Issues

• [Bug] Hard-coded trust_remote_code=True in the Cosmos3 runner lets a model bundle authorise its own Python

  bug
  ModelTC/LightX2V#1559 · Filter by author 3em0 opened on Sep 24, 2026

• Lightning LoRAs on Qwen-Image 2.1

  ModelTC/LightX2V#1548 · Filter by author htrvu opened on Sep 22, 2026

• May I know how to distill a light-weight VAE from one teacher like wan2.2?

  ModelTC/LightX2V#1509 · Filter by author iamweiweishi opened on Sep 10, 2026

• Missing FP8 quantized MiniMax-H3 models referenced in configs

  ModelTC/LightX2V#1445 · Filter by author nanwanj opened on Aug 26, 2026

• [Bug] Qwen-Image-Edit-2509 Scaled FP8 inference fails with BF16/FP8 dtype mismatch on RTX A6000

  bug
  ModelTC/LightX2V#1439 · Filter by author zddpk-cyber opened on Aug 26, 2026

• [Model Request] T3-Video (generates 2k videos very fast, almost realtime)

  ModelTC/LightX2V#1425 · Filter by author MeiYi-dev opened on Aug 24, 2026

• Replace deprecated v2.ToTensor() in wan_matrix_game2_runner.py with v2.ToImage() + v2.ToDtype(scale=True)

  ModelTC/LightX2V#1418 · Filter by author xyf5432 opened on Aug 23, 2026

• [Bug] Qwen2.5-VL image-edit encoder drops mm_token_type_ids with Transformers 5.15

  ModelTC/LightX2V#1416 · Filter by author llc1123 opened on Aug 22, 2026

• [model request] sensenova/SenseNova-U1.5-8B-MoT-Preview

  ModelTC/LightX2V#1347 · Filter by author dape1984 opened on Aug 10, 2026

• [Question] Scail-2 dedicated distill lora

  ModelTC/LightX2V#1295 · Filter by author bondobrus opened on Jul 24, 2026

• how can I get InfiniteTalk-4StepDistill-Mean-w-ITAudioAdaptorV6.1-fp8.safetensors

  ModelTC/LightX2V#1293 · Filter by author lianshiyong opened on Jul 24, 2026

• Release distilled SGMD video checkpoints on Hugging Face

  ModelTC/LightX2V#1286 · Filter by author NielsRogge opened on Jul 23, 2026

• [Bug] Wan2.2 LightX2V LoRA Merge Fails to Match cross_attn.k_img and cross_attn.v_img Keys

  bug
  ModelTC/LightX2V#1246 · Filter by author xuansu-genmo opened on Jul 14, 2026

• Do you have any plan to train a lightning lora for flux.2-klein-base 4B/9B?

  ModelTC/LightX2V#1182 · Filter by author 2502128021 opened on Jun 25, 2026

• 是否有计划加速mova模型呢？

  ModelTC/LightX2V#1178 · Filter by author yxqyyy opened on Jun 23, 2026

• Wan2.2-TI2V-5B I2V Generation Failed - Output is filled with meaningless color blocks (Official Model)

  ModelTC/LightX2V#1170 · Filter by author c-zt-p opened on Jun 19, 2026

• Is there any step distill plan for Bernini ？

  ModelTC/LightX2V#1166 · Filter by author wqf321 opened on Jun 18, 2026

• [Discussion] NVFP4 model quality issue

  bug
  ModelTC/LightX2V#1147 · Filter by author lobo1586 opened on Jun 12, 2026
  1

• [Feature] Add C500 Dockerfile

  ModelTC/LightX2V#1121 · Filter by author yl-1993 opened on Jun 5, 2026

• Do the wan22_moe_i2v_distill_model_4step_720p_260412 support the different target_video_length?

  ModelTC/LightX2V#1112 · Filter by author xinjiu612 opened on Jun 2, 2026
  2

• KeyError: 'target_video_length'

  ModelTC/LightX2V#1086 · Filter by author buptmengjj opened on May 22, 2026

• Any guidance / suggestion for B200 optimization?

  ModelTC/LightX2V#1054 · Filter by author nanwan03 opened on May 7, 2026
  4

• LightVAE Wan2.1

  ModelTC/LightX2V#1046 · Filter by author MHD-SMHD opened on Apr 29, 2026

• LoRa for Wan2.2-I2V-A14B-4step-720p

  bug
  ModelTC/LightX2V#1030 · Filter by author kingp0dd opened on Apr 21, 2026

• JoyAI runs on Intel XPU with Modern_Enable_Intel_XPU, but LightX2V does not fully support this model (noise output)

  ModelTC/LightX2V#1016 · Filter by author Tyr0727 opened on Apr 16, 2026