We are building a python library of procedural dataset generators and algorithmically verifiable reasoning environments for training reasoning models with reinforcement learning (RL).
The goal is to generate virtually infinite data with adjustable complexity.
Algorithmic verification allows to train on tasks like Rubik‘s cube or Countdown which have many correct solutions.
In GALLERY.md you find example outputs of all datasets available in reasoning-gym.
The reasoning-gym package requires Python >= 3.11.
Install via pip:
pip install reasoning-gym
For development setup see CONTRIBUTING.md.
Example:
import reasoning_gym
data = reasoning_gym.create_dataset('leg_counting', size=10, seed=42)
for i, x in enumerate(data):
print(f'{i}: q="{x['question']}", a="{x['answer']}"')
print('metadata:', x['metadata'])
# use the dataset's `score_answer` method for algorithmic verification
assert data.score_answer(answer=x['answer'], entry=x) == 1.0Output:
0: q="How many legs are there in total if you have 1 sea slug, 1 deer?", a="4"
metadata: {'animals': {'sea slug': 1, 'deer': 1}, 'total_legs': 4}
1: q="How many legs are there in total if you have 2 sheeps, 2 dogs?", a="16"
metadata: {'animals': {'sheep': 2, 'dog': 2}, 'total_legs': 16}
2: q="How many legs are there in total if you have 1 crab, 2 lobsters, 1 human, 1 cow, 1 bee?", a="42"
...
Please see CONTRIBUTING.md.
If you have ideas for dataset generators please create an issue here or contact us in the #reasoning-gym channel of the GPU-Mode discord server.