Reward Model Card for sarm

A Success-Aware Reward Model (SARM) predicts a dense reward signal from observations, typically used downstream for reinforcement learning or human-in-the-loop fine-tuning when task success is not directly observable.

This reward model has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs.

---

How to Get Started with the Reward Model

Train from scratch

lerobot-train \
  --dataset.repo_id=${HF_USER}/<dataset> \
  --reward_model.type=sarm \
  --output_dir=outputs/train/<desired_reward_model_repo_id> \
  --job_name=lerobot_reward_training \
  --reward_model.device=cuda \
  --reward_model.repo_id=${HF_USER}/<desired_reward_model_repo_id> \
  --wandb.enable=true

Writes checkpoints to outputs/train/<desired_reward_model_repo_id>/checkpoints/.

Load the reward model in Python

from lerobot.rewards import make_reward_model

reward_model = make_reward_model(pretrained_path="<hf_user>/<reward_model_repo_id>")
reward = reward_model.compute_reward(batch)

---

Model Details

• License: apache-2.0