fix: Change the translational layer from numpy to torch during conversion to handle additional data types #3445

Job	Run time
generate-matrix / generate	7s
substitute-runner	1s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / wheel-py3_11-cuda11_8	10m 2s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / wheel-py3_11-cuda12_6	10m 0s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / wheel-py3_11-cuda12_8	10m 8s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / upload / wheel-py3_11-cuda11_8	11s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / upload / wheel-py3_11-cuda12_6	13s
Build torch-tensorrt whl package (pytorch/tensorrt, packaging/pre_build_script_windows.sh, packag... / upload / wheel-py3_11-cuda12_8	19s
Test dynamo converters [Python]	1s
Test dynamo export serde [Python]	1s
Test dynamo core [Python]	1s
Test core [Python]	1s
Test dynamo cudagraphs [Python]	1s
Test dynamo frontend [Python]	1s
Test torchscript frontend [Python]	1s
Test torch compile backend [Python]	1s
	31m 9s

Provide feedback