Together AI enables fine-tuning any Hugging Face Hub model

AI Impact Summary

Together AI is introducing the ability to fine-tune any model from the Hugging Face Hub using a base model template paired with a user’s Hugging Face model. The workflow supports private repos via HF tokens and an example shows base model templates like togethercomputer/llama-2-7b-chat and a custom HF model such as HuggingFaceTB/SmolLM2-1.7B-Instruct, enabling production-ready, domain-specific variants without bespoke training infra. This shift reduces DevOps overhead, accelerates experimentation across coding, multilingual, and domain-specific use cases, and allows trained artifacts to be pushed back to the HF Hub for broader sharing and reuse. Overall, teams can move from discovery to customized, deployment-ready models faster while maintaining control over training resources and governance.