[DRAFT] refactor: PyExecutor uses a list-type for response handling #5406
Conversation
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
PyExecutor._enqueue_responses now accepts a list of LlmResponse instead of a dictionary mapping req_id to response. Since each response already contains its request_id, using req_id as a dictionary key is redundant and causes issues when num_return_sequences > 1, where multiple responses share the same request_id. This change aligns PyExecutor with C++ Executor's behavior, which returns std::vector<Response>. The previous dictionary-based approach would overwrite responses with the same request_id, losing all but the last response. With this fix, all responses for multi-sequence generation are properly preserved. Changes: - _enqueue_responses now accepts List[LlmResponse] instead of Dict[int, LlmResponse] - All callers updated: _handle_errors, _handle_cancelled_requests, _handle_first_token_response, _handle_responses Signed-off-by: Jaedeok Kim <[email protected]>
|
This is a draft PR for a discussion purpose. Please DO NOT MERGE yet. |
jaedeok-nvidia
changed the title
|
/bot run |
|
PR_Github #9611 [ run ] triggered by Bot |
|
PR_Github #9611 [ run ] completed with state |
|
Closing the PR because it's no longer needed for #5415. If needed, we will revisit later. |
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Outdated suggestions cannot be applied.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.
PyExecutor._enqueue_responses now accepts a list of LlmResponse instead of a dictionary mapping req_id to response. Since each response already contains its request_id, using req_id as a dictionary key is redundant and causes issues when num_return_sequences > 1, where multiple responses share the same request_id.
This change aligns PyExecutor with C++ Executor's behavior, which returns std::vector. The previous dictionary-based approach would overwrite responses with the same request_id, losing all but the last response. With this fix, all responses for multi-sequence generation are properly preserved.
Changes: