Mkulakow/legacy pipeline finish reason#4169
Open
michalkulakowski wants to merge 1 commit intomainfrom
Open
Conversation
Contributor
There was a problem hiding this comment.
Pull request overview
Updates OVMS legacy LLM pipeline response handling to propagate the actual GenAI finish_reason (instead of always reporting stop), and bumps OpenVINO/GenAI dependency pins to a newer nightly.
Changes:
- Propagate legacy pipeline
finish_reasoninto the final streamed chunk and into OpenAI Chat/Completions unary responses. - Enable finish-reason checking for the legacy LLM flow test case.
- Update OpenVINO / OpenVINO GenAI source pins and nightly package versions (including export-model requirements).
Reviewed changes
Copilot reviewed 5 out of 5 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
versions.mk |
Updates OV/GenAI commit pins and GenAI nightly package URLs to 2026.2.0.0.dev20260428. |
src/test/llm/llmnode_test.cpp |
Enables finish-reason validation for lm_legacy_regular in parameterized tests. |
src/llm/language_model/legacy/servable.cpp |
Uses results.finish_reasons[0] (when available) for the final streaming chunk finish reason. |
src/llm/apis/openai_completions.cpp |
Switches unary finish_reason serialization to use results.finish_reasons[...] for Encoded/VLM results. |
demos/common/export_models/requirements.txt |
Bumps openvino / openvino-tokenizers nightly versions to 2026.2.0.dev20260428. |
Comment on lines
409
to
418
| auto finishReason = mapFinishReason(results.finish_reasons[index], !parsedOutput.toolCalls.empty()); | ||
| jsonResponse.FinishReason(finishReason.value_or("unknown")); | ||
| // index: integer; Choice index, only n=1 supported anyway | ||
| jsonResponse.Index(index++); |
| jsonResponse.StartObject(); | ||
| // finish_reason: "stop" in regular scenario, "tool_calls" if output contains tool calls | ||
| auto finishReason = mapFinishReason(ov::genai::GenerationFinishReason::STOP, !parsedOutput.toolCalls.empty()); | ||
| auto finishReason = mapFinishReason(results.finish_reasons[index], !parsedOutput.toolCalls.empty()); |
michalkulakowski
force-pushed
the
mkulakow/legacy_pipeline_finish_reason
branch
from
4a740a6 to
31f3036
Compare
michalkulakowski
force-pushed
the
mkulakow/legacy_pipeline_finish_reason
branch
from
2d62b16 to
fbd018c
Compare
dkalinowski
reviewed
May 8, 2026
| jsonResponse.StartObject(); | ||
| // finish_reason: "stop" in regular scenario, "tool_calls" if output contains tool calls | ||
| auto finishReason = mapFinishReason(ov::genai::GenerationFinishReason::STOP, !parsedOutput.toolCalls.empty()); | ||
| const ov::genai::GenerationFinishReason finishReasonRaw = |
Collaborator
There was a problem hiding this comment.
is it possible to have results.finish_reasons.empty() ? in which situations?
dkalinowski
reviewed
May 8, 2026
| // finish_reason: "stop" in regular scenario, "tool_calls" if output contains tool calls | ||
| auto finishReason = mapFinishReason(ov::genai::GenerationFinishReason::STOP, !parsedOutput.toolCalls.empty()); | ||
| const ov::genai::GenerationFinishReason finishReasonRaw = | ||
| (!results.finish_reasons.empty()) ? results.finish_reasons[0] : ov::genai::GenerationFinishReason::STOP; |
Collaborator
There was a problem hiding this comment.
please add comment that we access [0] because we always use generate with batch=1
dkalinowski
reviewed
May 8, 2026
| parsedOutputs.push_back(parseOutputIfNeeded(tokens)); | ||
| } | ||
| return serializeUnaryResponseImpl(parsedOutputs); | ||
| for (const auto& finishReason : results.finish_reasons) { |
Collaborator
There was a problem hiding this comment.
why do we have different implementation than chat/completions? cant we just take [0]? I think we also have batch size=1 here always
dkalinowski
reviewed
May 8, 2026
| std::string serializedChunk = executionContext->apiHandler->serializeStreamingChunk(lastTextChunk, ov::genai::GenerationFinishReason::STOP); | ||
| ov::genai::GenerationFinishReason finishReason = ov::genai::GenerationFinishReason::STOP; | ||
| if (!legacyExecutionContext->results.finish_reasons.empty()) { | ||
| finishReason = legacyExecutionContext->results.finish_reasons[0]; |
Collaborator
There was a problem hiding this comment.
isnt 0 finish reasons internal error here? should we throw if it happens?
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
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.
🛠 Summary
JIRA/Issue if applicable.
Describe the changes.
🧪 Checklist
``