ManimBench v1 is a curated dataset designed to support fine-tuning and evaluation of models that generate or interpret Manim animations. It pairs Manim code snippets with natural language descriptions, enabling research in code generation, animation synthesis, and multimodal understanding.
🔗 GitHub Repository for Fine-Tuning: SuienS/manim-fine-tune-v1
📄 Research Paper: Large Language Model Approaches to Educational Video Generation Using Manim
The… See the full description on the dataset page:
https://huggingface.co/datasets/SuienR/ManimBench-v1.
