fix(rag): lift chat output cap to 32768 tokens — long answers no longer cut off

2026-08-22 11:30:19 -04:00
parent 6ec6181c7b
commit 0da5275eeb
9 changed files with 295 additions and 5 deletions
@@ -51,6 +51,10 @@ class Settings(BaseSettings):
    # AND no candidate chunk matches the question lexically (see A8).
    relevance_threshold: float = 0.62
    max_context_chars: int = 24_000
+    #: Maximum output tokens a chat answer may use (owner instruction
+    #: 2026-08-22: answers must run to their natural end — the old hard
+    #: 700-token cap cut long answers off mid-sentence).
+    max_output_tokens: int = 32_768
    chunk_target_chars: int = 2_000
    chunk_overlap_chars: int = 200
    embed_batch_size: int = 16