Fine-tune Any LLM from the Hugging Face Hub with Together AI

Summary

Together AI and Hugging Face have introduced an integration enabling developers to fine-tune compatible Hugging Face Hub models directly on Together AI's managed infrastructure. To launch a fine-tuning job via the Python SDK, users provide a base model from Together's catalog as a configuration template alongside the target Hugging Face repository identifier. This base model template dictates GPU allocation, memory configuration, training pipelines, and inference setup for custom models with matching architectures and sizes. The workflow operates bidirectionally, pulling from public or token-authenticated private repositories and optionally pushing completed checkpoints back to the Hub upon completion. Geared toward CausalLM models under 100 billion parameters, the capability enables faster iteration cycles and domain adaptation without custom DevOps infrastructure.

Takeaways

1. 1

   Fine-tuning requires specifying both a base model template from Together AI for GPU configuration and a custom Hugging Face model sharing a similar architecture.

2. 2

   The integration supports bidirectional Hub access, enabling downloads from private repositories using tokens and automatic output uploads to Hugging Face.

3. 3

   The fine-tuning infrastructure is designed to support all CausalLM model architectures under 100 billion parameters.