README.md

metadata

license: apache-2.0
base_model: Qwen/Qwen3-8B
tags:
  - manim
  - manim-voiceover
  - text-to-speech
  - aos
  - lora
  - sft

AOS-qwen3-8b-narrated-adapter

Fine-tuned LoRA adapter on 400 Narrated Manim Trajectories using manim-voiceover.

Features

• Pedagogical Audio Grounding: Synthesizes synchronized VoiceoverScene Python scripts.
• Timing Synchronization: Generates animations bounded by run_time=tracker.duration.
• Phonetic LaTeX: Spoken voiceover scripts express formulas verbally without raw LaTeX syntax errors.

Base Model

• Qwen/Qwen3-8B