ltx-2.3-22b

🎤 LTX-2.3 22B Distilled — Audio-to-Video / Lip Sync

<div align="center"> <img src="https://img.shields.io/badge/Maintained%20by-vinh--gogo-blue?style=for-the-badge&logo=github" hrep="https://github.com/vinh-gogo"> <img src="https://img.shields.io/badge/YouTube-Hinn%20Channel-red?style=for-the-badge&logo=youtube"> <img src="https://img.shields.io/badge/Optimized%20for-Google%20Colab-F9AB00?style=for-the-badge&logo=googlecolab"> </div>

📌 Overview / Tổng quan

This repository contains the backup weights for the LTX-2.3 22B Distilled model, optimized for Audio-to-Video (Lip Sync) tasks. All symlinks have been resolved to ensure safe and easy downloads.

🚀 Specifications / Đặc điểm kỹ thuật

• Architecture: LTX-Video (Distilled 8-steps)
• Format: GGUF (Q4_K_M) + Safetensors
• Text Encoder: Gemma 3 12B IT (bfloat16)
• Hardware Requirement: Recommended for Google Colab with T4 (16GB VRAM) or L4 (22.5GB VRAM) GPU.

📁 Directory Structure / Cấu trúc thư mục

• ltx-2.3-22b-distilled-Q4_K_M_light.gguf: Main Transformer weights.
• gemma-3-12b-it-qat-q4_0-unquantized/: Full Text Encoder.
• *.safetensors: Essential modules (VAE, Vocoder, Spatial Upscaler x2).

---

Maintained by vinh-gogo. Follow for more AI tutorials at Hinn Channel.