Insighta Mandala v13: Structured Learning Plan Generation
Insighta Mandala v13 is a specialized 4 billion parameter language model, built upon the Qwen3-4B architecture and fine-tuned by jamesjk4242. Its core capability lies in generating structured JSON-formatted mandala learning plans, making it unique among general-purpose LLMs. The model supports both Korean and English, allowing for versatile application in different linguistic contexts.
Key Capabilities
- Mandala Learning Plan Generation: Specifically designed to create detailed, organized learning plans in a mandala chart format.
- Structured JSON Output: Outputs plans as clean, parseable JSON, facilitating integration into applications or workflows.
- Multilingual Support: Processes prompts and generates content in both Korean and English.
- Optimized for Specific Task: Fine-tuned using LoRA for high performance on its niche task, rather than broad general knowledge.
Good For
- Educational Tools: Developing applications that require automated generation of structured study plans.
- Personalized Learning: Creating customized learning roadmaps for students based on specific goals (e.g., "TOEFL 100점 만다라트 차트").
- Structured Data Generation: Any use case requiring a model to reliably output complex, nested JSON structures based on natural language prompts.
Quantized versions, including Q4_K_M.gguf (recommended for CPU), are available for efficient deployment.