Skip to content

Latest commit

 

History

History
36 lines (24 loc) · 782 Bytes

Installation.md

File metadata and controls

36 lines (24 loc) · 782 Bytes

Installation

Claw-R1 relies on verl for the training backend. Follow the steps below to set up your environment.

Prerequisites

  • Conda (recommended for environment management)
  • Python 3.10+
  • CUDA (for GPU training)

Setup

1. Create a Conda Environment

conda create -n clawr1 python=3.10 -y
conda activate clawr1

2. Clone and Install veRL

Install the nightly version of veRL from source (recommended):

git clone https://github.com/volcengine/verl && cd verl
pip install --no-deps -e .
cd ..

Verify Installation

After installation, ensure your environment is ready:

python -c "import verl; print('veRL installed successfully')"