https://ericfillion.github.io/erictransformer/

Local pre-training, fine-tuning and inference for LLMs.

• Format your text data in JSONL and then use a few lines of code to train models.
• Full-parameter training of GPT-OSS-20b on a single H200.
• Use Apple's new MLX-LM framework for fast inference. Run GPT-OSS-120b locally.
• Enable RAG powered by Eric Search.
• Local experiment tracking that displays charts and metrics.

pip install erictransformer

Documentation

• Eric Fillion Lead Maintainer
• Ted Brownlow Maintainer

We are currently not accepting contributions.