Overview

This model is the instruction-tuned 3.09 billion parameter variant of the Qwen2.5 series, developed by Qwen. It builds upon the Qwen2 architecture with significant enhancements in several key areas. The model supports a full context length of 32,768 tokens and can generate up to 8,192 tokens, making it capable of handling and producing extensive text.

Key Capabilities

• Enhanced Knowledge & Reasoning: Significantly improved capabilities in coding and mathematics, leveraging specialized expert models.
• Instruction Following: Demonstrates substantial improvements in adhering to instructions and generating long texts.
• Structured Data Handling: Excels at understanding structured data, such as tables, and generating structured outputs, particularly JSON.
• Prompt Resilience: More robust against the diversity of system prompts, which enhances role-play implementation and condition-setting for chatbots.
• Multilingual Support: Offers comprehensive support for over 29 languages, including major global languages like Chinese, English, French, Spanish, and Japanese.

Good For

• Applications requiring strong coding and mathematical reasoning.
• Scenarios demanding precise instruction following and structured output generation (e.g., JSON).
• Chatbot development where role-play and diverse system prompts are crucial.
• Tasks involving long-context understanding and generation.
• Multilingual applications needing broad language support.