Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[usability] support qwen2.5 and deepseek #931

Merged
merged 2 commits into from
Feb 4, 2025
Merged

Conversation

wheresmyhair
Copy link
Collaborator

@wheresmyhair wheresmyhair commented Feb 3, 2025

  1. Support jinja template (requires transformers>=4.43.0), see here
  2. Support new models through jinja template (conversation dataset with tools, tool_calls):
  1. Support opt-in test
  2. Hymba template are now more consistant with the one provided in their tokenizer_config.json

Tests

image

@wheresmyhair wheresmyhair mentioned this pull request Feb 3, 2025
33 tasks
Copy link
Contributor

@research4pan research4pan left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The commit looks good to me in most parts. Some minor concerns,

src/lmflow/utils/conversation_template/deepseek.py

  • [Feature] line 7-8: may also retain the name "deepseek" for backward compatibility purposes.

src/lmflow/utils/conversation_template/hymba.py

  • [Question] line 40-70, 75-89: Why are they removed?

@wheresmyhair
Copy link
Collaborator Author

wheresmyhair commented Feb 4, 2025

src/lmflow/utils/conversation_template/deepseek.py

  • [Feature] line 7-8: may also retain the name "deepseek" for backward compatibility purposes.

🆗, and change JINJA_TEMPLATE in alphabetical order.

src/lmflow/utils/conversation_template/hymba.py

  • [Question] line 40-70, 75-89: Why are they removed?

Several changes have been made in ConversationTemplateForTool such that we don't need to overwrite the ConversationTemplateForTool.encode_conversation for hymba anymore. After this change, hymba template are now more consistant with the one provided in their official tokenizer_config.json

@wheresmyhair wheresmyhair merged commit 65b9125 into main Feb 4, 2025
2 checks passed
@wheresmyhair wheresmyhair deleted the yizhenjia-modelsupport branch February 10, 2025 16:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants