reaperdoesntknow/Qwen3-1.7B-Thinking-Distil
reaperdoesntknow/Qwen3-1.7B-Thinking-Distil is a 2 billion parameter Qwen3ForCausalLM model developed by Convergent Intelligence LLC: Research Division. It is a distilled version of the Qwen3-30B-A3B-Thinking teacher model, specifically fine-tuned on the longwriter-6k dataset to capture and reproduce extended reasoning chains and deliberative thought patterns. With a context length of 40,960 tokens, this model excels at generating long-form content that demonstrates internal monologue, reasoning through uncertainty, and re-evaluation before reaching a conclusion, making it suitable for complex analytical tasks.
Loading preview...
Overview
reaperdoesntknow/Qwen3-1.7B-Thinking-Distil is a 2 billion parameter model from Convergent Intelligence LLC: Research Division, designed to emulate the extended reasoning capabilities of its larger teacher model, Qwen3-30B-A3B-Thinking. This model is distilled via supervised fine-tuning on the longwriter-6k dataset, which preserves long-form generation samples with detailed reasoning chains. Unlike other distillation variants that focus on instruction following or hierarchical decomposition, this model specifically captures the teacher's internal monologue, including deliberation, backtracking, and re-evaluation processes.
Key Capabilities
- Extended Reasoning: Reproduces complex thought processes, uncertainty handling, and re-evaluation steps from a 30B parameter teacher model.
- Long-Form Generation: Capable of generating detailed and coherent responses up to 4096 tokens, with a maximum context length of 40,960 tokens.
- Deliberative Depth: Captures the nuanced deliberative patterns, making it suitable for tasks requiring in-depth analysis and explanation.
- Efficient Distillation: Achieves advanced reasoning capabilities in a compact 1.7B parameter size, making it efficient for deployment.
Good For
- Complex Problem Solving: Ideal for applications requiring models to "think aloud" or show their reasoning steps.
- Analytical Tasks: Suited for generating explanations, justifications, and detailed analyses where the process of arriving at an answer is as important as the answer itself.
- Content Generation: Useful for creating long-form content that benefits from a deliberative and reflective tone.
- Research and Development: Provides a compact model for exploring advanced reasoning patterns and knowledge distillation techniques, particularly those grounded in Discrepancy Calculus.