A streamlined, customizable, no-code web platform to manage fine-tuning workflows of DeepMind's Gemma 3 family of multimodal models. From dataset formatting to deployment, Facet AI lets you focus on what matters for your use case, not writing boilerplate code and scrolling through pages of documentations and cookbook.
Whether you're a complete beginner to LLM fine tuning, or experienced engineer looking to build a platform for your team, Facet AI is here to democratize SLM fine-tuning for everyone.
- Create your account and start now with our web platform
- Read our beginner-friendly docs and watch our video tutorials
Watch our demo video!! 👇👇
- For personal users, access our service through our no-code intuitive web interface to start training models in minutes
- For production and enterprise, easily deploy the entire platform on your cloud project with our Terraform IaC
- Powered by Google Cloud Platform (GCP), 100% open source and free to use!
- Supports local + huggingface datasets, all Gemma 3 & 3n models (from 270M to 27B), SFT + RL (GRPO, DPO, ORPO) post-training, evaluation + benchmarks, export to deploy on Ollama / llama.cpp / vLLM / SGLang and more
- Production-ready, robust, and centralized for your team to collaborate, unlike your pile of Colab notebooks
- Highly integrated with HuggingFace ecosystem --
hub, datasets, transformers, trackio, evaluate, lighteval, trl, peft, bitsandbytes
At Facet AI, we're on a journey to democratize small language model (SLM) fine-tuning. SLMs are geting more and more powerful, with new use caes emerging day by day. We hope that as more people start training SLM for their own task and domain specific adaptations, advancements in AI can more rapidly translate to real-world impact across industries and communities.
This project was built by Jet Chiang and Adarsh Dubey, two students developers from Canada and India, during the Google Summer of Code program in summer 2025. We would like to thank our organisation, Google DeepMind, our mentor, and the team at Google Cloud and Gemma for their continuous support throughout the term.
We also acknoweldge the team at transformers, TRL, Unsloth, vLLM, and much more open-source libraries that power our project. Facet AI would not exist without these amazing open-source AI communities.
We welcome contributions!! Refer to CONTRIBUTING.md in each sub-directory that you wish to contribute to.