2-5X faster 80% less memory LLM finetuning

Repository for Meta Chameleon, a mixed-modal early-fusion foundation model from FAIR.

Utils for Unsloth

Open weights LLM from Google DeepMind.

Felafax is building AI infra for non-NVIDIA GPUs

Unsloth Studio

Efficient Triton Kernels for LLM Training

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

A collaborative project to collect datasets in SEA languages, SEA regions, or SEA cultures.

A framework for few-shot evaluation of language models.

Puzzles for learning Triton

Open source codebase powering the HuggingChat app

Efficient Streaming Language Models with Attention Sinks

security probing tool for LLMs

Experiments testing the algebraic value-editing conjecture (AVEC) on GPT-2 models

An experimental open-source attempt to make GPT-4 fully autonomous.

Toolkit for Indobenchmark