How to convert a AMP trained model to get best performance and speed?

According to the doc: https://docs.pytorch.org/TensorRT/user_guide/mixed_precision.html We can convert model with this project where the param precision are explicitly said in the code. But when I train a model with torch AMP GradScaler where no value precision tagged in model code, Can we use this method to get a conerted chackpoint with best performance and inference speedup?

In fect, we had tried the torch pt->onnx-> tensorrt fp16 pipeline to convert pytorch AMP trained checkpoint into trt model format, but the inference results are noisey. while pt->onnx-> tensorrt fp32 pipeline will get a trt fp32 model the inference slower then what we need.

Start with the linked mixed-precision documentation and compare its explicitly tagged-precision workflow with the reported torch AMP GradScaler checkpoint. Reproduce the pt→ONNX→TensorRT FP16 and FP32 pipelines, then measure inference speed and output accuracy against the source model; done means a documented conversion path that avoids the reported noisy results while meeting the required speed.