feat: add frequency penalty to inference sampling pipeline

- Add FrequencyPenaltyStrategy (logit -= penalty * count) - Per-task rep_window for penalty history lookup - Wire through engine, task, executor, API layer - Add --frequency_penalty and --rep_window to stream_chat.py - 9 unit tests for frequency penalty strategy
2026-07-17 21:28:31 +08:00
parent a1ea26d367
commit d08a92c7bd
9 changed files with 370 additions and 20 deletions
@@ -21,7 +21,6 @@ logger = logging.getLogger(__name__)
 _UNSUPPORTED_PARAMS = (
    "n",
    "presence_penalty",
-    "frequency_penalty",
    "logit_bias",
    "user",
 )