🐛 [Bug] Encountered bug when using Torch-TensorRT

[QiusongYang]

Bug Description

torch.max to tensorrt operator failure

To Reproduce

Steps to reproduce the behavior:

1. batch_max_used_token_hidden_states.max(dim=1)[0]

Expected behavior

Environment

Build information about Torch-TensorRT can be found by turning on debug messages

• Torch-TensorRT Version (e.g. 1.0.0): 1.1.0
• PyTorch Version (e.g. 1.0): 1.12.0
• CPU Architecture:
• OS (e.g., Linux): Linux
• How you installed PyTorch (conda, pip, libtorch, source): pytorch official docker 22.05
• Build command you used (if compiling from source):
• Are you using local sources or building from archives:
• Python version:
• CUDA version: 11.3
• GPU models and configuration: tsal p40
• Any other relevant information:

Additional context

[narendasan]

Please provide a complete reproducing script

[QiusongYang]

batch_max_token_hidden_states = batch_max_used_token_hidden_states.max(dim=1).values

[ncomly-nvidia]

Hi @QiusongYang can you please provide your python script where you are seeing this failure? Without knowing the specific failure we won't be able to help debug with you.

[QiusongYang]

Hi @QiusongYang can you please provide your python script where you are seeing this failure? Without knowing the specific failure we won't be able to help debug with you.

actually, you can write any network and torch::max function, could you convert to tensorrt model?

[github-actions[bot]]

This issue has not seen activity for 90 days, Remove stale label or comment or this will be closed in 10 days