Thank you for your interest in our private beta!

As a beta participant, you will have early access to 4-bit quantization and paged optimizers code for efficient LLM finetuning. This code allows finetuning of the largest LLaMA models on a single GPU and is integrated with Huggingface transformers through bitsandbytes. We also support all other HuggingFace models. Please note that our code is still in development. We are looking for testers who are interested in experimenting with the code and reporting any bugs. We are particularly interested in help testing our examples and our own fine-tuning runs for new use-cases to debug edge-cases. Significant contributors will be acknowledged in the main repos and/or paper.

Your feedback and contributions will be invaluable to the release of the project!