%global _empty_manifest_terminate_build 0
Name:		torchrl
Version:	0.3.1
Release:	1
Summary:	TorchRL is a library of reusable components for deep learning with reinforcement learning
License:	MIT
URL:		https://torchrl.org/
Source0:	https://github.com/pytorch/rl/archive/refs/tags/v%{version}.tar.gz/rl-%{version}.tar.gz

BuildRequires:  g++
Requires:	python3-torch
Requires:	python3-gym

%description
TorchRL provides efficient, reusable components for Reinforcement Learning research with PyTorch.
Key features include:
- Data structures for storing and manipulating reinforcement learning environments
- Efficient operations on these structures (sampling, loss functions)
- A framework for policy gradient methods
- A framework for Q-learning methods.

%package -n python3-torchrl
Summary:	TorchRL is a library of reusable components for deep learning with reinforcement learning
Provides:	python-torchrl

BuildRequires:	python3-devel
BuildRequires:	python3-setuptools
BuildRequires:	python3-setuptools_scm
BuildRequires:	python3-pbr
BuildRequires:	python3-pip
BuildRequires:	python3-wheel
BuildRequires:	python3-pytorch
BuildRequires:	ninja-build

%description -n python3-torchrl
TorchRL provides efficient, reusable components for Reinforcement Learning research with PyTorch.
Key features include:
- Data structures for storing and manipulating reinforcement learning environments
- Efficient operations on these structures (sampling, loss functions)
- A framework for policy gradient methods
- A framework for Q-learning methods.

%prep
%autosetup -p1 -n rl-%{version}

%build
%py3_build

%install
%py3_install

%files -n python3-torchrl
%doc *.md
%license LICENSE
%{python3_sitearch}/*

%changelog
* Mon Apr 15 2024 weilaijishu 
- Initial package