Server: preserve raw prompts in /v1/completions - #1012
Open
kyuz0 wants to merge 1 commit into
Open
Conversation
Remove the injected system/user conversation from string prompts, preserve exact token-ID prompts, and keep compatibility thinking controls inert. Retain model stops without applying chat-only thinking-tag stops. Validate raw live and disk cache histories before reuse, preserving model start tokens and sampled tokenization boundaries. Add model-free prompt/cache/stop regressions and a model-backed tokenization check.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
/v1/completionscurrently wraps the supplied prompt in a system/user conversation and adds an assistant prefix. Continue the raw prefix instead, with only the model's start sequence added to string prompts.Validation: model-free DeepSeek/GLM prompt, token-ID, stop and cache regression fixtures; CPU server, agent and session tests; AddressSanitizer/UndefinedBehaviorSanitizer checks. The initial prompt/cache fixture fails against unchanged main. Model-backed inference has not been rerun for this server-only change.