Configuration reference - Baseten

Model deployments

Set resources, dependencies, secrets, and runtime options in config.yaml. For setup instructions, see Configure a model.

Inference engines

Compare supported models and features in the inference engines overview.

Engine configuration	Section in `config.yaml`
BIS-LLM (Baseten Inference Stack v2)	`bis_llm`
Baseten Embeddings Inference (BEI)	`trt_llm`
Engine-Builder-LLM	`trt_llm`

Training jobs

Use the Training SDK to define compute, runtime, storage, and checkpoint settings in a Python file such as config.py. The SDK ships in the truss Python package. For a first training job, follow the Training quickstart.

⌘I