modelscope/DiffSynth-Studio

View on GitHub

Enjoy the magic of Diffusion models!

Stars: 13.1k
Forks: 1.3k
Open beginner issues: 1
Indexed issues: 542
Avg merge: 13h 12m
Merged PRs (30d): 45

Dominant language: Python
License: Apache-2.0
Last GitHub push: Sep 20, 2026
Latest indexed: Sep 20, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: No beginner labels indexed

Beginner friendly Everything

Open All

542 open issues indexed

Regarding multi-machine multi-GPU training for FLUX kontext Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#1021 · 1 comment · Opened Oct 31, 2025
关于FLUX kontext的多机多卡训练 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#1019 · 2 comments · Opened Oct 30, 2025
请问有办法降低训练wan2.2 animate的lora的显存要求吗 Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

modelscope/DiffSynth-Studio#1016 · 4 comments · Opened Oct 29, 2025
Operation timeout when using Wan2.2-Fun-A14B-Control with multiple GPUs Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1015 · 2 comments · Opened Oct 28, 2025
Question regarding the noise scheduler and training objectives Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1014 · 1 comment · Opened Oct 28, 2025
zero3保存模型通信错误 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#1013 · 1 comment · Opened Oct 27, 2025
DiffSynth-Studio/examples/wanvideo/model_inference/Wan2.2-Fun-A14B-Control.py 多卡推理出现噪声 Open

Difficulty 4/5 3-5 days Newbie friendliness 30/100

modelscope/DiffSynth-Studio#1012 · 5 comments · Opened Oct 27, 2025
wan2.2-S2V training always got OOM on H100-80GB GPU Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1010 · 2 comments · Opened Oct 23, 2025
qwen-image-edit-2509训练lora有测试用的example数据集吗 Open

Difficulty 1/5 Under an hour Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1009 · 2 comments · Opened Oct 22, 2025
Qwen-Image-Edit-2509 Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

modelscope/DiffSynth-Studio#1007 · 2 comments · Opened Oct 21, 2025
Probs about VACE full training setting of Num_frames Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1003 · 2 comments · Opened Oct 18, 2025
Expected to have finished reduction in the prior iteration before starting a new one. Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#1001 · 2 comments · Opened Oct 17, 2025
wan-animated推理训练数据长度问题 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#999 · 4 comments · Opened Oct 17, 2025
Question about optimal hyper parameters for training Qwen-Image-Edit-2509 Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#994 · 12 comments · Opened Oct 15, 2025
Is there a way to combine Control and Control-Camera models? Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

modelscope/DiffSynth-Studio#993 · 1 comment · Opened Oct 15, 2025
support of hunyuan-image-3.0? Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#992 · 2 comments · Opened Oct 14, 2025
Qwen-Image-Edit-2509 Extra Input Problem Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#991 · 2 comments · Opened Oct 14, 2025
关于Wan训练时的timestep Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#990 · 1 comment · Opened Oct 13, 2025
不知道为什么同时训练controlnet 和 lora ，controlnet 权重无法被保存？ Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#988 · 1 comment · Opened Oct 12, 2025
Support resume training of `QwenImagePipeline` LoRA Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#987 · 1 comment · Opened Oct 11, 2025
wan2.2 training 的时候，无法支持batch_size > 1 吗 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#985 · 3 comments · Opened Oct 10, 2025
NaN Loss after Forward Pass on Step 4 when Full Fine-Tuning Wan 2.2 5B Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#983 · 9 comments · Opened Oct 9, 2025
How to load 3 wan2.2 dit and avoid the OOM error? Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#981 · 2 comments · Opened Oct 9, 2025
Qwen-image的lora微调时，报错No qwen_image_blockwise_controlnet models available. Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#980 · 2 comments · Opened Oct 9, 2025
There might be a bug for Wan2.2 TI2V? Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#978 · 3 comments · 4 reactions · Opened Oct 7, 2025
qwen-image-edit lora训练是否有显存优化手段 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#977 · 4 comments · Opened Oct 5, 2025
关于保存模型 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#973 · 2 comments · Opened Oct 3, 2025
Unexpected output from WanVideoPostUnit_AnimateFacePixelValues Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#971 · 1 comment · 1 reaction · Opened Oct 2, 2025
会支持fsdp和fsdp2吗 Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#970 · 1 comment · 1 reaction · Opened Sep 30, 2025
No match with original code and library！！！ Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#967 · Opened Sep 30, 2025
Error: AttributeError: 'list' object has no attribute 'size' in example code of Qwen-Image-Edit Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#965 · 2 comments · Opened Sep 28, 2025
Unstable loss during Kontext lora training Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#964 · 1 comment · Opened Sep 28, 2025
'NameError: name 'vace' is not defined' when run Wan2.2-T2V-A14B.sh for train Open

Difficulty 3/5 1-2 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#962 · 1 comment · Opened Sep 26, 2025
How to convert qwen-image LoRA to diffusers Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#961 · 4 comments · 1 reaction · Opened Sep 26, 2025
Cannot load Qwen-Image-Edit-2509 Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#959 · 2 comments · Opened Sep 26, 2025
WAN VACE comfy / diffusers lora compatibility Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#958 · 1 comment · Opened Sep 25, 2025
如何查看lora训练过程中的loss下降曲线，像tensorboard那样 Open

Difficulty 3/5 1-2 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#957 · 3 comments · Opened Sep 25, 2025
Wan2.2-S2V-14B Tea Cache coefficients Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#956 · 1 comment · Opened Sep 25, 2025
请问最新版本必须要大于python10吗？使用python9报错 Open

Difficulty 1/5 Under an hour Newbie friendliness 35/100

modelscope/DiffSynth-Studio#953 · 1 comment · Opened Sep 25, 2025
Why developers not replying to the Issues? Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

modelscope/DiffSynth-Studio#951 · 9 comments · Opened Sep 25, 2025
Error in LoRA fune VACE-14B Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

modelscope/DiffSynth-Studio#950 · 3 comments · Opened Sep 25, 2025
Flux full train的模型转opensource format Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#948 · 4 comments · Opened Sep 25, 2025
wan2.2能否支持teacache或magcache等加速方案？ Open

Difficulty 5/5 Over a week Newbie friendliness 20/100

modelscope/DiffSynth-Studio#946 · 1 comment · Opened Sep 24, 2025
Wan2.1 lora training loss Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#943 · 12 comments · 1 reaction · Opened Sep 24, 2025
为什么Qwen_Image的推理代码里面的text_encoder部分和modelscope以及huggingface里面的text_encoder的权重中的key值对不上？ Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

modelscope/DiffSynth-Studio#942 · 3 comments · 1 reaction · Opened Sep 23, 2025
Support for Flux.1-Fill Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

modelscope/DiffSynth-Studio#939 · 1 comment · Opened Sep 23, 2025
Cinescale Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#938 · 1 comment · Opened Sep 22, 2025
hello，求支持wan2.2 animate的训练 Open

Difficulty 5/5 Over a week Newbie friendliness 15/100

modelscope/DiffSynth-Studio#936 · 3 comments · Opened Sep 22, 2025
Qwen-Image-Edit FP8 inference without CPU offloading Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

modelscope/DiffSynth-Studio#935 · 1 comment · Opened Sep 22, 2025
Failed to load diffusers weights Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

modelscope/DiffSynth-Studio#934 · 1 comment · Opened Sep 22, 2025

modelscope/DiffSynth-Studio

Get new issues in your inbox