Pin wandb version in llm-finetuning repo

modal-labs / llm-finetuning

Guide for fine-tuning Llama/Mistral/CodeLlama models and more

MIT License

538 stars 84 forks source link

Closed yirenlu92 closed 7 months ago

yirenlu92 commented 7 months ago

need to pin wandb version, otherwise get this error: https://github.com/wandb/wandb/issues/7406

mwaskom commented 7 months ago

Ah nice, this bit me when I updated the Axolotl container version and I (falsely) attributed it to that.

I don't immediately understand why CI is failing but it seems not great?

yirenlu92 commented 7 months ago

oof yeah that is a lot of failing tests

mwaskom commented 7 months ago

oh you deleted src/inference.py by accident :D

yirenlu92 commented 7 months ago

🤦‍♀️ sorry I had some local changes that I didn't want to include and forgot about that.

charlesfrye commented 7 months ago

pulled this change into #48