reaperdoesntknow/Dualmind-Qwen-1.7B-Thinking
Dualmind-Qwen-1.7B-Thinking is a 2.03 billion parameter Qwen3ForCausalLM model developed by Convergent Intelligence LLC: Research Division, fine-tuned using the DualMind SFT methodology. It is trained on over 2.5 million tokens of Claude Opus 4.6 reasoning traces, specifically designed to absorb and reproduce extended deliberation and self-correction patterns. This model excels at complex reasoning tasks by mimicking the nuanced thought processes of a frontier AI, with a maximum context length of 40,960 tokens.
Loading preview...
Dualmind-Qwen-1.7B-Thinking: Opus-Inspired Reasoning
This model, developed by Convergent Intelligence LLC: Research Division, is a 2.03 billion parameter Qwen3ForCausalLM (1.7B effective) specifically engineered for advanced reasoning. It leverages the DualMind SFT methodology by training on over 2.5 million tokens of Claude Opus 4.6 reasoning traces from the Opus-4.6-Reasoning-3000x-filtered dataset. Unlike models trained on synthetic logic, this variant learns the "shape of deliberation" from a frontier model, including backtracking, hedging, and synthesizing multiple approaches.
Key Capabilities & Features
- Opus-like Reasoning: Absorbs the nuanced self-correction and deliberative structure of Claude Opus 4.6.
- Extended Deliberation: Designed to produce multi-phase reasoning, exploring, reconsidering, and concluding naturally.
- Strong Foundation: Built upon
Disctil-Qwen3-1.7B, which is DISC-refined, providing a robust structural base. - High Context Length: Supports a maximum context length of 40,960 tokens, allowing for longer, more complex reasoning chains.
- DualMind Family: Part of a series exploring different reasoning teachers; this is the Opus variant, complementing the LogicInference and TopologicalQwen models.
Ideal Use Cases
- Complex Problem Solving: Suited for tasks requiring detailed, multi-step reasoning and self-correction.
- Simulating Deliberative Thought: Useful for applications where an AI needs to demonstrate uncertainty navigation and nuanced thought processes.
- Research into AI Reasoning: Provides a compact model for studying how reasoning patterns from frontier models can be distilled into smaller architectures.