Execution framework for multi-task model parallelism. Enables the training of arbitrarily large models with a single GPU, with linear speedups for multi-gpu multi-task execution.
Geek Repo:Geek Repo
Github PK Tool:Github PK Tool