mondk/Safetensors.Claude-Qwen3.5-4B-Reasoning
The mondk/Safetensors.Claude-Qwen3.5-4B-Reasoning model is a 4.5 billion parameter language model fine-tuned for enhanced reasoning and coding capabilities. Developed by mondk, it leverages a Qwen3.5 base and incorporates datasets derived from Claude models, specifically targeting improved performance in complex problem-solving and code generation. This model demonstrates a notable uplift in coding accuracy, making it suitable for applications requiring precise technical understanding and output.
Loading preview...
Model Overview
The mondk/Safetensors.Claude-Qwen3.5-4B-Reasoning is a 4.5 billion parameter model built upon the Qwen3.5 architecture, specifically fine-tuned by mondk to excel in reasoning and coding tasks. It integrates diverse datasets, including those derived from Claude models, to enhance its analytical and problem-solving prowess.
Key Capabilities & Performance
This model shows significant improvements in coding performance, with a reported +23% better coding capability compared to its base. Evaluation on a custom question set (generated by Claude) rated by Gemini demonstrates:
- Easy problems (E1-E5): 100% pass rate (5/5)
- Medium problems (M1-M8): Improved from 6/8 to 8/8 pass rate
- Hard problems (H1-H7): Improved from 4/7 to 5/7 pass rate
Overall, the fine-tuned model achieved an 18/20 total pass rate on the evaluation set, up from 15/20 for the base model.
Good For
- Coding assistance: Generating and debugging code, especially for precise problem-solving.
- Reasoning tasks: Applications requiring logical deduction and analytical thinking.
- Technical problem-solving: Scenarios where accurate and structured outputs are critical.
This model is designed to provide precise coding assistance and will not identify itself as Claude.