Website Home (ChinaZ.com) June 21 News: Tencent's Hunyuan Text-to-Image Large Model (Hunyuan DiT Model) has announced the open-source release of its training code, including LoRA plugins and ControlNet plugins.
LoRA is a technology used for fine-tuning large language models, allowing the training of models with specific features using a small amount of data without modifying the original model or increasing its size.
ControlNet is a controllable generation algorithm that allows users to better control image generation by adding extra conditions. Tencent Hunyuan has introduced three initial ControlNet models that can extract and apply conditions such as image edges, depth, and human poses.
Additionally, the Hunyuan DiT has released a dedicated acceleration library to enhance inference efficiency and simplify usage. The Hunyuan DiT has been widely used in various fields including material creation, product synthesis, and game image generation, and is utilized by platforms like Tencent Advertising's Miaosi and multiple media outlets for content generation.
Official Website:
https://dit.hunyuan.tencent.com/
Code:
https://github.com/Tencent/HunyuanDiT
Model:
https://huggingface.co/Tencent-Hunyuan/HunyuanDiT
Paper:
https://tencent.github.io/HunyuanDiT/asset/Hunyuan_DiT_Tech_Report_05140553.pdf
Data Creation Process:
https://github.com/Tencent/HunyuanDiT/blob/main/IndexKits/docs/MakeDataset.md