Dongwei/DeepSeek-R1-Distill-Qwen-7B-GRPO_Math
TEXT GENERATIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.6BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Feb 3, 2025Architecture:Transformer Featherless Exclusive Cold
Dongwei/DeepSeek-R1-Distill-Qwen-7B-GRPO_Math is a 7.6 billion parameter language model, fine-tuned from deepseek-ai/DeepSeek-R1-Distill-Qwen-7B, with a 32768 token context length. It was trained using the GRPO method on the MATH-lighteval dataset, specializing in mathematical reasoning. This model is designed for tasks requiring advanced mathematical problem-solving capabilities.
Loading preview...
Overview
This model, DeepSeek-R1-Distill-Qwen-7B-GRPO_Math, is a specialized language model with 7.6 billion parameters and a 32768 token context length. It is a fine-tuned version of the deepseek-ai/DeepSeek-R1-Distill-Qwen-7B base model, developed by Dongwei.
Key Capabilities
- Enhanced Mathematical Reasoning: The model has been specifically fine-tuned on the DigitalLearningGmbH/MATH-lighteval dataset.
- GRPO Training Method: It leverages the GRPO (Gradient-based Reward Policy Optimization) method, as introduced in the paper "DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models" (arXiv:2402.03300), to improve its mathematical problem-solving abilities.
- TRL Framework: Training was conducted using the TRL (Transformer Reinforcement Learning) framework.
Good for
- Mathematical Problem Solving: Ideal for applications requiring robust mathematical reasoning and accurate solutions to complex math problems.
- Research in Mathematical LLMs: Useful for researchers exploring advanced training techniques like GRPO for improving mathematical capabilities in language models.