Can any one share the script what params should be passed to run_dpo.py

huggingface / alignment-handbook

Robust recipes to align language models with human and AI preferences

https://huggingface.co/HuggingFaceH4

Apache License 2.0

4.18k stars 354 forks source link

Can any one share the script what params should be passed to run_dpo.py #141

Open Oscarjia opened 3 months ago

Oscarjia commented 3 months ago

Hello, I'm interested in utilizing run_dpo, but I'm unsure about the required parameters. Could someone provide me with some guidance on which parameters need to be passed?

alvarobartt commented 2 months ago

Hi here @Oscarjia indeed to run the alignment handbook scripts you only need to provide the configuration file and that's it!

See some sample configurations at https://github.com/huggingface/alignment-handbook/tree/main/recipes