Piaget-8B: Specialized Reasoning in Psychology and Philosophy
Piaget-8B is an 8 billion parameter language model developed by gustavecortal, built upon the Qwen3 architecture. Its core distinction lies in its fine-tuning on a unique dataset of 15,000 psychological and philosophical reasoning traces, derived from subsets of the Dolphin R1 and General Reasoning datasets. This specialized training enables Piaget-8B to excel in tasks requiring deep understanding and generation of content related to complex human thought and existence.
Key Capabilities
- Psychological Reasoning: Proficient in analyzing and generating insights on concepts such as self-image, emotion, and cognitive processes.
- Philosophical Inquiry: Capable of engaging with and reasoning about abstract philosophical ideas and arguments.
- Contextual Understanding: Benefits from a 32,768 token context window, allowing for comprehensive analysis of detailed prompts.
Methodology Highlights
The model's development involved domain filtering using Qwen3-1.7B to identify and retain psychology and philosophy-related clusters from the base datasets. This targeted approach, utilizing k-means clustering and LoRA finetuning, ensures its specialized performance. The project was inspired by research into improving language models for emotion analysis, aiming to bridge cognitive science insights with AI capabilities.
Ideal Use Cases
Piaget-8B is particularly well-suited for applications in:
- Academic research in psychology and philosophy.
- Content generation requiring nuanced understanding of human thought and emotion.
- Developing tools for psychotherapeutic concept exploration.
- Educational platforms focusing on humanities and social sciences.