From 09c8f852fbeff021175180edafe2dae5949865fa Mon Sep 17 00:00:00 2001 From: helloyongyang Date: Tue, 1 Sep 2026 02:52:50 +0000 Subject: [PATCH 1/3] update h3 dmd2 config --- .../train/dmd/minimax_h3_t2av_dmd_lora.yaml | 24 ++++++------------- 1 file changed, 7 insertions(+), 17 deletions(-) diff --git a/lightx2v_train/configs/train/dmd/minimax_h3_t2av_dmd_lora.yaml b/lightx2v_train/configs/train/dmd/minimax_h3_t2av_dmd_lora.yaml index fffb313b1..441810278 100644 --- a/lightx2v_train/configs/train/dmd/minimax_h3_t2av_dmd_lora.yaml +++ b/lightx2v_train/configs/train/dmd/minimax_h3_t2av_dmd_lora.yaml @@ -90,7 +90,7 @@ training: train_type: lora lora: rank: 128 - alpha: 128 + alpha: 8 target_modules: - to_q - to_k @@ -99,33 +99,23 @@ training: - ff.net.0.proj - ff.net.2 optimizer: - learning_rate: 0.0001 - adam_beta1: 0.9 + learning_rate: 0.00005 + adam_beta1: 0.0 adam_beta2: 0.999 weight_decay: 0.01 adam_epsilon: 0.00000001 fake: - train_type: lora - lora: - rank: 128 - alpha: 128 - target_modules: - - to_q - - to_k - - to_v - - to_out.0 - - ff.net.0.proj - - ff.net.2 + train_type: full optimizer: - learning_rate: 0.00002 - adam_beta1: 0.9 + learning_rate: 0.0000004 + adam_beta1: 0.0 adam_beta2: 0.999 weight_decay: 0.01 adam_epsilon: 0.00000001 teacher: guidance_scale: 1.0 cfg_norm: none - output_dir: /data/nvme6/gushiqiao/codes/latest/LightX2V/lightx2v_train/output_train/minimax_h3_t2av_dmd_lora1 + output_dir: ./output_train/minimax_h3_t2av_dmd_zoe_fake_full inference: method: none From d4c3fa3c05aba226320424213926fe27e9a2ac5f Mon Sep 17 00:00:00 2001 From: helloyongyang Date: Tue, 1 Sep 2026 03:00:35 +0000 Subject: [PATCH 2/3] update readme --- README.md | 3 ++- README_zh.md | 3 ++- 2 files changed, 4 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 92bab06bf..39d8e83d7 100644 --- a/README.md +++ b/README.md @@ -7,7 +7,6 @@ [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/ModelTC/lightx2v) [![Doc](https://img.shields.io/badge/docs-English-99cc2)](https://lightx2v-en.readthedocs.io/en/latest) [![Doc](https://img.shields.io/badge/文档-中文-99cc2)](https://lightx2v-zhcn.readthedocs.io/zh-cn/latest) -[![Papers](https://img.shields.io/badge/论文集-中文-99cc2)](https://lightx2v-papers-zhcn.readthedocs.io/zh-cn/latest) [![Docker](https://img.shields.io/badge/Docker-2496ED?style=flat&logo=docker&logoColor=white)](https://hub.docker.com/r/lightx2v/lightx2v/tags) **\[ English | [中文](README_zh.md) \]** @@ -222,6 +221,7 @@ pipe.generate( ## 🤖 Supported Model Ecosystem ### Official Open-Source Models +- ✅ [MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3) - ✅ [LTX-2.3](https://huggingface.co/Lightricks/LTX-2.3) - ✅ [LTX-2](https://huggingface.co/Lightricks/LTX-2) - ✅ [HunyuanVideo-1.5](https://huggingface.co/tencent/HunyuanVideo-1.5) @@ -316,6 +316,7 @@ We sincerely thank all the model repositories and research communities that insp - [Tencent-Hunyuan](https://github.com/Tencent-Hunyuan) - [Wan-Video](https://github.com/Wan-Video) - [Qwen-Image](https://github.com/QwenLM/Qwen-Image) +- [MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3) - [LightLLM](https://github.com/ModelTC/LightLLM) - [sglang](https://github.com/sgl-project/sglang) - [vllm](https://github.com/vllm-project/vllm) diff --git a/README_zh.md b/README_zh.md index 1905ca872..1e65a084a 100644 --- a/README_zh.md +++ b/README_zh.md @@ -7,7 +7,6 @@ [![Ask DeepWiki](https://deepwiki.com/badge.svg)](https://deepwiki.com/ModelTC/lightx2v) [![Doc](https://img.shields.io/badge/docs-English-99cc2)](https://lightx2v-en.readthedocs.io/en/latest) [![Doc](https://img.shields.io/badge/文档-中文-99cc2)](https://lightx2v-zhcn.readthedocs.io/zh-cn/latest) -[![Papers](https://img.shields.io/badge/论文集-中文-99cc2)](https://lightx2v-papers-zhcn.readthedocs.io/zh-cn/latest) [![Docker](https://img.shields.io/badge/Docker-2496ED?style=flat&logo=docker&logoColor=white)](https://hub.docker.com/r/lightx2v/lightx2v/tags) **\[ [English](README.md) | 中文 \]** @@ -221,6 +220,7 @@ pipe.generate( ## 🤖 支持的模型生态 ### 官方开源模型 +- ✅ [MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3) - ✅ [LTX-2.3](https://huggingface.co/Lightricks/LTX-2.3) - ✅ [LTX-2](https://huggingface.co/Lightricks/LTX-2) - ✅ [HunyuanVideo-1.5](https://huggingface.co/tencent/HunyuanVideo-1.5) @@ -316,6 +316,7 @@ pipe.generate( - [Tencent-Hunyuan](https://github.com/Tencent-Hunyuan) - [Wan-Video](https://github.com/Wan-Video) - [Qwen-Image](https://github.com/QwenLM/Qwen-Image) +- [MiniMax-H3](https://github.com/MiniMax-AI/MiniMax-H3) - [LightLLM](https://github.com/ModelTC/LightLLM) - [sglang](https://github.com/sgl-project/sglang) - [vllm](https://github.com/vllm-project/vllm) From 6f8c3fc6ed27193218f10e49ef42eea4f1bddffe Mon Sep 17 00:00:00 2001 From: helloyongyang Date: Tue, 1 Sep 2026 03:05:08 +0000 Subject: [PATCH 3/3] Update LightX2V Studio models --- README.md | 2 +- README_zh.md | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 39d8e83d7..cb79910b5 100644 --- a/README.md +++ b/README.md @@ -17,7 +17,7 @@ **LightX2V** is an advanced lightweight image/video generation inference framework engineered to deliver efficient, high-performance image/video synthesis solutions. This unified platform integrates multiple state-of-the-art image/video generation techniques, supporting diverse generation tasks including text-to-video (T2V), image-to-video (I2V), text-to-image (T2I), image-editing (I2I). **X2V represents the transformation of different input modalities (X, such as text or images) into vision output (Vision)**. -> 🌐 **Try it online now!** Experience LightX2V without installation: **[LightX2V Studio](https://x2v.light-ai.top/)** — a free, lightweight AI video platform with **LTX 2.3**, **Wan 2.2**, **SekoTalk**, **Qwen-Image**, **SeedVR2**, and more models and tasks. +> 🌐 **Try it online now!** Experience LightX2V without installation: **[LightX2V Studio](https://x2v.light-ai.top/)** — a free, lightweight AI video platform with **Minimax H3**, **Wan 2.2**, **SekoTalk**, **Qwen-Image**, **SwiftVR**, and more models and tasks. > 🤗 **HuggingFace Model Repository: [LightX2V HuggingFace](https://huggingface.co/lightx2v)** diff --git a/README_zh.md b/README_zh.md index 1e65a084a..09ef3708f 100644 --- a/README_zh.md +++ b/README_zh.md @@ -17,7 +17,7 @@ **LightX2V** 是一个先进的轻量级图像视频生成推理框架，专为提供高效、高性能的图像视频生成解决方案而设计。该统一平台集成了多种前沿的图像视频生成技术，支持文本生成视频(T2V)和图像生成视频(I2V)，文本生图片(T2I)，图像编辑(I2I)等多样化生成任务。**X2V 表示将不同的输入模态(X，如文本或图像)转换为视觉输出(Vision)**。 -> 🌐 **立即在线体验！** 无需安装即可体验 LightX2V：**[LightX2V Studio](https://x2v.light-ai.top/)** - 免费、轻量、快速的AI视频生成平台，提供LTX2.3 / Wan 2.2 / SekoTalk / Qwen-Image / SeedVR2等多个模型多种任务的免费体验使用。 +> 🌐 **立即在线体验！** 无需安装即可体验 LightX2V：**[LightX2V Studio](https://x2v.light-ai.top/)** - 免费、轻量、快速的AI视频生成平台，提供 Minimax H3 / Wan 2.2 / SekoTalk / Qwen-Image / SwiftVR 等多个模型多种任务的免费体验使用。 > 🤗 **HuggingFace 模型仓库: [LightX2V HuggingFace](https://huggingface.co/lightx2v)**