot.PilotTests.test_ranged_download_preserves_prefix) ... Weight download 36/36 bytes (100.0%)
|
||||
ok
|
||||
|
||||
----------------------------------------------------------------------
|
||||
Ran 6 tests in 77.193s
|
||||
|
||||
OK
|
||||
@@ -0,0 +1,8 @@
|
||||
test_3d_reader_and_exact_block (test_pilot.PilotTests.test_3d_reader_and_exact_block) ... ok
|
||||
test_bfloat16 (test_pilot.PilotTests.test_bfloat16) ... ok
|
||||
test_directional_replacement_error_and_scaling (test_pilot.PilotTests.test_directional_replacement_error_and_scaling) ... ok
|
||||
test_end_to_end_synthetic_groups_and_savings (test_pilot.PilotTests.test_end_to_end_synthetic_groups_and_savings) ... Inventory: 4 tensors, 33792 bytes
|
||||
tensor: sampling 4 objects
|
||||
block: sampling 22 objects
|
||||
block: verified 100/109
|
||||
{"model": "synthetic", "revision": "test", "tensor_count": 4, "parameter_count": 8448, "model_weight_bytes": 33792, "ndim_counts": {"2": 2, "3": 2}, "elapsed_seconds": 0.16227321792393923, "analysis_budget_seconds": 60, "peak_rss_kib": 49700, "python": "3.14.4", "numpy": "2.5.4", "seed": 20261109, "limitations": ["No inference or functional equivalence measured", "No exhaustive pair search", "Block search uses per-role/shape/scale reservoirs of 256", "Different block scales overlap; savings must not be added", "Block savings count only verified disjoint sampled blocks, no extrapolation", "Spectral probes use aligned bounded subvolumes, not whole tensors", "Shared encoding is theoretical, not a runnable compressed checkpoint", "No cross-role grouping, neuron permutation alignment, or training"], "completed": true}
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user