single mention · inference & serving
TensorRT-LLM all-reduce backend
Selects TensorRT-LLM as the backend for all-reduce communication.
The sources on this page name it without explaining how it works. The line above is the extraction's one-line gloss. See serving parallelism for the methods filed beside it.
Also called trtllm all-reduce backend.
Filed under inference & serving › serving parallelism