Optimize KV cache size #1
Merged
Conversation
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
- Create `reload_from_cache_state` method - Still using LLamaState as container - Use low level `ctx.get_logits_ith` to get last calculated logits. - Add StateReloadError so that can be fallible. - Change Llama class to use this instead of `load_state` directly. - Default implementation still uses `load_state`.
Catch StateReloadError and add logging if runs into this when running.
- Check when saving that model doesn't need logits - Ad note in `reload_from_cache` state to revisit
- Make default to *not* save logits - Error if needed and save_logits False in build_cache - Handle reloading with/without scores if needed + available
- Add more tests - Make llama_state / small_model module scope (so don't need to reload for each test) - Setting env var in `.env` file
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Outdated suggestions cannot be applied.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.
Allow for saving / reloading KV cache without saving the logits, since not needed for my use case + sampling strategy.