You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is there a plan to support PEFT methods like LoRA training in maxtext to support larger model fine-tuning / continue pretraining so that bigger models like LLaMA-3-70B can be trainined even with small amount of TPU/GPUs?
The text was updated successfully, but these errors were encountered:
Is there a plan to support PEFT methods like LoRA training in maxtext to support larger model fine-tuning / continue pretraining so that bigger models like LLaMA-3-70B can be trainined even with small amount of TPU/GPUs?
The text was updated successfully, but these errors were encountered: