cs-552-2026-nlpowerpuffs/math_model
The Qwen3-1.7B model, developed by Qwen, is a 1.7 billion parameter causal language model with a 32,768 token context length. It uniquely supports seamless switching between a 'thinking mode' for complex logical reasoning, math, and coding, and a 'non-thinking mode' for general-purpose dialogue. This model excels in reasoning capabilities, agentic tasks, and multilingual instruction following across over 100 languages.
Loading preview...
Qwen3-1.7B Model Overview
Qwen3-1.7B is a 1.7 billion parameter causal language model from the Qwen series, featuring a substantial 32,768 token context length. A key differentiator is its dynamic thinking mode capability, allowing the model to switch between a dedicated reasoning mode for complex tasks like mathematics, coding, and logical problem-solving, and an efficient non-thinking mode for general conversational interactions. This dual-mode functionality aims to optimize performance across diverse scenarios.
Key Capabilities
- Enhanced Reasoning: Demonstrates significant improvements in mathematical problem-solving, code generation, and commonsense logical reasoning, outperforming previous Qwen models in their respective modes.
- Agentic Expertise: Achieves leading performance among open-source models in complex agent-based tasks, supporting precise integration with external tools in both thinking and non-thinking modes.
- Multilingual Support: Offers strong capabilities across over 100 languages and dialects for instruction following and translation.
- Human Preference Alignment: Excels in creative writing, role-playing, multi-turn dialogues, and general instruction following, providing a more natural conversational experience.
Usage Recommendations
For optimal performance, specific sampling parameters are recommended for each mode: Temperature=0.6, TopP=0.95, TopK=20 for thinking mode, and Temperature=0.7, TopP=0.8, TopK=20 for non-thinking mode. The model also supports dynamic switching between these modes via user input tags (/think and /no_think) in multi-turn conversations. Developers can leverage Qwen-Agent for streamlined agentic applications.