Add independent thinking-mode controls

This commit is contained in:
2026-09-23 00:36:45 +08:00
parent b645190ea5
commit 0f0594bc41
5 changed files with 18 additions and 2 deletions
+2 -1
View File
@@ -52,6 +52,7 @@ python3 -m logicbench run tasks/short-smoke.json
| `RENDERER_API_STYLE`, `SOLVER_API_STYLE` | `responses` (default) or `chat_completions` |
| `RENDERER_MODEL`, `SOLVER_MODEL` | Required, independently selected model IDs |
| `RENDERER_TIMEOUT_SECONDS`, `SOLVER_TIMEOUT_SECONDS` | Optional per-request timeout; default `240` |
| `RENDERER_ENABLE_THINKING`, `SOLVER_ENABLE_THINKING` | Optional `true`/`false` provider extension; omitted unless set |
Outputs under the task's `output_dir` are:
@@ -70,7 +71,7 @@ Configure repository Actions with:
2. Variables `RENDERER_BASE_URL`, `RENDERER_MODEL`, and `RENDERER_API_STYLE`.
3. Variables `SOLVER_BASE_URL`, `SOLVER_MODEL`, and `SOLVER_API_STYLE`.
For Qwen's OpenAI-compatible API, use the compatible-mode `/v1` endpoint and `chat_completions`. Keys, URLs, and models remain role-specific even when both roles currently use the same provider.
For Qwen's OpenAI-compatible API, use the compatible-mode `/v1` endpoint and `chat_completions`. Keys, URLs, and models remain role-specific even when both roles currently use the same provider. The checked-in short workflow explicitly disables solver thinking to avoid long non-streaming requests; remove or change that workflow value when thinking-mode performance is itself under test.
## Tests