Metadata-Version: 2.1
Name: atorch
Version: 0.1.4
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 ==2023.1.0
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.
