gustavecortal/Qwen3-psychological-reasoning-8B

Model Overview

The gustavecortal/Qwen3-psychological-reasoning-8B is an 8 billion parameter language model built upon the Qwen3 architecture. Its core distinction lies in its specialized fine-tuning on a dataset comprising 15,000 psychological and philosophical reasoning traces. This targeted training allows the model to engage with and reason about intricate concepts within psychology and philosophy.

Key Capabilities

• Specialized Reasoning: Excels in generating and understanding content related to psychological and philosophical concepts, including self-image, emotion, and existence.
• Domain-Specific Training: Fine-tuned using a filtered subset of open reasoning traces from Dolphin R1 and General Reasoning datasets.
• Qwen3 Base: Leverages the robust capabilities of the Qwen3 foundational model.

Methodology

The fine-tuning process involved domain filtering of prompts from the source datasets, embedding and clustering them using k-means, and then majority-voting for domain labels with Qwen3-1.7B. Clusters identified as psychology or philosophy were retained for LoRA fine-tuning, utilizing specific parameters (rank=8, alpha=16, max length=2048, epoch=1, batch size=16). This work was supported by HPC resources from GENCI-IDRIS.

Good For

• Applications requiring deep psychological analysis or philosophical discourse.
• Generating creative content or insights on human emotions, self-perception, and existential topics.
• Research and development in AI models with specialized reasoning capabilities in humanities.