gustavecortal/Qwen3-psychological-reasoning-4B
The gustavecortal/Qwen3-psychological-reasoning-4B is a 4 billion parameter language model, based on the Qwen3 architecture, fine-tuned by gustavecortal. It specializes in psychological and philosophical reasoning, trained on 15,000 reasoning traces from datasets like Dolphin R1 and General Reasoning. This model is designed to understand and generate content related to complex psychological and philosophical concepts, offering a 32768 token context length.
Loading preview...
Overview
The gustavecortal/Qwen3-psychological-reasoning-4B is a 4 billion parameter language model, developed by gustavecortal, specifically fine-tuned for psychological and philosophical reasoning. It is built upon the Qwen3 base model and leverages a 32768 token context length.
Key Capabilities
- Specialized Reasoning: The model is trained on 15,000 psychological and philosophical reasoning traces, enabling it to understand and generate content related to complex concepts such as self-image, emotion, and existence.
- Domain-Specific Fine-tuning: It was fine-tuned using a filtered subset of the Dolphin R1 and General Reasoning datasets, focusing on psychology and philosophy clusters.
- Qwen3 Base: Benefits from the underlying capabilities of the Qwen3 architecture.
Methodology
The fine-tuning process involved domain filtering using embedding and k-means clustering, with Qwen3-1.7B assisting in majority-voting for domain labels. Only clusters tagged as psychology or philosophy were used for LoRA fine-tuning (rank=8, alpha=16, max length=2048, epoch=1, batch size=16).
Good For
- Applications requiring nuanced understanding and generation of psychological concepts.
- Tasks involving philosophical inquiry and reasoning.
- Research and development in AI focused on cognitive and emotional intelligence.