Rumiii/Qwen3-8B-MedReasonPath
Rumiii/Qwen3-8B-MedReasonPath is an 8.2 billion parameter Qwen3-based causal language model with a 32768 token context length, fine-tuned for enhanced step-by-step medical reasoning. It specializes in clinical reasoning quality while retaining its native thinking mode and existing tool-calling capabilities. This model is designed for applications requiring robust medical problem-solving and agentic behavior.
Loading preview...
Overview
Rumiii/Qwen3-8B-MedReasonPath is an 8.2 billion parameter model built upon the Qwen3-8B architecture, specifically fine-tuned to improve step-by-step medical reasoning. It maintains the base model's native thinking mode (<think>...</think>) and its ability for agentic tool-calling. This is a full, standalone model, requiring no PEFT or additional libraries beyond transformers.
Key Capabilities
- Enhanced Medical Reasoning: Fine-tuned on the MedReason dataset for knowledge-graph grounded clinical reasoning.
- Preserved Agentic Behavior: Training included the xLAM Function-Calling dataset to ensure tool-calling functionality remains robust.
- Flexible Reasoning Modes: Supports both step-by-step reasoning (via
enable_thinking=True) and direct, non-reasoning answers (viaenable_thinking=False). - Standard Deployment: Can be loaded and used like any other
transformerscausal language model, with an available 8-bit quantized GGUF version forllama.cppor Ollama.
Good for
- Applications requiring detailed, step-by-step medical problem-solving.
- Scenarios where a model needs to perform both reasoning and utilize external tools.
- Research and educational projects in the medical domain, particularly for clinical reasoning tasks.