meta-llama / llama-recipes

Scripts for fine-tuning Meta Llama3 with composable FSDP & PEFT methods to cover single/multi-node GPUs. Supports default & custom datasets for applications such as summarization and Q&A. Supporting a number of candid inference solutions such as HF TGI, VLLM for local or cloud deployment. Demo apps to showcase Meta Llama3 for WhatsApp & Messenger.
10.4k stars 1.48k forks source link

Tensorboard/Wandb #95

Open zepmck opened 11 months ago

zepmck commented 11 months ago

🚀 The feature, motivation and pitch

Hi all!

Has anyone edited the code so to support logging on Tensorboard or WanDB?

Thanks!

Alternatives

No response

Additional context

No response

polym commented 9 months ago

I tried to add wandb to llama-recipes training, but I'm not sure whether it's useful?

Wandb web: https://wandb.ai/polym404/llama_recipes/workspace?workspace=user-polym404 PR: https://github.com/facebookresearch/llama-recipes/pull/200

@mreso could you give me some advice?

HamidShojanazeri commented 9 months ago

@polym its for sure useful in tracking your experiments, we are looking into adding some ML OP options, pls check this issue.

wukaixingxp commented 1 month ago

Hi! The wandb feature has been integrated. Please check here for more details