Repository navigation
修复说话人采样配置被 top2 加额覆盖 - #262
Merged
Merged
Conversation
Agent-Executor: codex-lead Agent-Session: 01a1481d-411a-78b2-bce7-3089f4f4ac06
Agent-Executor: codex-lead Agent-Session: 01a1481d-411a-78b2-bce7-3089f4f4ac06
zj1123581321
marked this pull request as ready for review
October 17, 2026 04:56
Required Gate v2 — 状态面板当前状态:pass · 可合并 当前裁决: PR 累计主审轮(影子): Disposition receipts consumed:
Gate 历史(v1;来源为持久化
|
| Run | Attempt | Head | 状态 | PR 累计主审轮(影子) | 收件人动作 |
|---|---|---|---|---|---|
| 39003361874 | 1 | e681476 |
skipped |
— |
无需动作(主审未跑,绿≠过审) |
| 39004390336 | 1 | dbf79e6 |
fail |
— |
要修代码 |
| 39004431028 | 1 | dbf79e6 |
fail |
— |
要修代码 |
| 39004431028 | 2 | dbf79e6 |
pass |
1/5 · collecting |
可合并 |
历史行按 run_id + run_attempt 去重并只增不删;删除本评论后可由 gate_terminal 制品重建。
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
说话人采样配置现在作为普通说话人的基准配额。发言量 top2 的条数和字符预算分别取配置值与 8 条/1200 字符下限中的较大值,例如配置 12/2000 时仍保留 12 条与 2000 字符上限;配置 12/400 时 top2 使用 12 条/1200 字符,其他人使用 12 条/400 字符。单条 120 字符限制和全局预算保持不变。
任一采样配置设为 0 时,只停用本轮新采样与新的说话人 LLM 调用;不会清除与输入指纹匹配的已验证历史姓名,identity_fallback 姓名恢复及分层缓存行为保持不变。top2 排名移到说话人循环外计算一次,并保留并列时的原顺序。
验证:
uv run pytest -q tests/unit/test_speaker_inferencer.py:46 passed。make test:通过。max临时注入为min后,较大配置回归按预期失败;还原后通过。OCR 维护意见(接受不修):
infer()/_identity_fallback()docstring 未列零配额来源;只影响说明完整性。_extract_sample_dialogs()被直接以零配额调用时仍可能采样 top2;唯一生产调用方infer()已早退,没有生产缺陷。findings=[]记录保持不变。Refs #244(子项 2、7)