here4dadata
here4dadata
+1 for even more visibility
+1
Set triton_backend to 'tensorrtllm' in the config.pbtxt for tensorrt_llm and it should work. I think this was introduced because there is now a model.py file in `tensorrt_llm/1` as of v0.10.0,...
Thanks for that clarification @byshiue.
Re-opening based on @VsonicV 's post