nabin2004/AOS-qwen3-8b-narrated-sft-merged

TEXT GENERATIONPricing:Input $0.468 / Output $1.82Concurrent Unit Cost:1Model Size:8BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 4, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

nabin2004/AOS-qwen3-8b-narrated-sft-merged is an 8 billion parameter Qwen3-based language model, fine-tuned for generating Manim Community Edition scripts. This model specializes in creating executable Manim `VoiceoverScene` code, integrating synchronized audio bookmarks and adhering to CE API compliance. It is designed for mathematical and scientific animation with narration.

Loading preview...

AOS Qwen3 8B Narrated (Continued SFT Merged)

This model is a supervised fine-tuned (SFT) version of the Qwen/Qwen3-8B base model, specifically designed for generating Manim Community Edition (CE) scripts. It is released as a bf16 Safetensors model, built upon an adapter trained by nabin2004. The primary focus of this model is to facilitate the creation of mathematical and scientific animations that are synchronized with voiceovers using manim-voiceover.

Key Capabilities

  • VoiceoverScene Architecture: Generates complete and executable Manim scripts that inherit from VoiceoverScene, enabling complex narrated animations.
  • Audio Bookmarking: Integrates precise synchronized audio markers (tracker.wait_until_bookmark(...)) within the generated Manim code, ensuring visual animations align perfectly with narration.
  • CE API Compliance: Adheres strictly to Manim Community Edition syntax, avoiding deprecated legacy APIs and ensuring compatibility with modern Manim workflows.

Good For

  • Developers and educators looking to automate the creation of narrated mathematical and scientific animations.
  • Generating Manim code that requires precise synchronization between visual elements and audio.
  • Ensuring Manim scripts are compliant with the latest Manim Community Edition API standards.

Finetunes
1 models