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

EliGen如何实现基于原图的inpaint Open

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

modelscope/DiffSynth-Studio#1098 · 1 comment · Opened Dec 5, 2025
Why there is no unconditional input while we training the model Open

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

modelscope/DiffSynth-Studio#1097 · 3 reactions · Opened Dec 5, 2025
编辑模型能使用lora差分训练吗？ Open

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

modelscope/DiffSynth-Studio#1095 · 1 comment · Opened Dec 4, 2025
flux2 多图编辑 Open

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

modelscope/DiffSynth-Studio#1094 · Opened Dec 4, 2025
Traning data of Qwen-Image-Edit-Lowres-Fix Open

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

modelscope/DiffSynth-Studio#1092 · 4 comments · Opened Dec 3, 2025
VACE finetune时候，如何断点训练？ Open

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

modelscope/DiffSynth-Studio#1090 · 1 comment · Opened Dec 2, 2025
FLUX.2-dev训练显存配置 Open

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

modelscope/DiffSynth-Studio#1089 · 5 comments · Opened Dec 2, 2025
Wan2.2 TI2V 的VAE compression ratio Open

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

modelscope/DiffSynth-Studio#1088 · Opened Dec 1, 2025
会有加入z-image模型吗 Open

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

modelscope/DiffSynth-Studio#1087 · 1 comment · 2 reactions · Opened Nov 30, 2025
DiffSynth-Studio/data/example_image_dataset/中的csv 和json有什么区别么？ Open

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

modelscope/DiffSynth-Studio#1086 · 2 comments · Opened Nov 28, 2025
Can the training of Z-Image-Turbo be included? Open

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

modelscope/DiffSynth-Studio#1085 · 2 comments · 1 reaction · Opened Nov 28, 2025
按照Wan2.1的推理手册运行出现问题 Open

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

modelscope/DiffSynth-Studio#1080 · 1 comment · 1 reaction · Opened Nov 26, 2025
When will Flux2 training be supported? Open

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

modelscope/DiffSynth-Studio#1079 · 7 comments · 2 reactions · Opened Nov 26, 2025
本repo的特点咨询 Open

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

modelscope/DiffSynth-Studio#1078 · Opened Nov 25, 2025
训练的时候报错 the size of tensor a (4) must match the size of tensor b (5) at non-singleton dimension 2 Open

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

modelscope/DiffSynth-Studio#1077 · Opened Nov 25, 2025
TypeError: WanS2VModel.patchify() takes 2 positional arguments but 3 were given Open

Difficulty 3/5 1-2 days Newbie friendliness 43/100

modelscope/DiffSynth-Studio#1076 · 1 comment · 1 reaction · Opened Nov 25, 2025
wan2.2-14b 如何用多卡推理？ Open

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

modelscope/DiffSynth-Studio#1075 · 4 comments · Opened Nov 25, 2025
wan2.2-vace模型decoder部分会oom Open

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

modelscope/DiffSynth-Studio#1074 · Opened Nov 25, 2025
关于如何修改模型结构的问题 Open

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

modelscope/DiffSynth-Studio#1073 · Opened Nov 25, 2025
how should I set it up to train and save the model checkpoint of the module I added? Open

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

modelscope/DiffSynth-Studio#1072 · Opened Nov 24, 2025
旧版本框架代码支持wan2.2系列吗？ Open

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

modelscope/DiffSynth-Studio#1071 · Opened Nov 24, 2025
[feature request] add pre-calculating latent / text encoder outputs Open

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

modelscope/DiffSynth-Studio#1070 · 2 comments · 1 reaction · Opened Nov 22, 2025
QwenImageVAE tiled decode未实现？ Open

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

modelscope/DiffSynth-Studio#1069 · Opened Nov 21, 2025
How to load Qwen-Image-Lightning's acceleration lora in qwen-image-eidt-2509? Open

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

modelscope/DiffSynth-Studio#1068 · 10 comments · Opened Nov 21, 2025
LoRA trained on Wan2.1-I2V-14B-480P does not work Open

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

modelscope/DiffSynth-Studio#1066 · 2 comments · Opened Nov 20, 2025
Ran Wan2.1 VACE1.3B lora finetuning but have weird result Open

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

modelscope/DiffSynth-Studio#1064 · 1 comment · Opened Nov 20, 2025
wan_video_dit.py中的patchify函数是错误实现 Open

Difficulty 3/5 1-2 days Newbie friendliness 50/100

modelscope/DiffSynth-Studio#1063 · 5 reactions · Opened Nov 19, 2025
The full-parameter training H20 of Flux Konetxt shows OOM Open

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

modelscope/DiffSynth-Studio#1062 · Opened Nov 19, 2025
有关训练集的疑问：使用qwen-image-edit lora微调时，我的输入图片和目标图片，除了要编辑的部分，其他部分必须保持一样吗？ Open

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

modelscope/DiffSynth-Studio#1060 · 1 comment · Opened Nov 18, 2025
After training DiffSynth-Studio/Qwen-Image-Distill-LoRA on the data, the inferred images are blurry. What could be the reason? Open

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

modelscope/DiffSynth-Studio#1059 · 7 comments · Opened Nov 18, 2025
OOM when LORA training Wan2.2-I2V-A14B with 2-H20 Open

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

modelscope/DiffSynth-Studio#1056 · Opened Nov 17, 2025
H100 GPU - FP8 inference acceleration Open

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

modelscope/DiffSynth-Studio#1055 · 1 comment · 1 reaction · Opened Nov 17, 2025
为什么max_timestep_boundary和switch_DiT_boundary不同？ Open

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

modelscope/DiffSynth-Studio#1052 · 4 reactions · Opened Nov 13, 2025
[Feature Request] Add reinforcement learning (RL) training support for Qwen-Image-Edit-2509 in DiffSynth. Open

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

modelscope/DiffSynth-Studio#1051 · 2 comments · 1 reaction · Opened Nov 12, 2025
wan2.1i2v 14b lora training loss Open

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

modelscope/DiffSynth-Studio#1050 · 2 reactions · Opened Nov 12, 2025
Low GPU-Util and Save Promblem During Full training on Qwen-Image-Edit-2509 Open

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

modelscope/DiffSynth-Studio#1049 · 2 comments · Opened Nov 11, 2025
关于 Wan2.1-Fun-V1.1-14B-Control-Camera 的 LoRA 训练 Open

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

modelscope/DiffSynth-Studio#1047 · 2 comments · Opened Nov 10, 2025
When using gradient accumulation, does the order of optimizer.zero_grad() affect training? Open

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

modelscope/DiffSynth-Studio#1046 · 2 comments · Opened Nov 10, 2025
CPU memory explosion during LoRA training — all training processes duplicate full model weights (no shared memory / lazy loading) Open

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

modelscope/DiffSynth-Studio#1043 · 1 comment · Opened Nov 6, 2025
qwen image lora training problem. Open

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

modelscope/DiffSynth-Studio#1040 · 1 comment · Opened Nov 5, 2025
Wan 2.2 TI2V lora训练速度 Open

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

modelscope/DiffSynth-Studio#1039 · 4 comments · Opened Nov 5, 2025
什么时候支持Wan2.2-Fun-5B-Control微调？ Open

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

modelscope/DiffSynth-Studio#1038 · 1 comment · Opened Nov 5, 2025
为什么使用qwen-image-edit微调时不收敛呢 Open

Difficulty 4/5 3-5 days Newbie friendliness 20/100

modelscope/DiffSynth-Studio#1037 · 1 comment · 1 reaction · Opened Nov 5, 2025
Request the support for FastVideo Open

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

modelscope/DiffSynth-Studio#1036 · 2 comments · Opened Nov 4, 2025
Wan2.2 animate inference errrors Open

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

modelscope/DiffSynth-Studio#1035 · 2 comments · Opened Nov 4, 2025
SDXL 训练完成后输出lora权重，如何与原始模型进行lora合并，有没有相关的示例代码 Open

Difficulty 4/5 3-5 days Newbie friendliness 20/100

modelscope/DiffSynth-Studio#1033 · 3 comments · Opened Nov 4, 2025
Question about krea realtime Open

Difficulty 2/5 1-3 hours Newbie friendliness 20/100

modelscope/DiffSynth-Studio#1031 · 2 comments · Opened Nov 3, 2025
I have published a very detailed Qwen Image models training video for average technical people on Windows Open

Difficulty 1/5 Under an hour Newbie friendliness 20/100

modelscope/DiffSynth-Studio#1029 · 4 comments · 1 reaction · Opened Nov 3, 2025
why drop last 4 frames in wan-animate training? Open

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

modelscope/DiffSynth-Studio#1027 · 7 comments · Opened Nov 3, 2025
Distill Open

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

modelscope/DiffSynth-Studio#1023 · 2 comments · Opened Oct 31, 2025

modelscope/DiffSynth-Studio

Get new issues in your inbox