The latest series of Code-Specific Qwen models, with significant improvements in code generation, code reasoning, and code fixing.
ollama run qwen2.5-coderQwen 2.5 Coder series of models are now updated in 6 sizes: 0.5B, 1.5B, 3B, 7B, 14B and 32B.
There are significant improvements in code generation, code reasoning and code fixing. The 32B model has competitive performance with OpenAI’s GPT-4o.
- 32B:
ollama run qwen2.5-coder:32b - 14B:
ollama run qwen2.5-coder:14b - 7B:
ollama run qwen2.5-coder:7b - 3B:
ollama run qwen2.5-coder:3b - 1.5B:
ollama run qwen2.5-coder:1.5b - 0.5B:
ollama run qwen2.5-coder:0.5b
Qwen2.5 Coder 32B Instruct: Advancing Code Generation, Repair, and Reasoning
Code Generation:
Qwen2.5 Coder 32B Instruct, the flagship model of this open-source release, sets a new standard in code generation. It outperforms other open-source models across popular benchmarks, including EvalPlus, LiveCodeBench, and BigCodeBench, while demonstrating performance competitive with GPT-4o.
Code Repair:
Efficiently fixing code errors is a crucial skill for programmers, and Qwen2.5 Coder 32B Instruct excels in this area. On the widely recognized Aider benchmark for code repair, the model achieved an impressive score of 73.7, delivering results on par with GPT-4o. This capability makes it an indispensable tool for streamlining development workflows.
Code Reasoning:
The ability to understand code execution and predict inputs and outputs is vital for advanced programming tasks. Qwen2.5 Coder 32B Instruct builds on the success of its 7B counterpart, offering enhanced code reasoning capabilities that push the boundaries of what AI can achieve in this domain.
With its state-of-the-art performance in generation, repair, and reasoning, Qwen2.5 Coder 32B Instruct is a powerful tool for developers seeking cutting-edge solutions.