RLHFlow/Llama3.1-8B-PRM-Mistral-Data

RLHFlow/Llama3.1-8B-PRM-Mistral-Data is an 8 billion parameter process-supervised reward model (PRM) developed by RLHFlow, fine-tuned from Meta's Llama-3.1-8B-Instruct. It is trained on Mistral-generated data with a 32768 token context length, specifically optimized for evaluating and improving mathematical reasoning. This model excels at providing process-level feedback for complex problem-solving, demonstrating strong performance on benchmarks like GSM8K and MATH.