Prevent special token injection from user-provided text - #770
Open
niraj-mx07 wants to merge 1 commit into
Open
Conversation
…not accidentally be interpreted as chat structure
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Fixes the prompt-structure injection issue described in #768 by escaping characters that can introduce Gemma's special control sequences when processing user-provided text.
Changes
escape()method to the tokenizer to escape characters used in special token sequences.Motivation
User-controlled text containing sequences such as
<|turn>could previously be interpreted as special Gemma control tokens rather than ordinary text. This could allow untrusted input to alter the intended conversation structure.This change ensures that special sequences originating from user-provided text are treated as text instead of being interpreted as control tokens.
Fixes #768