TorchCompileModel - ComfyUI Built-in Node Documentation - ComfyUI

The TorchCompileModel node applies PyTorch compilation to a model to optimize its performance. It creates a copy of the input model and wraps it with PyTorch’s compilation functionality using the specified backend. This can improve the model’s execution speed during inference.

Inputs

Parameter	Description	Data Type	Required	Range
`model`	The model to be compiled and optimized	MODEL	Yes	-
`backend`	The PyTorch compilation backend to use for optimization (default: “inductor”)	STRING	Yes	”inductor" "cudagraphs”

Outputs

Output Name	Description	Data Type
`model`	The compiled model with PyTorch compilation applied	MODEL

This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 240c6eb84ce2b93fe4d7c04fcf69df571e662f17893a4fd8e721241ea082edc8