Pyxidis-Manim-CodeGen-1.7B: Specialized Math Animation Code Generation
Pyxidis-Manim-CodeGen-1.7B is an experimental 1.7 billion parameter model developed by prithivMLmods, specifically fine-tuned on Qwen/Qwen3-1.7B using Manim-CodeGen code traces. Its core purpose is to generate Python code for mathematical animations using the Manim library, making it a highly specialized tool for educational and research-driven visualization.
Key Capabilities
- Manim-Specific Code Generation: Optimized for Python-based animation scripting of mathematical concepts and visual proofs.
- Math + Code Synergy: Generates step-by-step mathematical derivations alongside corresponding Manim animation code.
- Animation Workflow Optimization: Produces structured code for Manim scenes, transformations, graphs, and equations, reducing manual effort.
- Python-Centric Reasoning: Focuses on generating clean, modular, and reusable Python code for animation pipelines.
- Structured Output: Capable of outputting in Python, Markdown, and LaTeX formats, suitable for tutorials and automated content generation.
- Lightweight & Specialized: Designed for Manim coding efficiency with a deployable footprint, ideal for GPU clusters and research labs.
Intended Use Cases
- Manim-based math animation coding for research, teaching, and content creation.
- Educational visualization assistant to convert mathematical problems into animated explanations.
- Python tutoring tool for workflows involving mathematical animations.
- Prototype generator for interactive STEM video content.
Limitations
As an experimental model, it may require manual debugging of generated code. It is limited to Manim coding workflows and is not a general-purpose code assistant. Complex multi-scene projects may require iterative refinement, and its optimization prioritizes structured math and animation reasoning over general dialogue.