Key Features
Everything you need to fine-tune
Facet AI handles the boilerplate, so you can focus on creating powerful, specialized models.
Dataset Preprocessing
Process vision and text data from custom uploads and Hugging Face.
Optimized Training
Utilize PEFT, LoRA, QLoRA, and full tuning with 4-bit/8-bit quantization.
Flexible Model Export
Export to adapters, merged, quantized, GGUF, and deploy to the Hub or GCS.
Use Cases
Tailored for Your Needs
Fine tune your model for a variety of tasks, without writing a single line of code.
Instruction Tuning (SFT)
Adapt models to follow specific instructions and formats for precise outputs.
Reasoning (GRPO)
Enhance reasoning capabilities for complex, multi-step tasks and problem-solving.
Preference Tuning (DPO)
Align models with human preferences for better performance, safety, and user experience.