Has anyone had this extension work properly with models converted to TensorRT format?

[Tliks]

Normally, a model converted to TensorRT format only reduces the generation time and does not change the generated image, but when using a converted model in this extension, the image seems to be different from the unconverted model and the generated image is not segmented correctly. Has anyone confirmed that this extension is working correctly in the generation using TensorRT? I used SDXL for testing