Tensor_parallel
Visit Tooltensor_parallel automatically splits PyTorch models across multiple GPUs for training and inference, enabling large models to run with potentially linear speedup. It reduces memory load by only loading parts of the model at once.
At a glance
Trending
Also listed in