Issues

• No valid video file is generated with Wan2.1-1.3B

  thu-ml/TurboDiffusion#142 · Filter by author RhythmOfTheRain-Byte opened on Oct 15, 2026

• RuntimeError: Triton Error [CUDA]: an illegal memory access was encountered

  thu-ml/TurboDiffusion#141 · Filter by author RhythmOfTheRain-Byte opened on Oct 15, 2026

• DreamX-World: A General-Purpose Interactive World Model

  thu-ml/TurboDiffusion#135 · Filter by author shuidongliu opened on Jun 18, 2026

• 请求新增 RTX 5090 平台 FP8 推理优化分支

  thu-ml/TurboDiffusion#131 · Filter by author Jordanyang opened on Jun 10, 2026

• 关于merge的一些疑问

  thu-ml/TurboDiffusion#130 · Filter by author lxer66 opened on Jun 3, 2026

• 能否支持多卡并行推理呢？

  thu-ml/TurboDiffusion#128 · Filter by author yxqyyy opened on May 29, 2026

• 请问当前支持在NPU安装，推理吗？

  thu-ml/TurboDiffusion#127 · Filter by author ysuLihua opened on May 27, 2026

• Strategy for applying rCM with Autoregressive Student

  thu-ml/TurboDiffusion#126 · Filter by author lapp0 opened on Apr 26, 2026
  1

• 关于wan 2.2 5B模型支持

  thu-ml/TurboDiffusion#124 · Filter by author weiyithu opened on Mar 20, 2026

• net.load_state_dict(state_dict, assign=True) 失败

  thu-ml/TurboDiffusion#122 · Filter by author JackBlue302 opened on Mar 9, 2026
  1

• Generalization of the merging operation

  thu-ml/TurboDiffusion#121 · Filter by author toyot-li opened on Mar 9, 2026
  2

• hello,请教下关于模型兼容性的问题

  thu-ml/TurboDiffusion#120 · Filter by author godwenbin opened on Mar 5, 2026
  3

• Illegal Memory Access on 5090

  thu-ml/TurboDiffusion#119 · Filter by author loaydatrain opened on Mar 5, 2026
  1

• QA What benchmark/dataset is used for evaluations?

  thu-ml/TurboDiffusion#118 · Filter by author mikasa955 opened on Feb 25, 2026

• sla是全参微调吗，还是只训练SparseLinearAttention？

  thu-ml/TurboDiffusion#117 · Filter by author wzr0108 opened on Feb 25, 2026
  3

• Question about different patch embedding implementation compared to official Wan-Video

  thu-ml/TurboDiffusion#114 · Filter by author ChenJiongxu355 opened on Feb 14, 2026
  1

• How is the performance if we train with sCM loss only?

  thu-ml/TurboDiffusion#111 · Filter by author meng-ustc opened on Feb 1, 2026
  1

• Sparse Linear Layer weights are 0 for 14B 480p model

  thu-ml/TurboDiffusion#109 · Filter by author ulucsahin opened on Jan 27, 2026

• 帧率比较低？

  thu-ml/TurboDiffusion#108 · Filter by author ScottishFold007 opened on Jan 27, 2026
  1

• 推理效果

  thu-ml/TurboDiffusion#107 · Filter by author marmalade-s opened on Jan 27, 2026
  1

• merge逻辑

  thu-ml/TurboDiffusion#105 · Filter by author OO848 opened on Jan 23, 2026
  2

• [Benchmark] Inference Performance on RTX 5090 with Wan2.1-1.3B (PyTorch 2.8.0)

  thu-ml/TurboDiffusion#102 · Filter by author haoyuhe04 opened on Jan 19, 2026
  1

• 你好，请教一下作者是怎么将原始wan2.1的tensor 格式，转成pth格式的

  thu-ml/TurboDiffusion#101 · Filter by author Qicheng-Lin opened on Jan 16, 2026
  1

• 你好，为什么在尝试SLA训练时，会卡在这里

  thu-ml/TurboDiffusion#100 · Filter by author Qicheng-Lin opened on Jan 16, 2026
  2

• loss 0.0000

  thu-ml/TurboDiffusion#99 · Filter by author vje526 opened on Jan 16, 2026
  5