Issues

• 考虑支持fasth3-omni的代码推理和训练吗？

  modelscope/DiffSynth-Studio#1776 · Filter by author SudaLiruiqi opened on Oct 21, 2026

• minimax h3 训练 关于训练视频帧率10hz存在运动叠影

  modelscope/DiffSynth-Studio#1762 · Filter by author stihuangyuan opened on Oct 19, 2026

• wan2.2-I2V-A14B umt5_xxl_fp8_e4m3fn_scaled.safetensors

  modelscope/DiffSynth-Studio#1760 · Filter by author yitianlige-spec opened on Oct 18, 2026

• 基于Qwen-Image-2.1进行Lora微调，效果很快变差，而同样的任务，基于FLUX2-Klein-base-9B微调效果在变好而没有变差，是什么问题？该如何解决？

  modelscope/DiffSynth-Studio#1756 · Filter by author gobigrassland opened on Oct 15, 2026
  3

• support VDN-Turbo ref2va?

  modelscope/DiffSynth-Studio#1751 · Filter by author 01-111162-173 opened on Oct 14, 2026

• lingbot-video-moe-30b-a3b_ti2v.sh多机训练问题，每个机器各跑个的，没有并行

  modelscope/DiffSynth-Studio#1744 · Filter by author chenhangcal opened on Oct 13, 2026
  3

• MiniMax-H3 CFG 蒸馏模型 SFT 后发糊，可能是什么原因？

  modelscope/DiffSynth-Studio#1735 · Filter by author reaper19991110 opened on Oct 9, 2026
  9

• [Reproducibility] Training script / config for MiniMax-H3-TrainingAdapter (DeCFG adapter) itself

  modelscope/DiffSynth-Studio#1713 · Filter by author kmatzen opened on Sep 28, 2026
  1

• 麻烦问下，qwen image 2.1 如何进行图像编辑的训练？没看到相关数据集

  modelscope/DiffSynth-Studio#1709 · Filter by author whyPeanutbutter opened on Sep 24, 2026
  1

• 怎么能使用minimax h3 turbo lora、我测试了最新的 1.2版本、输出视频都是噪声

  modelscope/DiffSynth-Studio#1691 · Filter by author 01-111162-173 opened on Sep 14, 2026

• Silently slower attention kernel when training MiniMax-H3 LoRA on B200

  modelscope/DiffSynth-Studio#1687 · Filter by author TarzanZhao opened on Sep 11, 2026

• FLUX LoRA Fine-Tuning: Border Outpainting and Pixel-Level Misalignment Issues in Image Editing

  modelscope/DiffSynth-Studio#1685 · Filter by author lllllialois opened on Sep 11, 2026
  2

• oom on minimax h3

  modelscope/DiffSynth-Studio#1682 · Filter by author perryyang-hue opened on Sep 10, 2026
  1

• Partial download surfaces as "Cannot detect the model type" rather than a download error

  modelscope/DiffSynth-Studio#1668 · Filter by author rakhimovv opened on Sep 2, 2026

• LingBot-Video 30B LoRA training OOM on H100

  modelscope/DiffSynth-Studio#1664 · Filter by author Zanghu-ze opened on Sep 2, 2026
  1

• train_scale_cfg

  modelscope/DiffSynth-Studio#1663 · Filter by author smith-fugui opened on Sep 2, 2026
  3

• minimax-H3 FL2VA train error

  modelscope/DiffSynth-Studio#1660 · Filter by author stihuangyuan opened on Sep 1, 2026
  5

• Minimax h3 多卡如何推理？

  modelscope/DiffSynth-Studio#1645 · Filter by author 01-111162-173 opened on Aug 26, 2026
  1

• 训练minimax- h3的nf4模型的lora，使用adapter训练时融合lora会报错

  modelscope/DiffSynth-Studio#1642 · Filter by author professorhuanghaojing opened on Aug 26, 2026
  1

• minimax h3我看新增了 adater 训练，ref2av 目前是不支持这种训练方式吗

  modelscope/DiffSynth-Studio#1639 · Filter by author 01-111162-173 opened on Aug 26, 2026
  1

• DMD2训练损失震荡

  modelscope/DiffSynth-Studio#1617 · Filter by author 123lcy123 opened on Aug 19, 2026
  3

• [Proposal] 探讨对 Diffusion Language Model 的支持

  modelscope/DiffSynth-Studio#1609 · Filter by author tutao0123 opened on Aug 17, 2026
  1

• ltx2.5 会支持吗

  modelscope/DiffSynth-Studio#1595 · Filter by author 01-111162-173 opened on Aug 14, 2026
  2

• 会考虑支持d-opsd吗

  modelscope/DiffSynth-Studio#1588 · Filter by author NancyFyong opened on Aug 13, 2026
  9

• 训练minimax-h3的lora没办法在comfyui中使用

  modelscope/DiffSynth-Studio#1587 · Filter by author professorhuanghaojing opened on Aug 12, 2026
  3