| Process Reward Model trained by OpenRLHF | |
| - Dataset: Math-Shepherd (https://huggingface.co/datasets/peiyi9979/Math-Shepherd) | |
| - Learning Rate: 1e-6 | |
| - Training Accuracy: 0.922 |
| Process Reward Model trained by OpenRLHF | |
| - Dataset: Math-Shepherd (https://huggingface.co/datasets/peiyi9979/Math-Shepherd) | |
| - Learning Rate: 1e-6 | |
| - Training Accuracy: 0.922 |