diff --git a/models/qwen3/14b/qwen3_14b_prefill_tq_draft.py b/models/qwen3/14b/qwen3_14b_prefill_tq_draft.py index e892b81a..155bebe9 100644 --- a/models/qwen3/14b/qwen3_14b_prefill_tq_draft.py +++ b/models/qwen3/14b/qwen3_14b_prefill_tq_draft.py @@ -74,7 +74,7 @@ # --------------------------------------------------------------------------- # Prefill-specific tiling constants (local, may differ from config.py) # --------------------------------------------------------------------------- -MAX_SEQ = 128 +MAX_SEQ = M.max_seq # rope_cos/sin ctx cap; was 128 (draft placeholder) -> long prompts >128 read OOB rope -> garbage BATCH_TILE = 16 K_CHUNK = 128 Q_OUT_CHUNK = 64