lplalbert commented on PR #412:
URL: https://github.com/apache/hugegraph-ai/pull/412#issuecomment-6899725357

   Confirmed unfixed at cc6d391 — pushed `f1b9aab` for it. I also owe a 
correction on my earlier note: reading an explicit request value as a per-call 
override of *both* chunk retrieval and keyword matching was wrong; the field is 
scoped to keyword matching.
   
   Root cause: `VectorQueryNode` forwarded `wk_input.vector_dis_threshold` into 
`VectorIndexQuery`, where the explicit value overrode `CHUNK_DIS_THRESHOLD` — 
exactly the `0.9`-filters-chunks behavior in 
[r4516553841](https://github.com/apache/hugegraph-ai/pull/412#discussion_r4516553841).
   
   Structural fix: the node no longer forwards the field; chunk retrieval 
always prunes at `CHUNK_DIS_THRESHOLD`, and the operator parameter is renamed 
`chunk_dis_threshold` so the coupling cannot return through an ambiguous name. 
No separate request field was added.
   
   Validation: red-green 
`test_node_init_keeps_request_threshold_out_of_chunk_retrieval` plus 
API-to-store chain tests (`test_rag_request_defaults.py`: request -> flow 
`prepare` -> node -> chunk store search at `CHUNK_DIS_THRESHOLD` for both 
omitted and explicit thresholds; the explicit value still reaches 
`SemanticIdQuery`). Full suite 1310 passed, 12 skipped, 2 known pre-existing 
failures (Windows mode-bit pair in `test_external_config`). Also on this head: 
`129cc48` preserves JSON-shaped chunk text through the Milvus round trip 
(r4578310176).


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to