Lightx2v
lightx2v
AI & ML interests
We are the Model-ToolChain team. We are interested in model compression and inference deployment for AI models.
Experience page: https://x2v.light-ai.top. Homepage: https://light-ai.top
Recent Activity
new activity about 11 hours ago
lightx2v/Minimax-h3-Turbo:8-step 768p V1.0 LoRA released — improved video and audio quality updated a model about 11 hours ago
lightx2v/Minimax-h3-Turbo published a model 1 day ago
lightx2v/MiniMax-H3-Prompt-Rewriter-LoRA-OmniOrganizations
FewStep RL for Wan models
-
lightx2v/Qwen-Image-Lightning
Text-to-Image • Updated • 435k • • 822 -
lightx2v/Qwen-Image-Edit-2511-Lightning
Image-to-Image • Updated • 336k • • 516 -
lightx2v/Qwen-Image-2512-Lightning
Text-to-Image • Updated • 77k • • 228 -
lightx2v/Qwen-Image-Edit-Causal
Image-Text-to-Image • 20B • Updated • 1.46k • 18
Efficient autoregressive video generation (i.e., the Self-Forcing family) checkpoints.
FewStep RL for Wan models
-
lightx2v/Qwen-Image-Lightning
Text-to-Image • Updated • 435k • • 822 -
lightx2v/Qwen-Image-Edit-2511-Lightning
Image-to-Image • Updated • 336k • • 516 -
lightx2v/Qwen-Image-2512-Lightning
Text-to-Image • Updated • 77k • • 228 -
lightx2v/Qwen-Image-Edit-Causal
Image-Text-to-Image • 20B • Updated • 1.46k • 18
Efficient autoregressive video generation (i.e., the Self-Forcing family) checkpoints.
Model collections trained with our framework: https://github.com/ModelTC/GenRL