Generalizing the chat_template prompt strategy #1660
Merged
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 strategy now supports configuring several fields: * The data field holding message arrays * the role and content fields for each message * role mapping from source to target types additionally this adds a sample llama3-8b instruct template using the chat template
winglian
approved these changes
May 28, 2024
Comment on lines
+9
to
+16
| chat_template: llama3 | ||
| datasets: | ||
| - path: fozziethebeat/alpaca_messages_2k_test | ||
| type: chat_template | ||
| chat_template: llama3 | ||
| field_messages: messages | ||
| message_field_role: role | ||
| message_field_content: content |
There was a problem hiding this comment.
Do we really need to assign the chat_template twice in dataset and outside? I'm testing this PR. Any difference between the two "chat_template" settings?
I feel that passing type: chat_template and the related field keys is already specifying how to load the data. The value of the chat_template should be the template used for tokenization for training.
There was a problem hiding this comment.
we already have some handling when using this with sharegpt and chatml, so I've updated that to handle it automatically for the general case here: #1664
|
Great! Thanks for merging! |
5 tasks
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Outdated suggestions cannot be applied.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.
Description
The strategy now supports configuring several fields:
additionally this adds a sample llama3-8b instruct template using the chat template
Fixes #1654
Motivation and Context
#1654
How has this been tested?
Tested via
Further it was tested by running
And manually inspecting the emitted sample tokens
Screenshots (if appropriate)
Types of changes
Social Handles (Optional)
fozziethebeat