Release Notes - ChatLLaMA 0.0.4

Major release of ChatLLaMA adding support to efficient training using LoRA.

New Features

• HF-based models can now be trained on both actor and RLHF using LoRA. This will help users in reducing the memory needed for training.