[BFCL] Update Llama Prompting Handler for Llama 3.0 vs. 3.1 Chat Template Differences #752
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Outdated suggestions cannot be applied.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.
Llama 3.1 models in prompting mode have a slightly different chat template than Llama 3.0. Specifically, they need to carry the
Cutting Knowledge Date
and theToday Date
information in the system prompt. This PR accounts for that difference in the Llama prompting handler (llama.py
).Here is the difference in how we should format the system prompt:
Note: Although Llama 3.2 has the same chat template as Llama 3.1, their model card doesn't contain such additional information, so we adhere to that, and thus Llama 3.2 still have the same processing logic with Llama 3.0.
For reference, below is the Llama 3.0 chat template.
And this is the Llama 3.1 chat template.