ErtasAI/DeepSeek-R1-Distill-Qwen-7B
DeepSeek-R1-Distill-Qwen-7B is a 7.6 billion parameter language model developed by DeepSeek AI, distilled from the larger DeepSeek-R1 reasoning model and based on Qwen2.5-Math-7B. It is fine-tuned using reasoning data generated by DeepSeek-R1, enabling it to achieve strong performance on complex reasoning, math, and code tasks. This model is designed to bring advanced reasoning capabilities to a smaller, more efficient architecture, making it suitable for applications requiring robust problem-solving in a compact form factor.
Loading preview...
DeepSeek-R1-Distill-Qwen-7B: Distilled Reasoning Power
DeepSeek-R1-Distill-Qwen-7B is a 7.6 billion parameter model from DeepSeek AI, part of their DeepSeek-R1 series. It is a distilled version of the larger DeepSeek-R1 reasoning model, built upon the Qwen2.5-Math-7B base model. This distillation process leverages reasoning patterns discovered by the 671 billion parameter DeepSeek-R1, which was developed using large-scale reinforcement learning (RL) without initial supervised fine-tuning (SFT).
Key Capabilities & Features
- Reasoning Distillation: Inherits advanced reasoning patterns from the powerful DeepSeek-R1, demonstrating that complex reasoning can be effectively transferred to smaller models.
- Strong Performance: Achieves competitive results across various benchmarks, particularly in mathematical and coding tasks, as shown in evaluations against models like GPT-4o and Claude-3.5-Sonnet.
- Optimized for Math & Code: Benchmarks indicate strong performance in areas such as AIME 2024 (55.5% pass@1), MATH-500 (92.8% pass@1), and LiveCodeBench (37.6% pass@1).
- Efficient Architecture: Provides robust reasoning capabilities within a 7.6B parameter count, making it more efficient than larger counterparts while retaining significant problem-solving prowess.
- Flexible Deployment: Can be run locally using frameworks like vLLM or SGLang, with specific recommendations for optimal performance, including temperature settings and prompt structuring.
Usage Recommendations
To achieve optimal performance with DeepSeek-R1-Distill-Qwen-7B, DeepSeek AI recommends:
- Setting the temperature between 0.5-0.7 (0.6 recommended).
- Avoiding system prompts; all instructions should be in the user prompt.
- For mathematical problems, include a directive like: "Please reason step by step, and put your final answer within \boxed{}".
- Enforcing the model to start its response with "\n" to ensure thorough reasoning.