188 lines
3.8 KiB
JSON
188 lines
3.8 KiB
JSON
{
|
|
"layers": [
|
|
3,
|
|
7,
|
|
11,
|
|
15,
|
|
19,
|
|
23
|
|
],
|
|
"layer_pairs": [
|
|
[
|
|
3,
|
|
7
|
|
],
|
|
[
|
|
3,
|
|
11
|
|
],
|
|
[
|
|
3,
|
|
15
|
|
],
|
|
[
|
|
3,
|
|
19
|
|
],
|
|
[
|
|
3,
|
|
23
|
|
],
|
|
[
|
|
7,
|
|
11
|
|
],
|
|
[
|
|
7,
|
|
15
|
|
],
|
|
[
|
|
7,
|
|
19
|
|
],
|
|
[
|
|
7,
|
|
23
|
|
],
|
|
[
|
|
11,
|
|
15
|
|
],
|
|
[
|
|
11,
|
|
19
|
|
],
|
|
[
|
|
11,
|
|
23
|
|
],
|
|
[
|
|
15,
|
|
19
|
|
],
|
|
[
|
|
15,
|
|
23
|
|
],
|
|
[
|
|
19,
|
|
23
|
|
]
|
|
],
|
|
"heads": 8,
|
|
"kv_heads": 2,
|
|
"head_dim": 256,
|
|
"calibration_samples": [
|
|
"cal-zh-1",
|
|
"cal-zh-2",
|
|
"cal-en-1",
|
|
"cal-en-2"
|
|
],
|
|
"validation_samples": [
|
|
"test-zh-1",
|
|
"test-zh-2",
|
|
"test-en-1",
|
|
"test-en-2"
|
|
],
|
|
"fresh_test_file": "crossqkv/fresh_samples.json",
|
|
"fresh_test_count": 4,
|
|
"tokens_per_text": 32,
|
|
"query_positions_for_primary_metrics": [
|
|
7,
|
|
8,
|
|
9,
|
|
10,
|
|
11,
|
|
12,
|
|
13,
|
|
14,
|
|
15,
|
|
16,
|
|
17,
|
|
18,
|
|
19,
|
|
20,
|
|
21,
|
|
22,
|
|
23,
|
|
24,
|
|
25,
|
|
26,
|
|
27,
|
|
28,
|
|
29,
|
|
30,
|
|
31
|
|
],
|
|
"candidate_types": [
|
|
"full_attention",
|
|
"qkv_head",
|
|
"qk_map",
|
|
"qk_head",
|
|
"v_group"
|
|
],
|
|
"discrete_head_matching": "Exact 8-head minimum mean-JS bijection, two source-layer calibration input distributions; frozen before validation; P injection only",
|
|
"continuous_alignment": false,
|
|
"training": false,
|
|
"scale_fitting": false,
|
|
"gpu": false,
|
|
"cost_required": false,
|
|
"tiers": {
|
|
"strict": {
|
|
"output_error": 0.05,
|
|
"output_cosine": 0.99,
|
|
"js": 0.01,
|
|
"js_p95": 0.05,
|
|
"top5": 0.8,
|
|
"value_error": 0.05,
|
|
"value_cosine": 0.99
|
|
},
|
|
"approximate": {
|
|
"output_error": 0.2,
|
|
"output_cosine": 0.98,
|
|
"js": 0.05,
|
|
"js_p95": 0.15,
|
|
"top5": 0.6,
|
|
"value_error": 0.2,
|
|
"value_cosine": 0.98
|
|
}
|
|
},
|
|
"max_qk_head_output_candidates": 32,
|
|
"max_v_group_output_candidates": 16,
|
|
"all_qk_map_output_candidates": 30,
|
|
"max_final_candidate": 1,
|
|
"selection": "strict before approximate; type priority full_attention/qkv_head/qk_map/qk_head/v_group; then worst output error, JS, ID; no validation-driven fallback",
|
|
"head_output_activity_floor": "max(1e-6, 0.1 * within-layer median head RMS) for every calibration/validation/test text",
|
|
"fresh_test_run_only_if_frozen_candidate_passes_validation": true,
|
|
"network_interventions": "two same-input donor-function substitution directions; plus one earlier-to-later actual cached-state reuse test reported separately",
|
|
"network_quality": {
|
|
"mean_token_nll_delta_max": 0.01,
|
|
"per_text_p95_nll_delta_max": 0.05
|
|
},
|
|
"retention": "all weights and caches retained until explicit user deletion request",
|
|
"seconds": 7140,
|
|
"controls": [
|
|
"identity",
|
|
"uniform causal P",
|
|
"seeded within-causal-row key shuffle",
|
|
"first-key ablation",
|
|
"self/recent3 mass",
|
|
"value Gram to input",
|
|
"value Gram of two seeded Gaussian projections"
|
|
],
|
|
"positional_repeat_labels": {
|
|
"sink_dominated": 0.9,
|
|
"self_dominated": 0.9,
|
|
"recent3_dominated": 0.9
|
|
},
|
|
"limitations": [
|
|
"Old validation texts were used in prior experiments; four fresh texts are reserved for this new goal",
|
|
"Direct V-coordinate disagreement does not disprove V/O functional equivalence; centered value Gram CKA is diagnostic only",
|
|
"Arbitrary P-head assignment does not authorize arbitrary GQA KV-weight permutation",
|
|
"Short text-only pilot does not establish broad task equivalence or linear-attention reuse"
|
|
],
|
|
"registered_unix": 1794208453.0746515,
|
|
"source_run": "/opt/exp-model-parameter-similarity/runs/cpu-qwen35-2b",
|
|
"fresh_sample_sha256": "d3f53fc98fa9d061a3df42f24fe307b693b41a9ddf74c8377dd52096663c40a9"
|
|
}
|