Gemini: raise a NON-retryable error on a deterministically answer-less generation. - #760
Closed
copybara-service[bot] wants to merge 1 commit into
Closed
Gemini: raise a NON-retryable error on a deterministically answer-less generation.#760copybara-service[bot] wants to merge 1 commit into
copybara-service[bot] wants to merge 1 commit into
Conversation
…s generation. When a Gemini candidate returns no answer content because thinking consumed the decode budget (finishReason=MAX_TOKENS) or because the model stopped having emitted nothing (finishReason=STOP with candidatesTokenCount=0), the result is a function of the request, not of server health: re-sending the identical prompt reproduces it. Today that empty text reaches LanguageModel.sample(), which raises the retryable lf.EmptyGenerationError and re-sends the full prompt up to max_attempts times before failing with an opaque error. Gemini.result() now detects exactly those two signatures and raises the new GeminiEmptyGenerationError, an lf.LMError (deliberately not an lf.EmptyGenerationError / lf.RetryableLMError), so it escapes the retry wrappers on the first attempt carrying the model, finish reason, token counts, and the knobs to change. All other answer-less cases keep their existing behavior. PiperOrigin-RevId: 969538244
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Gemini: raise a NON-retryable error on a deterministically answer-less generation.
When a Gemini candidate returns no answer content because thinking consumed the decode budget (finishReason=MAX_TOKENS) or because the model stopped having emitted nothing (finishReason=STOP with candidatesTokenCount=0), the result is a function of the request, not of server health: re-sending the identical prompt reproduces it. Today that empty text reaches LanguageModel.sample(), which raises the retryable lf.EmptyGenerationError and re-sends the full prompt up to max_attempts times before failing with an opaque error.
Gemini.result() now detects exactly those two signatures and raises the new GeminiEmptyGenerationError, an lf.LMError (deliberately not an lf.EmptyGenerationError / lf.RetryableLMError), so it escapes the retry wrappers on the first attempt carrying the model, finish reason, token counts, and the knobs to change. All other answer-less cases keep their existing behavior.