Metadata-Version: 2.1
Name: atorch
Version: 0.1.7
Summary: A pytorch extension for efficient deep learning.
Home-page: https://github.com/intelligent-machine-learning/dlrover/tree/master/atorch
Author: Ant Group
Requires-Python: >=3.8
Requires-Dist: fairscale (==0.4.1)
Requires-Dist: apex
Requires-Dist: flash-attn
Requires-Dist: deepspeed (==0.10.0)
Requires-Dist: transformers (==4.31.0)
Requires-Dist: networkx
Requires-Dist: pyomo
Requires-Dist: pynvml (==11.4.1)
Requires-Dist: grpcio (==1.34.1)
Requires-Dist: grpcio-tools (==1.34.1)
Requires-Dist: fsspec (==2023.10.0)
Requires-Dist: pyarrow (==12.0.0)
Requires-Dist: pandas (==2.0.1)
Requires-Dist: tensorboard (==2.11.0)
Requires-Dist: dlrover[torch]
Requires-Dist: protobuf (==3.20.3)
Requires-Dist: safetensors
Requires-Dist: GPy
Requires-Dist: pymoo (==0.5.0)

ATorch supports efficient and easy-to-use model training experience. ATorch provides performance optimizations in aspects such as I/O, preprocessing, computation, and communication (including automatic optimization), and has supported large-scale pretraining and finetuning of LLMs with over 100 billion parameters and thousands of advanced GPUs.
