tensorrtllm_backend
tensorrtllm_backend copied to clipboard
InternVL deploy
I have completed the generation of the engine according to the tutorial, is there a relevant example tutorial for the end to end triton deploy. In addition, when making the visual encoder engine, I found that the pre-processing class of the image is pulled from the hf website initialization information, which seems not very friendly, why not directly use the local configuration file for initialization.