PyPI page
Summary:
The OpenPipe Agent Reinforcement Training (ART) library
Latest version:
0.5.20
Required dependencies:
aiohttp
|
anthropic
|
litellm
|
nest-asyncio
|
numpy
|
openai
|
polars
|
pydantic
|
requests
|
scipy
|
setproctitle
|
tblib
|
typer
|
typing-extensions
|
weave
Optional dependencies:
accelerate
|
aiohttp
|
awscli
|
bitsandbytes
|
datrie
|
duckdb
|
fastapi
|
gql
|
hf-xet
|
huggingface-hub
|
langchain-core
|
langchain-openai
|
langgraph
|
matplotlib
|
msgspec
|
nbclient
|
nbmake
|
nixl-cu12
|
nixl-cu13
|
numpy
|
nvidia-cudnn-frontend
|
nvidia-nccl-cu13
|
nvidia-resiliency-ext
|
peft
|
pillow
|
protobuf
|
pyarrow
|
pydantic
|
pytest
|
seaborn
|
setuptools
|
tinker
|
tinker-cookbook
|
torch
|
torchao
|
torchmonarch
|
transformers
|
trl
|
unsloth
|
unsloth-zoo
|
uv
|
uvicorn
|
wandb
Downloads last day:
1,357
Downloads last week:
4,842
Downloads last month:
8,470
7-day smoothing