open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-19 12:58:07 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions
reasoning-gym/examples/veRL
History
Andreas Koepf 8e25a4585f add grpo launch script
2025-02-17 20:52:03 +00:00
..
config add grpo launch script 2025-02-17 20:52:03 +00:00
.gitignore first bits of veRL example 2025-02-01 21:20:36 +00:00
launch_on_4gpu.sh add grpo launch script 2025-02-17 20:52:03 +00:00
main_ppo_custom_reward.py reduce veRL example size 2025-02-01 23:56:11 +00:00
README.md update config to latest veRL version 2025-02-17 18:43:51 +00:00
train_grpo.sh add grpo launch script 2025-02-17 20:52:03 +00:00
train_ppo.sh add grpo launch script 2025-02-17 20:52:03 +00:00

README.md

env setup

conda create --name verl python=3.11 -y
conda activate verl

pip install flash-attn --no-build-isolation
pip install ray wandb
# pip3 install vllm==0.7.0
pip3 install vllm --pre --extra-index-url https://wheels.vllm.ai/nightly

Regarding vllm>0.7 see: docs

clone and install veRL

tested with verl HEAD 0dfcb7f99e299940e1792a386df13c7591df351a

git clone https://github.com/volcengine/verl.git
cd verl
pip install -e .

Optionally log in to huggingface hub and wandb with your keys:

huggingface-cli login
wandb login