MiniLLM Geometry Engine is a specialized, fine-tuned version of the TinyLlama 1.1B language model, optimized for generating geometry-related functions for a geometry engine. By fine-tuning TinyLlama 1.1B on a custom Geometry Chain-of-Thought (CoT) dataset, MiniLLM Geometry Engine excels at producing accurate and efficient mathematical functions tailored for geometric computations.
Key Features
Base Model: TinyLlama 1.1B, a lightweight and efficient language model.
Fine-Tuning: Trained on a Geometry CoT dataset, enabling step-by-step reasoning for geometry problems.
Output: Generates precise geometry functions (e.g., calculating areas, volumes, distances, or intersections) in a format compatible with geometry engines.
Applications: Ideal for educational tools, CAD software, game development, and computational geometry tasks.
Efficiency: Optimized for low-resource environments, balancing performance and computational cost.
Capabilities
MiniLLM Geometry Engine can interpret geometry-related queries and output functional code or mathematical expressions. For example, given a prompt like "Generate a function to calculate the area of a triangle," it produces executable code or formulas with clear reasoning, leveraging its CoT fine-tuning to ensure logical accuracy.
Technical Details
Dataset: Custom Geometry CoT dataset, including problems on 2D/3D shapes, trigonometry, and coordinate geometry, with step-by-step solutions.
Architecture: Inherits TinyLlama's transformer-based structure, fine-tuned to prioritize geometric reasoning.
Output Format: Produces functions in Python or pseudocode, compatible with geometry engine APIs.
Performance: Enhanced accuracy on geometry tasks compared to the base TinyLlama model, with minimal latency.
MiniLLM Geometry Engine is a powerful, compact solution for developers and researchers needing reliable geometry function generation in resource-constrained environments.
