[None][feat] Layer-wise benchmarks: make model init more general and support weights loading#10562

yuantailing merged 10 commits intoNVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

yuantailing:layer_wise_benchmarksyuantailing/TensorRT-LLM:layer_wise_benchmarksCopy head branch name to clipboard