Logprobs Refractor #331
Merged
Logprobs Refractor #331
Conversation
This file contains 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
comaniac
approved these changes
Mar 27, 2024
This was referenced May 12, 2024
merrymercy
reviewed
May 15, 2024
| if extend_seq_lens_cpu[i] == 0: | ||
| continue | ||
| k = input_metadata.top_logprobs_nums[i] | ||
| t = all_logprobs[pt : pt + extend_seq_lens_cpu[i]].topk(k) |
| extend_seq_lens_cpu = input_metadata.extend_seq_lens | ||
| for i in range(len(input_metadata.extend_seq_lens)): | ||
| if extend_seq_lens_cpu[i] == 0: | ||
| continue |
There was a problem hiding this comment.
continue will result in a out-of-bound error later.
| normalized_prompt_logprobs, | ||
| prefill_token_logprobs, | ||
| decode_token_logprobs, | ||
| ) = self.backend.select(self, expr.choices, expr.temperature) |
There was a problem hiding this comment.
did not modify the openai backend.
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Outdated suggestions cannot be applied.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.
What does this PR do?
*_token_logprobsstands for each token' logprobs,*_prompt_logprobstands for the sum of logprobs of a segment of prompt.*_token_logprobsintoprefill_token_logprobsanddecode_token_logprobs.List[(logprob, token_id, token_text)]logprob_start_lenand respond inmeta_data["prefill_token_logprobs"]andmeta_data["decode_token_logprobs"]respectively.sgl.selectresponse which is the only use case for multiple requests in a single HTTP request.top_logprobssupport.