server : fix crash when system prompt is bigger than batch size #5714
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
The system prompt is now decoded in batches. * server : fix off-by-one n_past when start of prompt matches whole cache The tokens right after the matching part would otherwise skip a pos value.
ggerganov
approved these changes
Feb 25, 2024
There was a problem hiding this comment.
Maybe llama_decode should eventually have some kind of built-in auto-batching since forgetting to split a batch seems common in the examples.
It should be added to (see below)common and eventually the new llamax library will abstract this
|
I plan to extend |
|
@slaren Hi, thanks for the fix, would it be possible to add a simple test ? |
|
After it is implemented sure, but it is not merged yet. It still needs more work, but it should be done soon. |
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Mar 13, 2024
…ganov#5714) The system prompt is now decoded in batches. * server : fix off-by-one n_past when start of prompt matches whole cache The tokens right after the matching part would otherwise skip a pos value.
hodlen
pushed a commit
to hodlen/llama.cpp
that referenced
this pull request
Apr 1, 2024
…ganov#5714) The system prompt is now decoded in batches. * server : fix off-by-one n_past when start of prompt matches whole cache The tokens right after the matching part would otherwise skip a pos value.
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Outdated suggestions cannot be applied.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.
The system prompt is now decoded in batches.
Maybe
llama_decodeshould eventually have some kind of built-in auto-batching since forgetting to split a batch seems common in the examples.I also fixed a problem where
n_pastwould skip aposvalue when the prefix of a prompt is fully matching the tokens in the cache.