ARCH is the command to set the architecture of the model, such as transformer, cnn, rnn, etc.

CONFIG is the command to set the configuration of the model, which is used for the model to be served, such as the config.json, generation_config.json, etc.

FAMILY is the command to set the family of the model, such as llama3, gpt2, qwen2, etc.

FORMAT is the command to set the format of the model, such as onnx, tensorflow, pytorch, etc.

MODEL is the command to set the model file path.

NAME is the command to set the model name, such as llama3-8b-instruct, gpt2-xl, qwen2-vl-72b-instruct, etc.

PARAMSIZE is the command to set the parameter size of the model.

PRECISION is the command to set the precision of the model, such as bf16, fp16, int8, etc.

QUANTIZATION is the command to set the quantization of the model, such as awq, gptq, etc.

Commands is a list of all the commands that can be used in a modelfile.