Spaces:

Aravindhan11
/

Distributed-Transformer-Framework

Sleeping

Deploy Intelligent Distributed LLaMA Framework

52510e8 verified 4 months ago

270 Bytes

The examples can run on both Nvidia GPU (using NCCL backend) and CPU (using GLOO backend).

uv run torchrun --nproc_per_node=3 examples/row_parallel_linear_demo.py

uv run torchrun --nproc_per_node=4 examples/row_parallel_linear_demo.py