repo stringclasses 1
value | github_id int64 1.27B 4.67B | github_node_id stringlengths 18 24 | number int64 8 14k | html_url stringlengths 49 53 | api_url stringlengths 59 63 | title stringlengths 1 402 | body stringlengths 1 62.9k ⌀ | state stringclasses 2
values | state_reason stringclasses 4
values | locked bool 2
classes | comments_count int64 0 99 | labels listlengths 0 5 | assignees listlengths 0 5 | created_at stringdate 2022-06-09 16:28:35 2026-06-15 20:11:23 | updated_at stringdate 2022-06-12 22:18:01 2026-06-16 23:50:48 | closed_at stringdate 2022-06-12 22:18:01 2026-06-16 00:53:57 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 80
values | extracted_at stringdate 2026-04-07 13:34:13 2026-06-17 00:35:21 | author_login stringlengths 3 28 | author_id int64 1.54k 290M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 2,457,640,170 | I_kwDOHa8MBc6SfJzq | 9,141 | https://github.com/huggingface/diffusers/issues/9141 | https://api.github.com/repos/huggingface/diffusers/issues/9141 | diffusers.FluxPipeline has no .from_single_file() function | ### Describe the bug
At the moment one cannot use a **local** Flux1-dev or -schnell file for lack of an essential function "from_single_file()" in the diffusers.FluxPipeline library. The provided flag "local_files_only" for some reason doesn't solve this problem.
### Reproduction
```
pipe = FluxPipeline.from_single... | closed | completed | false | 16 | [
"bug"
] | [] | 2024-08-09T10:43:23Z | 2024-11-18T20:14:54Z | 2024-11-18T20:14:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lutusp | 15,413,489 | MDQ6VXNlcjE1NDEzNDg5 | User | false |
huggingface/diffusers | 2,457,745,797 | I_kwDOHa8MBc6SfjmF | 9,142 | https://github.com/huggingface/diffusers/issues/9142 | https://api.github.com/repos/huggingface/diffusers/issues/9142 | AttributeError: 'list' object has no attribute '__module__' :hunyuan dit cannot load multiple controlnets | ### Describe the bug
Can not use multiple controlnets, https://huggingface.co/docs/diffusers/main/en/api/pipelines/controlnet_hunyuandit. According to the docs above, it seems that I just need to use a list of controlnets to init the pipeline.
<img width="1049" alt="image" src="https://github.com/user-attachments... | closed | completed | false | 4 | [
"bug",
"Good second issue",
"contributions-welcome"
] | [] | 2024-08-09T11:49:07Z | 2025-05-15T04:39:59Z | 2025-05-15T04:39:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chuck-ma | 74,402,255 | MDQ6VXNlcjc0NDAyMjU1 | User | false |
huggingface/diffusers | 2,458,820,974 | I_kwDOHa8MBc6SjqFu | 9,145 | https://github.com/huggingface/diffusers/issues/9145 | https://api.github.com/repos/huggingface/diffusers/issues/9145 | @tin2tin | @tin2tin
I think we updated the doc for this recently - happy to accept a PR if you spotted places that we missed:)
_Originally posted by @yiyixuxu in https://github.com/huggingface/diffusers/issues/6869#issuecomment-1934770937_
Hey, just as an FYI, the erroneous code is st... | closed | completed | false | 2 | [
"stale"
] | [] | 2024-08-10T00:33:41Z | 2024-10-31T19:12:45Z | 2024-10-31T19:12:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ktchani | 37,028,292 | MDQ6VXNlcjM3MDI4Mjky | User | false |
huggingface/diffusers | 2,459,513,923 | I_kwDOHa8MBc6SmTRD | 9,147 | https://github.com/huggingface/diffusers/issues/9147 | https://api.github.com/repos/huggingface/diffusers/issues/9147 | autopipeline for flux error | ### Describe the bug
autopipeline gives error for flux
### Reproduction
```python
from diffusers import AutoPipelineForText2Image
import torch
pipe_txt2img = AutoPipelineForText2Image.from_pretrained("black-forest-labs/FLUX.1-dev", token="hf_vwWJsiLMb",torch_dtype=torch.float16, use_safetensors=True).to("cuda... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-08-11T09:44:12Z | 2024-09-18T00:56:13Z | 2024-09-17T20:22:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AbhinavJangra29 | 107,471,490 | U_kgDOBmfigg | User | false |
huggingface/diffusers | 2,459,633,330 | I_kwDOHa8MBc6Smway | 9,148 | https://github.com/huggingface/diffusers/issues/9148 | https://api.github.com/repos/huggingface/diffusers/issues/9148 | Add Flux inpainting pipeline | ### Model/Pipeline/Scheduler description
There is a lot of discussion and colabs here: https://old.reddit.com/r/StableDiffusion/comments/1epi1d0/flux_better_than_photoshop_for_inpainting/
### Open source status
- [X] The model implementation is available.
- [X] The model weights are available (Only relevant if addit... | closed | completed | false | 4 | [] | [] | 2024-08-11T15:09:04Z | 2024-09-07T15:52:27Z | 2024-09-07T15:52:27Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 2,459,692,310 | I_kwDOHa8MBc6Sm-0W | 9,149 | https://github.com/huggingface/diffusers/issues/9149 | https://api.github.com/repos/huggingface/diffusers/issues/9149 | Support FLUX nf4 & pf8 for GPUs with 6GB/8GB VRAM (method and checkpoints by lllyasviel) | https://github.com/lllyasviel/stable-diffusion-webui-forge/discussions/981
> Flux Checkpoints
> The currently supported Flux checkpoints are
>
> [flux1-dev-bnb-nf4.safetensors](https://huggingface.co/lllyasviel/flux1-dev-bnb-nf4/blob/main/flux1-dev-bnb-nf4.safetensors) Full flux-dev checkpoint with main model in... | closed | completed | false | 16 | [
"stale"
] | [] | 2024-08-11T17:43:44Z | 2024-10-23T18:53:15Z | 2024-10-23T18:53:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 1,415,305,255 | I_kwDOHa8MBc5UW9gn | 915 | https://github.com/huggingface/diffusers/issues/915 | https://api.github.com/repos/huggingface/diffusers/issues/915 | Improve docs | There are still plenty of outdated, non-existing docs.
E.g.: - https://huggingface.co/docs/diffusers/using-diffusers/loading | closed | completed | false | 2 | [
"stale"
] | [] | 2022-10-19T16:55:39Z | 2022-11-27T15:03:01Z | 2022-11-27T15:03:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 2,459,916,328 | I_kwDOHa8MBc6Sn1go | 9,151 | https://github.com/huggingface/diffusers/issues/9151 | https://api.github.com/repos/huggingface/diffusers/issues/9151 | please add flux img2img with controlnet pipeline | please add flux img2img with controlnet in a single pipeline | closed | completed | false | 4 | [
"stale"
] | [] | 2024-08-12T02:24:10Z | 2024-10-21T15:53:51Z | 2024-10-21T15:53:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | loboere | 15,265,895 | MDQ6VXNlcjE1MjY1ODk1 | User | false |
huggingface/diffusers | 2,460,285,245 | I_kwDOHa8MBc6SpPk9 | 9,154 | https://github.com/huggingface/diffusers/issues/9154 | https://api.github.com/repos/huggingface/diffusers/issues/9154 | Loading PT Files for VAE with the AutoEncoder is still broken!!! | ### Describe the bug
Just doesn't load pt files anymore. Really frustrating as it's been broken for a long time now. I keep posting about it, so now I'll just open a issue / bug instead of messaging in update threads. Last working version is 0.27.2
There is not a working safetensors or diffusers version of the VAE ... | closed | completed | false | 15 | [
"bug",
"stale"
] | [] | 2024-08-12T07:53:12Z | 2025-04-24T05:00:55Z | 2025-01-12T06:13:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JemiloII | 5,192,788 | MDQ6VXNlcjUxOTI3ODg= | User | false |
huggingface/diffusers | 2,460,390,755 | I_kwDOHa8MBc6SppVj | 9,156 | https://github.com/huggingface/diffusers/issues/9156 | https://api.github.com/repos/huggingface/diffusers/issues/9156 | CudaOutOfMemory on Flux Lora Training | ### Describe the bug
I tried to train the flux-dev model with Lora on A100 40GB. But it raises the CudaOutOfMemory exception.
### Reproduction
```
# Accelerate command
export MODEL_NAME="black-forest-labs/FLUX.1-dev"
export INSTANCE_DIR="woman"
export OUTPUT_DIR="trained-flux-lora-woman"
accelerate laun... | closed | completed | false | 13 | [
"bug",
"stale"
] | [] | 2024-08-12T08:46:24Z | 2024-12-15T11:06:00Z | 2024-12-15T11:05:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | m-pektas | 15,729,405 | MDQ6VXNlcjE1NzI5NDA1 | User | false |
huggingface/diffusers | 2,460,640,956 | I_kwDOHa8MBc6Sqma8 | 9,158 | https://github.com/huggingface/diffusers/issues/9158 | https://api.github.com/repos/huggingface/diffusers/issues/9158 | Flux with controlnet and image2image support | **Is your feature request related to a problem? Please describe.**
I'm always frustrated when I can't leverage the powerful capabilities of Flux ControlNet with the img2img pipeline in the diffusers library. Flux ControlNet has recently emerged as a potential new state-of-the-art model for img2img tasks, and the lack ... | closed | completed | false | 9 | [] | [] | 2024-08-12T10:40:18Z | 2024-10-04T15:25:59Z | 2024-10-04T15:25:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | BasimBashir | 45,428,008 | MDQ6VXNlcjQ1NDI4MDA4 | User | false |
huggingface/diffusers | 2,461,912,653 | I_kwDOHa8MBc6Svc5N | 9,161 | https://github.com/huggingface/diffusers/issues/9161 | https://api.github.com/repos/huggingface/diffusers/issues/9161 | train_dreambooth_lora_flux.py distributed bugs | ### Describe the bug
AttributeError when running model parallel distributed training with accelerate
### Reproduction
```
accelerate launch --config_file train_dreambooth_lora_flux.py
--resolution=1024
--mixed_precision=bf16
--pretrained_model_name_or_path=black-forest-labels/FLUX.1-dev
--num_validation_images=... | open | null | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-12T21:18:32Z | 2024-09-15T15:02:46Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 2,461,998,465 | I_kwDOHa8MBc6Svx2B | 9,163 | https://github.com/huggingface/diffusers/issues/9163 | https://api.github.com/repos/huggingface/diffusers/issues/9163 | more tensor bugs in train_dreambooth_lora_flux.py | ### Describe the bug
after hardcoding a workaround regarding transformer.config.guidance_embeds, more tensor errors appear
### Reproduction
```accelerate launch --config_file train_dreambooth_lora_flux.py
--resolution=1024
--mixed_precision=bf16
--pretrained_model_name_or_path=black-forest-labels/FLUX.1-dev
--nu... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-12T22:29:37Z | 2025-01-12T06:08:47Z | 2025-01-12T06:08:47Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 2,462,288,811 | I_kwDOHa8MBc6Sw4ur | 9,164 | https://github.com/huggingface/diffusers/issues/9164 | https://api.github.com/repos/huggingface/diffusers/issues/9164 | the dog example of train_dreambooth_lora_flux.py can not convergence | ### Describe the bug
```
export MODEL_NAME="black-forest-labs/FLUX.1-dev"
export INSTANCE_DIR="dog"
export OUTPUT_DIR="trained-flux-lora"
accelerate launch train_dreambooth_lora_flux.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--instance_data_dir=$INSTANCE_DIR \
--output_dir=$OUTPUT_DIR \
-... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-08-13T03:08:10Z | 2024-08-13T10:23:23Z | 2024-08-13T10:23:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chongxian | 18,145,013 | MDQ6VXNlcjE4MTQ1MDEz | User | false |
huggingface/diffusers | 2,462,431,761 | I_kwDOHa8MBc6SxboR | 9,165 | https://github.com/huggingface/diffusers/issues/9165 | https://api.github.com/repos/huggingface/diffusers/issues/9165 | NF4 Flux params in diffusers | @SunMarc
Since the Flux params are quite huge (if we include the text encoder2, autoencoder, and the diffusion model itself) -- it totals to more than 30GB.
https://huggingface.co/lllyasviel/flux1-dev-bnb-nf4 ships a single safetensors file that has the diffusion model in NF4.
Now, I was able to get this co... | closed | completed | false | 63 | [] | [] | 2024-08-13T05:41:49Z | 2025-03-03T13:29:22Z | 2024-08-15T05:28:40Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,463,164,400 | I_kwDOHa8MBc6S0Ofw | 9,166 | https://github.com/huggingface/diffusers/issues/9166 | https://api.github.com/repos/huggingface/diffusers/issues/9166 | Unexpected behavior of `image_processor.apply_overlay` | ### Describe the bug
The `apply_overlay` method of `image_processor` works unexpectedly if `init_image` and `image` have different shapes. For example it can appear using `StableDiffusionXLInpaintPipeline` with `padding_mask_crop` value filled.
According to `apply_overlay` code, it resizes `init_image` to the `imag... | open | null | false | 12 | [
"bug",
"stale"
] | [
"asomoza"
] | 2024-08-13T11:55:14Z | 2024-11-09T15:03:54Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cortwave | 8,936,357 | MDQ6VXNlcjg5MzYzNTc= | User | false |
huggingface/diffusers | 2,463,183,161 | I_kwDOHa8MBc6S0TE5 | 9,167 | https://github.com/huggingface/diffusers/issues/9167 | https://api.github.com/repos/huggingface/diffusers/issues/9167 | Cannot import nameStableDiffusion3Pipeline | ### Describe the bug
Hello,
I'm getting this error when try to launch the demo-code from the 'stabilityai/stable-diffusion-3-medium-diffusers' model:
_ImportError: cannot import name 'StableDiffusion3Pipeline' from 'diffusers'_
Just installed the latest version of diffusers via `pip3 install git+https://gith... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-08-13T12:05:01Z | 2024-08-13T15:29:39Z | 2024-08-13T15:29:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mmxhello | 60,787,898 | MDQ6VXNlcjYwNzg3ODk4 | User | false |
huggingface/diffusers | 1,415,856,990 | I_kwDOHa8MBc5UZENe | 917 | https://github.com/huggingface/diffusers/issues/917 | https://api.github.com/repos/huggingface/diffusers/issues/917 | onnx inpainting error | ### Describe the bug
With the latest code, I was able to convert the SD1.4 checkpoint into onnx and successfully run txt2img and img2img using the new onnx pipelines. However the onnx inpainting isn't working.
Thank you!
### Reproduction
```
from diffusers import OnnxStableDiffusionInpaintPipeline
import ... | closed | completed | false | 6 | [
"bug"
] | [
"anton-l"
] | 2022-10-20T03:02:10Z | 2022-11-07T22:54:57Z | 2022-11-07T14:46:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pythoninoffice | 90,436,829 | MDQ6VXNlcjkwNDM2ODI5 | User | false |
huggingface/diffusers | 2,464,719,520 | I_kwDOHa8MBc6S6KKg | 9,170 | https://github.com/huggingface/diffusers/issues/9170 | https://api.github.com/repos/huggingface/diffusers/issues/9170 | sdxl and contronet must has a GPU memory more than 36G? | ### Describe the bug
https://github.com/huggingface/diffusers/blob/15eb77bc4cf2ccb40781cb630b9a734b43cffcb8/src/diffusers/pipelines/controlnet/pipeline_controlnet_sd_xl.py
line73---line113
I run the demo with 24G GPU, then OOM everytime.
so I must run SDXl with 48G?
@yiyixuxu @sayakpaul @DN6 tks
### Reprod... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-14T01:46:35Z | 2024-11-13T08:49:22Z | 2024-08-14T12:13:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | henbucuoshanghai | 37,647,094 | MDQ6VXNlcjM3NjQ3MDk0 | User | false |
huggingface/diffusers | 2,464,801,062 | I_kwDOHa8MBc6S6eEm | 9,171 | https://github.com/huggingface/diffusers/issues/9171 | https://api.github.com/repos/huggingface/diffusers/issues/9171 | [regression] StableDiffusionPipeline.from_single_file() does not handle SD 1.5 models with prediction_type `v_prediction` | ### Describe the bug
There are a few Stable Diffusion 1.5 models that use a prediction type of `v_prediction` rather than `epsilon`. In version 0.27.0, `StableDiffusionPipeline.from_single_file()` correctly detected and rendered images from such models. However, in version 0.30.0, these models are always treated as ... | closed | completed | false | 13 | [
"bug",
"stale"
] | [] | 2024-08-14T03:13:06Z | 2025-01-16T11:41:58Z | 2025-01-16T11:41:57Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lstein | 111,189 | MDQ6VXNlcjExMTE4OQ== | User | false |
huggingface/diffusers | 2,464,959,195 | I_kwDOHa8MBc6S7Erb | 9,172 | https://github.com/huggingface/diffusers/issues/9172 | https://api.github.com/repos/huggingface/diffusers/issues/9172 | why rebuild a vae in inference stage? | Thanks for ur effort for diffusion model.
I want to know why we need to rebuild a vae in inference stage. I think it will introduce extra GPU cost.
https://github.com/huggingface/diffusers/blob/a85b34e7fdc0a5fceb11aa0fa6199bd9afaca396/examples/text_to_image/train_text_to_image_sdxl.py#L1217C16-L1223C24
| open | null | false | 2 | [
"stale"
] | [] | 2024-08-14T05:52:38Z | 2024-11-14T15:03:55Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WilliammmZ | 18,674,350 | MDQ6VXNlcjE4Njc0MzUw | User | false |
huggingface/diffusers | 2,465,169,749 | I_kwDOHa8MBc6S74FV | 9,174 | https://github.com/huggingface/diffusers/issues/9174 | https://api.github.com/repos/huggingface/diffusers/issues/9174 | [Quantization] bring quantization to diffusers core | Now that we have a working PoC (#9165) of NF4 quantization through `bitsandbytes` and also [this](https://huggingface.co/blog/quanto-diffusers) through `optimum.quanto`, it's time to bring in quantization more formally in `diffusers` 🎸
In this issue, I want to devise a rough plan to attack the integration. We are g... | closed | completed | false | 15 | [
"quantization"
] | [
"sayakpaul"
] | 2024-08-14T08:05:34Z | 2024-10-21T04:42:46Z | 2024-10-21T04:42:46Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,465,927,234 | I_kwDOHa8MBc6S-xBC | 9,178 | https://github.com/huggingface/diffusers/issues/9178 | https://api.github.com/repos/huggingface/diffusers/issues/9178 | flux lora train with diffusers peft target modules error | ### Describe the bug
[message (1).txt](https://github.com/user-attachments/files/16613911/message.1.txt)
when i inference with the following code the state_dict value error showed up,some layers not found in the base model,the attached file is error message,i trained with simpletuner project set config parameter --fl... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-14T13:59:36Z | 2024-08-15T07:47:12Z | 2024-08-15T07:46:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | frankchieng | 130,369,523 | U_kgDOB8VH8w | User | false |
huggingface/diffusers | 2,466,012,038 | I_kwDOHa8MBc6S_FuG | 9,179 | https://github.com/huggingface/diffusers/issues/9179 | https://api.github.com/repos/huggingface/diffusers/issues/9179 | Training results do not make sense for sdxl+controlnet | Hi, when i use the latest code to train the SDXL+controlnet using 4xA100. The results seem not my expection.
* training script, the only change is that i download the dataset locally and use 4xA100 GPUs for the training process.
```sh
export MODEL_DIR="/mnt/bn/multimodel/models/t2i/stable-diffusion-xl-base-1.0/"... | closed | completed | false | 8 | [
"stale"
] | [] | 2024-08-14T14:27:55Z | 2024-09-27T06:23:21Z | 2024-09-27T06:00:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | littletomatodonkey | 14,270,174 | MDQ6VXNlcjE0MjcwMTc0 | User | false |
huggingface/diffusers | 1,415,898,592 | I_kwDOHa8MBc5UZOXg | 918 | https://github.com/huggingface/diffusers/issues/918 | https://api.github.com/repos/huggingface/diffusers/issues/918 | --train-text-encoder in train_dreambooth.py yields type error | ### Describe the bug
specifying the --train_text_encoder flag in `train_dreambooth.py` yields an error:
```
RuntimeError: Expected tensor for argument #1 'indices' to have one of the following scalar types: Long, Int; but got torch.cuda.FloatTensor instead (while checking arguments for embedding)
```
Remov... | closed | completed | false | 3 | [
"bug"
] | [
"patil-suraj"
] | 2022-10-20T03:55:13Z | 2022-10-23T17:33:33Z | 2022-10-23T17:33:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MarkRich | 2,020,546 | MDQ6VXNlcjIwMjA1NDY= | User | false |
huggingface/diffusers | 2,466,066,593 | I_kwDOHa8MBc6S_TCh | 9,180 | https://github.com/huggingface/diffusers/issues/9180 | https://api.github.com/repos/huggingface/diffusers/issues/9180 | Pipeline has no attribute '_execution_device' | ### Describe the bug
Hello, I implemented my own custom pipeline referring StableDiffusionPipeline (RepDiffusionPipeline), but there are some issues
I called "accelerator.prepare" properly, and mapped the models on device (with "to.(accelerator.device)")
But when I call pipeline and the '__call__' function is call... | open | reopened | false | 37 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2024-08-14T14:43:15Z | 2026-01-29T10:18:20Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | choidaedae | 105,369,646 | U_kgDOBkfQLg | User | false |
huggingface/diffusers | 2,466,245,259 | I_kwDOHa8MBc6S_-qL | 9,181 | https://github.com/huggingface/diffusers/issues/9181 | https://api.github.com/repos/huggingface/diffusers/issues/9181 | strange noise dot in StableDiffusionControlNetPipeline | ### Describe the bug
Hi team, thanks for your wonderful work.
When I used StableDiffusionControlNetPipeline, I found that some pictures had strange noise, similar to black and white dots.
the input is:
the output is:
![ou... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-14T16:01:24Z | 2024-08-15T06:11:01Z | 2024-08-15T06:11:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sonia-kong | 171,914,194 | U_kgDOCj8z0g | User | false |
huggingface/diffusers | 2,467,474,558 | I_kwDOHa8MBc6TEqx- | 9,184 | https://github.com/huggingface/diffusers/issues/9184 | https://api.github.com/repos/huggingface/diffusers/issues/9184 | What is the correct way to apply the dictionary with the control strengths (called “scales”) but with blocks? | ### Describe the bug
I have managed to apply the basic dictionary. as the documentation mentions
```
adapter_weight_scales = { "unet": { "down": 1, "mid": 0, "up": 0} }
pipe.set_adapters("Lora1", adapter_weight_scales)
```
and it already works for N number of LORAS that I want to load, for example
```
ada... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-15T06:05:42Z | 2024-08-17T00:54:28Z | 2024-08-17T00:54:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Eduardishion | 47,488,420 | MDQ6VXNlcjQ3NDg4NDIw | User | false |
huggingface/diffusers | 2,468,025,484 | I_kwDOHa8MBc6TGxSM | 9,186 | https://github.com/huggingface/diffusers/issues/9186 | https://api.github.com/repos/huggingface/diffusers/issues/9186 | Differential Diffusion with ControlNet pipeline | **Is your feature request related to a problem? Please describe.**
Many users have asked me whether diff diff goes well with ControlNet :
https://github.com/exx8/differential-diffusion/issues/29,
https://github.com/exx8/differential-diffusion/issues/19
It does!
I have conducted an experiment in the past, but it w... | open | null | false | 4 | [
"stale"
] | [
"asomoza"
] | 2024-08-15T13:09:52Z | 2024-11-09T15:03:46Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | exx8 | 8,540,180 | MDQ6VXNlcjg1NDAxODA= | User | false |
huggingface/diffusers | 2,468,438,171 | I_kwDOHa8MBc6TIWCb | 9,187 | https://github.com/huggingface/diffusers/issues/9187 | https://api.github.com/repos/huggingface/diffusers/issues/9187 | IP Adapter is not working with SDXL when using `image_encoder_folder=None` after Kolors update | ### Describe the bug
When loading IP Adapter into SDXL Text2Image pipeline with `image_encoder_folder=None`, I encounter the following error:
```
clip_image_size = self.image_encoder.config.image_size
AttributeError: 'NoneType' object has no attribute 'config'
```
As I can see from the code changes in the `ip_a... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-08-15T16:29:58Z | 2024-08-17T06:00:06Z | 2024-08-17T06:00:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | detkov | 34,102,070 | MDQ6VXNlcjM0MTAyMDcw | User | false |
huggingface/diffusers | 2,469,164,735 | I_kwDOHa8MBc6TLHa_ | 9,188 | https://github.com/huggingface/diffusers/issues/9188 | https://api.github.com/repos/huggingface/diffusers/issues/9188 | Flux Outpainting | **Is your feature request related to a problem? Please describe.**
Outpainting doesn't seem to be supported in Flux
**Describe the solution you'd like.**
Add outpainting to Flux
**Describe alternatives you've considered.**
N/A
**Additional context.**
#9135 introduces Flux inpainting | closed | completed | false | 15 | [
"stale"
] | [] | 2024-08-15T23:50:12Z | 2024-12-09T02:13:59Z | 2024-09-17T21:23:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fakerybakery | 76,186,054 | MDQ6VXNlcjc2MTg2MDU0 | User | false |
huggingface/diffusers | 2,469,329,751 | I_kwDOHa8MBc6TLvtX | 9,190 | https://github.com/huggingface/diffusers/issues/9190 | https://api.github.com/repos/huggingface/diffusers/issues/9190 | Can pipe.load_lora_weights have scale value? For StableDiffusion | As far as I know with Diffusers (0.30.0) latest version we are able to load multiple lora weights for StableDiffusion like
```
pipe.load_lora_weights(lora_path1)
pipe.load_lora_weights(lora_path2)
pipe.load_lora_weights(lora_path3)
```
And it works
And then we can use
`pipe.unload_lora_weights()`
... | closed | completed | false | 2 | [] | [] | 2024-08-16T02:52:12Z | 2024-08-16T12:42:04Z | 2024-08-16T12:42:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | FurkanGozukara | 19,240,467 | MDQ6VXNlcjE5MjQwNDY3 | User | false |
huggingface/diffusers | 2,469,534,431 | I_kwDOHa8MBc6TMhrf | 9,192 | https://github.com/huggingface/diffusers/issues/9192 | https://api.github.com/repos/huggingface/diffusers/issues/9192 | load_ip_adapter() doesn't respect cache_dir or local_files_only | ### Describe the bug
When calling ``load_ip_adapter()`` with a custom ``cache_dir`` param, if the specified model requires the image encoder (i.e. [ip-adapter-plus_sd15.bin](https://huggingface.co/h94/IP-Adapter/blob/main/models/ip-adapter-plus_sd15.bin)), then diffusers will install the ip adapter model (i.e. ``ip-... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-16T05:48:10Z | 2024-08-26T20:19:13Z | 2024-08-26T19:03:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jonahclarsen | 97,376,233 | U_kgDOBc3X6Q | User | false |
huggingface/diffusers | 2,469,593,009 | I_kwDOHa8MBc6TMv-x | 9,195 | https://github.com/huggingface/diffusers/issues/9195 | https://api.github.com/repos/huggingface/diffusers/issues/9195 | Problem with Flux Schnell bfloat16 multiGPU | ### Describe the bug
Hello! I set device_map='balanced' and get images generated in 2.5 minutes (expected in 12-20 seconds), while in pipe.hf_device_map it shows that the devices are distributed like this:
```
{
"transformer": "cuda:0",
"text_encoder_2": "cuda:2",
"text_encoder": "cuda:0",
"vae": "cuda:1"
... | closed | completed | false | 26 | [
"bug"
] | [] | 2024-08-16T06:30:54Z | 2025-12-05T06:38:14Z | 2024-09-15T15:04:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | OlegRuban-ai | 61,088,564 | MDQ6VXNlcjYxMDg4NTY0 | User | false |
huggingface/diffusers | 2,469,762,442 | I_kwDOHa8MBc6TNZWK | 9,197 | https://github.com/huggingface/diffusers/issues/9197 | https://api.github.com/repos/huggingface/diffusers/issues/9197 | Flux Inference memory optimisation by pre-computing modulation parameters ahead of time | DiffusionKit has an excellent memory optimisation for Flux where it calculates modulation parameters ahead of time
then offloads the adaLN_modulation parameters which for fp16 saves ~ 6.5 GB peak memory usage during inference.
https://github.com/argmaxinc/DiffusionKit/pull/15/
It would be nice if a similar chan... | open | null | false | 3 | [
"stale"
] | [] | 2024-08-16T08:18:43Z | 2024-10-12T15:03:46Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vargol | 62,868 | MDQ6VXNlcjYyODY4 | User | false |
huggingface/diffusers | 1,415,973,632 | I_kwDOHa8MBc5UZgsA | 920 | https://github.com/huggingface/diffusers/issues/920 | https://api.github.com/repos/huggingface/diffusers/issues/920 | Separating legacy inpaint pipeline | The new fine-tuned inpaint pipeline in 0.6.0 release is very good, but its UNet architecture is different from original pipeline.
I think the inpaint pipeline in 0.5.1 release has some advantage such as sharing the same model with txt2img on GPU or inheriting DreamBooth UNet trained on txt2img model.
I actually often... | closed | completed | false | 26 | [
"stale"
] | [] | 2022-10-20T05:22:44Z | 2023-08-18T12:24:52Z | 2022-12-24T15:03:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | juno-hwang | 52,326,857 | MDQ6VXNlcjUyMzI2ODU3 | User | false |
huggingface/diffusers | 2,469,919,997 | I_kwDOHa8MBc6TN_z9 | 9,200 | https://github.com/huggingface/diffusers/issues/9200 | https://api.github.com/repos/huggingface/diffusers/issues/9200 | diffusers 0.30.0 StableDiffusionXLControlNetPipeline dtype error | ### Describe the bug
**Description:**
I am defining two `ControlNetModel` instances using the `diffusers` library and creating a pipeline using `StableDiffusionXLControlNetPipeline`. When I try to generate with this pipeline, I encounter an error in `diffusers` version 0.30.0, although it works fine in version 0.27... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-16T09:52:04Z | 2025-02-25T14:46:55Z | 2024-10-07T07:01:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yagmurx | 46,970,065 | MDQ6VXNlcjQ2OTcwMDY1 | User | false |
huggingface/diffusers | 2,470,046,974 | I_kwDOHa8MBc6TOez- | 9,202 | https://github.com/huggingface/diffusers/issues/9202 | https://api.github.com/repos/huggingface/diffusers/issues/9202 | `StableDiffusionXLPAGPipeline` does not work with FaceID Plus V2 IP-Adapter | ### Describe the bug
When switching from `StableDiffusionXLPipeline` to it's PAG version `StableDiffusionXLPAGPipeline`, the workflow using an additional IP-Adapter (ip-adapter-faceid-plusv2_sdxl) starts throwing Unet-related issue with some shapes mismatching.
### Reproduction
Work fine with raw SDXL:
```python
i... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-16T11:08:54Z | 2024-08-22T09:49:25Z | 2024-08-22T09:49:25Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | detkov | 34,102,070 | MDQ6VXNlcjM0MTAyMDcw | User | false |
huggingface/diffusers | 2,470,515,161 | I_kwDOHa8MBc6TQRHZ | 9,204 | https://github.com/huggingface/diffusers/issues/9204 | https://api.github.com/repos/huggingface/diffusers/issues/9204 | DreamBooth training script / FLUX.1 [dev] | ### Describe the bug
Thanks @bghira @sayakpaul for sharing the FLUX DreamBooth training scripts at https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README_flux.md
I was trying to run the examples and also train on my own data (I have successfully used it to get fine-tunes using SD2.x, SDXL, ... | closed | completed | false | 26 | [
"bug"
] | [
"sayakpaul"
] | 2024-08-16T15:43:37Z | 2024-08-26T19:23:45Z | 2024-08-26T11:59:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lamm-mit | 101,393,859 | U_kgDOBgslww | User | false |
huggingface/diffusers | 2,470,670,723 | I_kwDOHa8MBc6TQ3GD | 9,205 | https://github.com/huggingface/diffusers/issues/9205 | https://api.github.com/repos/huggingface/diffusers/issues/9205 | Can we pass output_attentions=True to DiT model such as pixart to get attention output? | Can we pass output_attentions=True to DiT model such as pixart to get attention output? Like using output_attentions=True in transformer? | open | null | false | 1 | [
"stale"
] | [] | 2024-08-16T17:26:14Z | 2024-09-16T15:02:42Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | foreverpiano | 29,483,524 | MDQ6VXNlcjI5NDgzNTI0 | User | false |
huggingface/diffusers | 2,471,623,878 | I_kwDOHa8MBc6TUfzG | 9,208 | https://github.com/huggingface/diffusers/issues/9208 | https://api.github.com/repos/huggingface/diffusers/issues/9208 | Cannot load most controlnet models using from_single_file() | ### Describe the bug
The controlnet models found at https://huggingface.co/lllyasviel/sd_control_collection fail to load with `from_single_file()`. The error message is `OSError: runwayml/stable-diffusion-v1-5 does not appear to have a file named config.json.`
Maybe these models are not intended to be loaded with... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-17T18:29:10Z | 2024-08-31T17:47:56Z | 2024-08-31T17:47:56Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lstein | 111,189 | MDQ6VXNlcjExMTE4OQ== | User | false |
huggingface/diffusers | 2,471,638,737 | I_kwDOHa8MBc6TUjbR | 9,209 | https://github.com/huggingface/diffusers/issues/9209 | https://api.github.com/repos/huggingface/diffusers/issues/9209 | We couldn't connect to '"https://hf-mirror.com"' to load this model | ### Describe the bug
env: MODEL_NAME="runwayml/stable-diffusion-v1-5"
env: DATASET_NAME="/content/drive/MyDrive/Newfolder"
env: HF_ENDPOINT="https://hf-mirror.com/"
env: TRANSFORMERS_OFFLINE=0
2024-08-17 19:12:18.804427: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factor... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-08-17T19:18:16Z | 2024-08-18T08:58:16Z | 2024-08-18T08:58:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SoumyaMB10 | 174,590,283 | U_kgDOCmgJSw | User | false |
huggingface/diffusers | 2,471,866,449 | I_kwDOHa8MBc6TVbBR | 9,210 | https://github.com/huggingface/diffusers/issues/9210 | https://api.github.com/repos/huggingface/diffusers/issues/9210 | ValueError: --image_column' value 'image' needs to be one of: text | ### Describe the bug
env: MODEL_NAME=runwayml/stable-diffusion-v1-5
env: INSTANCE_DIR=/content/drive/MyDrive/Newfolder
env: HF_ENDPOINT=https://hf-mirror.com/
2024-08-18 08:46:08.308678: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory f... | open | null | false | 26 | [
"bug"
] | [] | 2024-08-18T08:57:17Z | 2025-04-18T03:02:35Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SoumyaMB10 | 174,590,283 | U_kgDOCmgJSw | User | false |
huggingface/diffusers | 2,473,002,446 | I_kwDOHa8MBc6TZwXO | 9,215 | https://github.com/huggingface/diffusers/issues/9215 | https://api.github.com/repos/huggingface/diffusers/issues/9215 | error of num_images_per_prompt in the flux pipeline | in the `_get_clip_prompt_embeds` in flux pipeline, it is:
`
prompt_embeds = self.text_encoder(text_input_ids.to(device), output_hidden_states=False)
# Use pooled output of CLIPTextModel
prompt_embeds = prompt_embeds.pooler_output
prompt_embeds = prompt_embeds.to(dtype=self.text_en... | closed | completed | false | 6 | [
"stale"
] | [] | 2024-08-19T10:21:24Z | 2024-09-21T16:00:42Z | 2024-09-21T16:00:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rardz | 4,254,977 | MDQ6VXNlcjQyNTQ5Nzc= | User | false |
huggingface/diffusers | 2,473,160,641 | I_kwDOHa8MBc6TaW_B | 9,216 | https://github.com/huggingface/diffusers/issues/9216 | https://api.github.com/repos/huggingface/diffusers/issues/9216 | I made a pipeline that lets you use any number of models at once | ### Model/Pipeline/Scheduler description
Here's how to do it:
from rubberDiffusers import StableDiffusionRubberPipeline
pipe=StableDiffusionRubberPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5", torch_dtype=torch.float32,local_files_only=True,safety_checker=None, requires_safety_checker=False,
)
... | open | null | false | 3 | [
"stale"
] | [] | 2024-08-19T11:46:08Z | 2024-09-21T15:03:31Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 2,474,025,146 | I_kwDOHa8MBc6TdqC6 | 9,218 | https://github.com/huggingface/diffusers/issues/9218 | https://api.github.com/repos/huggingface/diffusers/issues/9218 | PAG with guess mode | ### Describe the bug
To me, this [code block](https://github.com/huggingface/diffusers/blob/67f5cce2943e0f32f4e0c6b53177dc3107a7955f/src/diffusers/pipelines/pag/pipeline_pag_controlnet_sd.py#L1195-L1205) seems to ignore the case `guess_mode=False`. Before that, only in the case of `self.do_classifier_free_guidance and... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2024-08-19T19:16:04Z | 2024-10-14T18:21:07Z | 2024-10-14T18:21:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,474,280,124 | I_kwDOHa8MBc6TeoS8 | 9,220 | https://github.com/huggingface/diffusers/issues/9220 | https://api.github.com/repos/huggingface/diffusers/issues/9220 | undefined variable in PAG pipeline_pag_sd | ### Describe the bug
`noise_pred_text` is not defined [here ](https://github.com/huggingface/diffusers/blob/v0.30.0/src/diffusers/pipelines/pag/pipeline_pag_sd.py#L1006C64-L1006C79) in the case of `do_perturbed_attention_guidance=True`, `guidance_rescale >0` and `do_classifier_free_guidance=True`
### Reproduction... | closed | completed | false | 7 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2024-08-19T21:55:34Z | 2024-10-14T21:46:31Z | 2024-10-14T21:46:31Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,474,516,231 | I_kwDOHa8MBc6Tfh8H | 9,221 | https://github.com/huggingface/diffusers/issues/9221 | https://api.github.com/repos/huggingface/diffusers/issues/9221 | kohya produced loras -> diffusers supported format | enables lora loading onto diffuser's FluxPipeline from loras produced by kohya training scripts | closed | completed | false | 2 | [
"stale"
] | [] | 2024-08-20T01:01:29Z | 2024-09-19T23:10:56Z | 2024-09-19T23:10:56Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 2,474,936,237 | I_kwDOHa8MBc6ThIet | 9,223 | https://github.com/huggingface/diffusers/issues/9223 | https://api.github.com/repos/huggingface/diffusers/issues/9223 | cudf-cu12 24.4.1, ibis-framework 8.0.0 requires pyarrow<15.0.0a0,>=14.0.1,pyarrow<16,>=2 and datasets 2.21.0 requires pyarrow>=15.0.0 | ### Describe the bug
!pip install accelerate>=0.16.0 torchvision transformers>=4.25.1 datasets>=2.19.1 ftfy tensorboard Jinja2 peft==0.7.0
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
c... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-08-20T07:38:45Z | 2024-08-24T13:55:22Z | 2024-08-20T07:49:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SoumyaMB10 | 174,590,283 | U_kgDOCmgJSw | User | false |
huggingface/diffusers | 2,474,960,892 | I_kwDOHa8MBc6ThOf8 | 9,224 | https://github.com/huggingface/diffusers/issues/9224 | https://api.github.com/repos/huggingface/diffusers/issues/9224 | hardcoded torch.float64 isn't supported on Metal (device="mps") | ### Describe the bug
Running transformer_flux on macOS device="mps" fails with error:
`Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead`
### Reproduction
Run on Mac:
```import torch
from diffusers import FluxPipeline
def print_hi(name):
... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2024-08-20T07:51:16Z | 2024-09-19T15:39:03Z | 2024-09-19T15:39:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | RoyLeviLangware | 163,380,355 | U_kgDOCbz8gw | User | false |
huggingface/diffusers | 2,475,281,900 | I_kwDOHa8MBc6Tic3s | 9,225 | https://github.com/huggingface/diffusers/issues/9225 | https://api.github.com/repos/huggingface/diffusers/issues/9225 | mismatching size of latents in StableDiffusionXLInstructPix2PixPipeline with rgba images | ### Describe the bug
there is size mismatch between latents and latens reconstructed from the input image
```
Traceback (most recent call last):
File "/home/imgen/projects/metafusion/examples/ip2p.py", line 20, in <module>
images = pipe(prompt, image=image,
File "/home/imgen/miniconda3/envs/py31/lib/p... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2024-08-20T10:25:03Z | 2024-09-19T17:18:55Z | 2024-09-19T17:18:54Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | noskill | 733,626 | MDQ6VXNlcjczMzYyNg== | User | false |
huggingface/diffusers | 2,475,373,640 | I_kwDOHa8MBc6TizRI | 9,227 | https://github.com/huggingface/diffusers/issues/9227 | https://api.github.com/repos/huggingface/diffusers/issues/9227 | Problem running train_dreambooth_lora_flux.py with model "black-forest-labs/FLUX.1-dev" | ### Describe the bug
I am using the training script documented here https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README_flux.md to train a LORA on my dataset.
here is the command I am running:
accelerate launch train_dreambooth_lora_flux.py
--pretrained_model_name_or_path="black-for... | open | null | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-20T11:11:14Z | 2024-12-01T15:04:10Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lizi85 | 1,999,292 | MDQ6VXNlcjE5OTkyOTI= | User | false |
huggingface/diffusers | 2,476,976,430 | I_kwDOHa8MBc6To6ku | 9,235 | https://github.com/huggingface/diffusers/issues/9235 | https://api.github.com/repos/huggingface/diffusers/issues/9235 | Is there any way to get diffusers-v0.27.0.dev0? | Is there any way to get diffusers-v0.27.0.dev0? I want to compare the difference between diffusers-v0.27.0.dev0 and branches that develop on it in another project, but I didn't find it on the releases or tags page. | closed | completed | false | 2 | [] | [] | 2024-08-21T03:42:11Z | 2024-08-21T05:10:26Z | 2024-08-21T05:10:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | D222097 | 50,061,868 | MDQ6VXNlcjUwMDYxODY4 | User | false |
huggingface/diffusers | 2,477,887,320 | I_kwDOHa8MBc6TsY9Y | 9,237 | https://github.com/huggingface/diffusers/issues/9237 | https://api.github.com/repos/huggingface/diffusers/issues/9237 | Issue on flux dreambooth lora training | ### Describe the bug
`RuntimeError: Input type (float) and bias type (c10::Half) should be the same`
whenever I follow the readme of flux https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/README_flux.md
I tried to do exactly it is written but still error
I tried the SDXL version it work... | closed | completed | false | 17 | [
"bug"
] | [
"linoytsaban"
] | 2024-08-21T12:14:04Z | 2024-09-29T02:05:58Z | 2024-09-14T10:59:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | arvnoodle | 76,831,102 | MDQ6VXNlcjc2ODMxMTAy | User | false |
huggingface/diffusers | 2,479,632,393 | I_kwDOHa8MBc6TzDAJ | 9,239 | https://github.com/huggingface/diffusers/issues/9239 | https://api.github.com/repos/huggingface/diffusers/issues/9239 | Generate black image when using ControlNet-Canny | ### Describe the bug
Hi,
I test SD1.5 with sd-controlnet-canny, but always generate black image.
The models I used:
ControlNet model: https://huggingface.co/lllyasviel/sd-controlnet-canny
SD1.5 model: https://huggingface.co/runwayml/stable-diffusion-v1-5
Thanks a lot for the help!
### Reproduction
F... | closed | completed | false | 14 | [
"bug"
] | [] | 2024-08-22T02:24:55Z | 2024-08-27T05:37:09Z | 2024-08-27T05:37:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xduzhangjiayu | 73,270,275 | MDQ6VXNlcjczMjcwMjc1 | User | false |
huggingface/diffusers | 2,479,714,885 | I_kwDOHa8MBc6TzXJF | 9,241 | https://github.com/huggingface/diffusers/issues/9241 | https://api.github.com/repos/huggingface/diffusers/issues/9241 | Convert From XLab lora to diffusers lora script | **Is your feature request related to a problem? Please describe.**
I want to use Lora made by XLab. But I can't make it work in Diffusers because there are many differences in Lora structure
**Describe the solution you'd like.**
Make a script so that we can convert Lora structures easily from XLAB to Diffusers form... | closed | completed | false | 14 | [] | [] | 2024-08-22T04:03:15Z | 2024-08-29T03:05:08Z | 2024-08-28T10:39:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chuck-ma | 74,402,255 | MDQ6VXNlcjc0NDAyMjU1 | User | false |
huggingface/diffusers | 2,481,925,989 | I_kwDOHa8MBc6T7y9l | 9,248 | https://github.com/huggingface/diffusers/issues/9248 | https://api.github.com/repos/huggingface/diffusers/issues/9248 | pip install --upgrade "diffusers[torch]" | ### Describe the bug
Installation instructions say
`pip install --upgrade diffusers[torch]`
when it should be
`pip install --upgrade "diffusers[torch]"`
### Reproduction
Installation instructions say
`pip install --upgrade diffusers[torch]`
when it should be
`pip install --upgrade "diffusers[torch]"`
### Logs... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-22T22:33:48Z | 2024-09-22T20:20:56Z | 2024-09-22T20:20:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | matheth | 47,308,104 | MDQ6VXNlcjQ3MzA4MTA0 | User | false |
huggingface/diffusers | 2,482,960,341 | I_kwDOHa8MBc6T_vfV | 9,258 | https://github.com/huggingface/diffusers/issues/9258 | https://api.github.com/repos/huggingface/diffusers/issues/9258 | Kohya SS FLUX LoRA training is way faster on Linux than Windows any ideas to debug? Same settings, libraries and GPU | ### Describe the bug
I am using Kohya SS to train FLUX LoRA
On Linux RTX 3090 gets like 5.5 second / it - batch size 1 and 1024x1024 px resolution
On Windows RTX 3090 TI gets 7.7 second / it - has the most powerful CPU 13900 K
This speed dispercany is huge between Windows and Linux for some reason
Torch... | closed | completed | true | 1 | [
"bug"
] | [] | 2024-08-23T11:42:53Z | 2024-08-23T11:55:18Z | 2024-08-23T11:55:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | FurkanGozukara | 19,240,467 | MDQ6VXNlcjE5MjQwNDY3 | User | false |
huggingface/diffusers | 2,483,462,785 | I_kwDOHa8MBc6UBqKB | 9,260 | https://github.com/huggingface/diffusers/issues/9260 | https://api.github.com/repos/huggingface/diffusers/issues/9260 | Storage-Efficient Bulk Conversion of Single-File Model Layout to Diffusers-Multifolder | **Is your feature request related to a problem? Please describe.**
A major drawback of using single-file models is their inefficient use of disk storage, as a user who has downloaded several models in the single-file format is likely storing many redundant copies of individual model components that were re-used across... | closed | completed | true | 6 | [
"stale"
] | [] | 2024-08-23T16:16:44Z | 2024-11-08T18:17:50Z | 2024-11-08T18:17:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wbclark | 7,719,077 | MDQ6VXNlcjc3MTkwNzc= | User | false |
huggingface/diffusers | 2,483,688,963 | I_kwDOHa8MBc6UChYD | 9,261 | https://github.com/huggingface/diffusers/issues/9261 | https://api.github.com/repos/huggingface/diffusers/issues/9261 | "previous_timestep()" in DDPM scheduling not compatible with "trailing" option. DDIM bugged too | ### Describe the bug
I'm currently on a quest to remove all noise from my very noise-sensitive diffusion application, so checking all the equations right now and I think I found one error:
in scheduling_ddmp.py, previous_timesteps() currently assumes two cases
- either you have fully custom timesteps, in which ... | closed | completed | false | 8 | [
"bug",
"scheduler"
] | [] | 2024-08-23T18:48:42Z | 2024-12-03T23:57:53Z | 2024-12-03T23:57:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | RochMollero | 3,669,517 | MDQ6VXNlcjM2Njk1MTc= | User | false |
huggingface/diffusers | 2,484,743,657 | I_kwDOHa8MBc6UGi3p | 9,264 | https://github.com/huggingface/diffusers/issues/9264 | https://api.github.com/repos/huggingface/diffusers/issues/9264 | Could you make an inpainting model for flux? | ### Model/Pipeline/Scheduler description
The [stable-diffusion-xl-1.0-inpainting-0.1](https://huggingface.co/diffusers/stable-diffusion-xl-1.0-inpainting-0.1) model helps a lot. Could you make a similar inpainting model for flux?
https://huggingface.co/black-forest-labs/FLUX.1-dev
### Open source status
- [ ] The... | closed | completed | false | 2 | [] | [] | 2024-08-24T17:32:32Z | 2024-08-24T17:37:59Z | 2024-08-24T17:37:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tiptinker | 32,961,592 | MDQ6VXNlcjMyOTYxNTky | User | false |
huggingface/diffusers | 2,485,150,497 | I_kwDOHa8MBc6UIGMh | 9,267 | https://github.com/huggingface/diffusers/issues/9267 | https://api.github.com/repos/huggingface/diffusers/issues/9267 | AttributeError in `diffusers.models.transformers.transformer_flux` - Missing `rope` Attribute | ### Describe the bug
I encountered an issue while working with the diffusers library where an `AttributeError` is raised because the `transformer_flux` module does not have an `attribute` named `rope`. This problem arises when trying to modify the` rop`e function to handle CUDA devices.
```AttributeError: module ... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-25T09:44:04Z | 2024-09-24T15:16:46Z | 2024-09-24T15:16:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rajveer43 | 64,583,161 | MDQ6VXNlcjY0NTgzMTYx | User | false |
huggingface/diffusers | 1,416,697,893 | I_kwDOHa8MBc5UcRgl | 927 | https://github.com/huggingface/diffusers/issues/927 | https://api.github.com/repos/huggingface/diffusers/issues/927 | [Community] Make safety model end-to-end compileable - Inference time of JAX / Flax pipeline | I have a question about JAX / Flax pipeline. We run it on a v3-8 Cloud TPU VM and it works pretty great. The first inference as expected takes around a minute because of the compilation. And the next ones take around 5-6 seconds, which is really impressive. Either way, if we change one of the input parameters other tha... | closed | completed | false | 7 | [
"good first issue",
"help wanted"
] | [] | 2022-10-20T14:08:48Z | 2023-03-13T14:55:35Z | 2023-03-13T14:55:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ngaer | 1,527,711 | MDQ6VXNlcjE1Mjc3MTE= | User | false |
huggingface/diffusers | 2,485,320,494 | I_kwDOHa8MBc6UIvsu | 9,270 | https://github.com/huggingface/diffusers/issues/9270 | https://api.github.com/repos/huggingface/diffusers/issues/9270 | Incomplete Lora support for Flux Dev from SimpleTuner | ### Describe the bug
I trained a lora with simpletuner using ai-toolkit preset (I used all+ffs and others and it doesnt train correctly on hard concepts).
And Now I have this issue when loading the lora:
File "/home/axel/Documents/Flux/venv/lib/python3.10/site-packages/torch/nn/modules/module.py", line 2183, in ... | open | null | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-25T15:08:42Z | 2024-12-12T15:04:32Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | axel578 | 44,734,791 | MDQ6VXNlcjQ0NzM0Nzkx | User | false |
huggingface/diffusers | 2,485,737,632 | I_kwDOHa8MBc6UKVig | 9,271 | https://github.com/huggingface/diffusers/issues/9271 | https://api.github.com/repos/huggingface/diffusers/issues/9271 | The different quality between ComfyUI and Diffusers ? | ### Discussed in https://github.com/huggingface/diffusers/discussions/9265
<div type='discussions-op-text'>
<sup>Originally posted by **vuongminh1907** August 25, 2024</sup>
I had a problem using InstantID (https://github.com/instantX-research/InstantID), which uses Diffusers as its base. Additionally, I tried C... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-08-26T02:53:23Z | 2024-10-15T18:10:42Z | 2024-10-15T18:10:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vuongminh1907 | 119,489,204 | U_kgDOBx9CtA | User | false |
huggingface/diffusers | 2,486,190,970 | I_kwDOHa8MBc6UMEN6 | 9,276 | https://github.com/huggingface/diffusers/issues/9276 | https://api.github.com/repos/huggingface/diffusers/issues/9276 | How can I manually update some of their checkpoints of UNet2/3DConditionModel objects? | ### Discussed in https://github.com/huggingface/diffusers/discussions/9273
<div type='discussions-op-text'>
<sup>Originally posted by **justin4ai** August 26, 2024</sup>
Hello, I'm quite new to diffusers package and trying to implement fine-tuning code that uses the saved checkpoints initialized with ```UNet2/3D... | open | null | false | 1 | [
"stale"
] | [] | 2024-08-26T07:49:23Z | 2024-09-25T15:03:01Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | justin4ai | 63,603,383 | MDQ6VXNlcjYzNjAzMzgz | User | false |
huggingface/diffusers | 2,486,686,091 | I_kwDOHa8MBc6UN9GL | 9,278 | https://github.com/huggingface/diffusers/issues/9278 | https://api.github.com/repos/huggingface/diffusers/issues/9278 | FLUX.1-dev dreambooth train on multigpu | ### Discussed in https://github.com/huggingface/diffusers/discussions/9277
<div type='discussions-op-text'>
<sup>Originally posted by **MehmetcanTozlu** August 26, 2024</sup>
I have 4 A6000 Ada Generation cards. but I get an error when training Flux.1-dev with dreambooth(train_dreambooth_flux.py). can anyone hel... | closed | completed | false | 48 | [] | [] | 2024-08-26T11:58:56Z | 2024-10-23T06:04:01Z | 2024-10-23T04:16:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MehmetcanTozlu | 65,121,799 | MDQ6VXNlcjY1MTIxNzk5 | User | false |
huggingface/diffusers | 2,487,105,612 | I_kwDOHa8MBc6UPjhM | 9,279 | https://github.com/huggingface/diffusers/issues/9279 | https://api.github.com/repos/huggingface/diffusers/issues/9279 | Support dynamic LoRA loading with `torch.compile` model | **Is your feature request related to a problem? Please describe.**
Would be great to be able to load a LoRA to a model compiled with `torch.compile`
**Describe the solution you'd like.**
Do `load_lora_weights` with a compiled `pipe` (ideally without triggering recompilation)
Currently, running this code:
```p... | open | null | false | 37 | [
"wip"
] | [] | 2024-08-26T15:10:04Z | 2025-03-31T12:36:18Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,487,921,152 | I_kwDOHa8MBc6USqoA | 9,283 | https://github.com/huggingface/diffusers/issues/9283 | https://api.github.com/repos/huggingface/diffusers/issues/9283 | Implementing training-free RB-Modulation pipeline for most used models | ### Model/Pipeline/Scheduler description
The [RB-Modulation algorithm](https://rb-modulation.github.io/) is **training-free** technique to produce image 2 image style and content transfer in diffusion model. It has two components:
1. Stochastic Optimization Control (SOC): This component requires an [evaluator](https:... | open | null | false | 4 | [
"wip",
"consider-for-modular-diffusers"
] | [] | 2024-08-26T22:48:27Z | 2025-01-12T06:03:50Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AnandK27 | 63,339,285 | MDQ6VXNlcjYzMzM5Mjg1 | User | false |
huggingface/diffusers | 1,417,421,012 | I_kwDOHa8MBc5UfCDU | 929 | https://github.com/huggingface/diffusers/issues/929 | https://api.github.com/repos/huggingface/diffusers/issues/929 | Why is the square root in attention block calculated twice? | [This is the line of code in question](https://github.com/huggingface/diffusers/blob/32bf4fdc4386809c870528cb261028baae012d27/src/diffusers/models/attention.py#L87).
Is it just a bug or is there a reason to do so? As far as I understand, it should be:
```python
scale = 1 / math.sqrt(self.channels / self.num_head... | closed | completed | false | 2 | [] | [] | 2022-10-20T22:47:36Z | 2022-10-24T16:20:37Z | 2022-10-24T16:20:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vinsis | 6,437,295 | MDQ6VXNlcjY0MzcyOTU= | User | false |
huggingface/diffusers | 2,490,507,329 | I_kwDOHa8MBc6UciBB | 9,291 | https://github.com/huggingface/diffusers/issues/9291 | https://api.github.com/repos/huggingface/diffusers/issues/9291 | kohya-ss FLUX LoRA keys not compatible with diffusers | **Is your feature request related to a problem? Please describe.**
kohya-ss has added FLUX.1 LoRA training, however the keys are not working with diffusers (silent erroring, it doesn't error but the style isn't applied)
**Describe the solution you'd like.**
Support `kohya-ss` FLUX.1 trained keys natively
**Addi... | closed | completed | false | 0 | [] | [] | 2024-08-27T23:01:24Z | 2024-08-29T02:11:47Z | 2024-08-29T02:11:47Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,490,529,501 | I_kwDOHa8MBc6Ucnbd | 9,292 | https://github.com/huggingface/diffusers/issues/9292 | https://api.github.com/repos/huggingface/diffusers/issues/9292 | X-Labs FLUX LoRA keys not compatible with diffusers | **Is your feature request related to a problem? Please describe.**
X-Labs was the first lab to add FLUX.1 LoRA training (https://github.com/XLabs-AI/x-flux), however the keys are not working with diffusers (silent erroring - I just caught because I tried after investigating #9291), it doesn't error but the style isn't... | closed | completed | false | 1 | [] | [] | 2024-08-27T23:19:10Z | 2024-08-29T02:11:48Z | 2024-08-29T02:11:48Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,490,648,715 | I_kwDOHa8MBc6UdEiL | 9,293 | https://github.com/huggingface/diffusers/issues/9293 | https://api.github.com/repos/huggingface/diffusers/issues/9293 | TypeError: couldn't find storage object Float8_e4m3fnStorage | ### Describe the bug
latest version of diffusers
torch Version: 2.4.0+cu124 OR Version: 2.3.1
transformers Version: 4.44.2
3090 CUDA Version: 12.5
## Error:
Traceback (most recent call last):
File "/home/ma/work/flux/demo/TotoroUI/flux-demo-hyper.py", line 552, in gen
img_path, image = gen_with_api... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-28T01:01:33Z | 2024-08-29T20:44:25Z | 2024-08-29T20:44:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | maxin9966 | 4,105,807 | MDQ6VXNlcjQxMDU4MDc= | User | false |
huggingface/diffusers | 2,491,350,044 | I_kwDOHa8MBc6Ufvwc | 9,297 | https://github.com/huggingface/diffusers/issues/9297 | https://api.github.com/repos/huggingface/diffusers/issues/9297 | load multiple LORAs, and the load time increases linearly | ### Describe the bug
I try to load multiple LORAs into the SDXL pipe，and the load time increases linearly.
But when i uninstall the peft according to the previous issue，the load time becomes stable.
### Reproduction
```python
from time import perf_counter
from diffusers import StableDiffusionXLPipeline, ... | closed | completed | false | 11 | [
"bug"
] | [] | 2024-08-28T07:52:34Z | 2024-09-03T08:13:31Z | 2024-09-03T08:13:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | javey-q | 49,089,869 | MDQ6VXNlcjQ5MDg5ODY5 | User | false |
huggingface/diffusers | 2,491,604,070 | I_kwDOHa8MBc6Ugtxm | 9,298 | https://github.com/huggingface/diffusers/issues/9298 | https://api.github.com/repos/huggingface/diffusers/issues/9298 | TypeError: couldn't find storage object Float8_e4m3fnStorage | ### System Info
Name: torch
Version: 2.4.0+cu124 OR Version: 2.3.1
Name: transformers
Version: 4.44.2
CUDA Version: 12.5
## Error:
Traceback (most recent call last):
File "/home/ma/work/flux/demo/TotoroUI/flux-demo-hyper.py", line 552, in gen
img_path, image = gen_with_api(prompt, steps, guida... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2024-08-28T00:36:54Z | 2025-06-25T22:38:30Z | 2024-09-29T08:13:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | maxin9966 | 4,105,807 | MDQ6VXNlcjQxMDU4MDc= | User | false |
huggingface/diffusers | 2,491,830,847 | I_kwDOHa8MBc6UhlI_ | 9,299 | https://github.com/huggingface/diffusers/issues/9299 | https://api.github.com/repos/huggingface/diffusers/issues/9299 | CUDAGRAPHs for Flux position embeddings | @yiyixuxu
Is it possible to refactor the Flux positional embeddings so that we can fully make use of CUDAGRAPHs?
```bash
skipping cudagraphs due to skipping cudagraphs due to cpu device (device_put). Found from :
File "/home/sayak/diffusers/src/diffusers/models/transformers/transformer_flux.py", line 469,... | closed | completed | false | 0 | [] | [
"yiyixuxu"
] | 2024-08-28T11:33:16Z | 2024-08-29T19:37:17Z | 2024-08-29T19:37:16Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,306,491,593 | I_kwDOHa8MBc5N33rJ | 93 | https://github.com/huggingface/diffusers/issues/93 | https://api.github.com/repos/huggingface/diffusers/issues/93 | Some tests fail on CPU / M1 | For example, in debugging the test `test_modeling_utils.py::PipelineTesterMixin::test_score_sde_ve_pipeline`, we added a specific value for CPU, but the values on m1 are different slightly again. Here are the values from m1:
```
if model.device.type == "cpu":
expected_image_sum = 3384805888.0
expected_im... | closed | completed | false | 2 | [] | [] | 2022-07-15T20:52:20Z | 2022-08-03T10:49:46Z | 2022-08-03T10:49:45Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | natolambert | 10,695,622 | MDQ6VXNlcjEwNjk1NjIy | User | false |
huggingface/diffusers | 2,491,947,497 | I_kwDOHa8MBc6UiBnp | 9,301 | https://github.com/huggingface/diffusers/issues/9301 | https://api.github.com/repos/huggingface/diffusers/issues/9301 | Convert From XLab controlnet to diffusers format script | **Is your feature request related to a problem? Please describe.**
convert from XLab controlnet to diffusers format
**Describe the solution you'd like.**
convert from XLab controlnet to diffusers format
**Additional context.**
I use the script as below:
```python
import torch
import safetensors.torch
... | closed | completed | false | 6 | [] | [] | 2024-08-28T12:30:06Z | 2024-10-15T22:17:43Z | 2024-10-15T22:17:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chuck-ma | 74,402,255 | MDQ6VXNlcjc0NDAyMjU1 | User | false |
huggingface/diffusers | 2,492,303,031 | I_kwDOHa8MBc6UjYa3 | 9,303 | https://github.com/huggingface/diffusers/issues/9303 | https://api.github.com/repos/huggingface/diffusers/issues/9303 | [Add] VEnhancer - the interpolation and upscaler for CogVideoX-5b | ### Model/Pipeline/Scheduler description
VEnhancer, a generative space-time enhancement framework that can improve the existing T2V results.
https://github.com/Vchitect/VEnhancer
### Open source status
- [X] The model implementation is available.
- [X] The model weights are available (Only relevant if addition is... | open | null | false | 3 | [
"stale"
] | [] | 2024-08-28T14:43:32Z | 2024-12-11T15:04:32Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 2,492,885,459 | I_kwDOHa8MBc6UlmnT | 9,306 | https://github.com/huggingface/diffusers/issues/9306 | https://api.github.com/repos/huggingface/diffusers/issues/9306 | [ERROR] CogVideoXPipeline.from_pretrained wrong device selected | ### Describe the bug
Running the new `CogVideoXPipeline.from_pretrained` with `to` device will not send all layers to specified device:
```python
torch.cuda.set_device( device )
device = f'cuda:{device}'
print(f'USING DEVICE {device} current_device:{torch.cuda.current_device()}')
pipe = Co... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-28T19:58:00Z | 2024-08-29T19:38:05Z | 2024-08-29T19:38:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | loretoparisi | 163,333 | MDQ6VXNlcjE2MzMzMw== | User | false |
huggingface/diffusers | 2,493,385,390 | I_kwDOHa8MBc6Ungqu | 9,309 | https://github.com/huggingface/diffusers/issues/9309 | https://api.github.com/repos/huggingface/diffusers/issues/9309 | [Flux] FluxControlNetPipeline run on Flux 1. schnell bug | ### Describe the bug
When I use FluxControlNetPipeline,
`Loading pipeline components...: 0%| | 0/7 [00:00<?, ?it/s]You set `add_prefix_space`. The tokenizer needs to be converted fro... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-29T03:44:58Z | 2024-09-04T02:00:33Z | 2024-09-04T02:00:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yinian-lw | 125,951,313 | U_kgDOB4HdUQ | User | false |
huggingface/diffusers | 1,417,498,031 | I_kwDOHa8MBc5UfU2v | 931 | https://github.com/huggingface/diffusers/issues/931 | https://api.github.com/repos/huggingface/diffusers/issues/931 | Conversion Script: Update to Stable Diffusion 1.5 (EMA and non-EMA) | Since stable diffusion 1.5 was out, I was hoping to use it with the imagic colab. Could you upgrade diffusers to use the new model instead of the old one?
| closed | completed | false | 17 | [
"stale"
] | [
"pcuenca",
"patil-suraj"
] | 2022-10-21T00:07:20Z | 2023-01-22T15:03:32Z | 2023-01-22T15:03:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | redwraith2 | 4,700,187 | MDQ6VXNlcjQ3MDAxODc= | User | false |
huggingface/diffusers | 2,494,019,475 | I_kwDOHa8MBc6Up7eT | 9,314 | https://github.com/huggingface/diffusers/issues/9314 | https://api.github.com/repos/huggingface/diffusers/issues/9314 | for the FluxTransformer2DModel.from_single_file(), | for the FluxTransformer2DModel.from_single_file(),
when i change the path to local, it doesnt work:
OSError: runwayml/stable-diffusion-v1-5 does not appear to have a file named config.json.
_Originally posted by @tomsin9 in https://github.com/huggingface/diffusers/issues/9141#issuecomment-229598373... | closed | not_planned | false | 2 | [] | [] | 2024-08-29T10:10:52Z | 2024-09-18T20:14:50Z | 2024-08-29T10:11:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sxy771 | 36,506,077 | MDQ6VXNlcjM2NTA2MDc3 | User | false |
huggingface/diffusers | 2,494,112,236 | I_kwDOHa8MBc6UqSHs | 9,315 | https://github.com/huggingface/diffusers/issues/9315 | https://api.github.com/repos/huggingface/diffusers/issues/9315 | support kolors with controlnet and ipadapter-Faceid | please support kolors with controlnet and ipadapter-Faceid
| closed | completed | false | 6 | [] | [] | 2024-08-29T10:57:52Z | 2024-11-18T18:14:53Z | 2024-11-18T03:10:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | akk-123 | 98,469,560 | U_kgDOBd6GuA | User | false |
huggingface/diffusers | 2,494,284,452 | I_kwDOHa8MBc6Uq8Kk | 9,317 | https://github.com/huggingface/diffusers/issues/9317 | https://api.github.com/repos/huggingface/diffusers/issues/9317 | Finetuning on dataset | dear @thedarkzeno and @patil-suraj
Thank you so much for putting your work out there. I wanted to ask, how would the training be for training on a dataset and not a single instance image as mentioned in train_dreambooth_inpaint. And can I finetune models trained from https://github.com/CompVis/latent-diffusion repo... | closed | completed | false | 4 | [
"stale"
] | [] | 2024-08-29T12:20:51Z | 2024-10-23T16:10:47Z | 2024-10-23T16:10:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ultiwinter | 95,130,956 | U_kgDOBauVTA | User | false |
huggingface/diffusers | 2,495,138,667 | I_kwDOHa8MBc6UuMtr | 9,320 | https://github.com/huggingface/diffusers/issues/9320 | https://api.github.com/repos/huggingface/diffusers/issues/9320 | Finetuned models aren't saved or loaded properly in train_custom_diffusion.py | ### Describe the bug
Thank you for your amazing work. It seems like models are not saved or loaded properly after finetuning train_custom_diffusion.py in a new dataset. Generated validation images are expected but final test (inference) results are not expected as finetuned models are not used properly. The bug most p... | open | null | false | 7 | [
"bug"
] | [] | 2024-08-29T17:28:10Z | 2025-05-10T06:05:12Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mostafij-rahman | 42,969,132 | MDQ6VXNlcjQyOTY5MTMy | User | false |
huggingface/diffusers | 2,495,647,098 | I_kwDOHa8MBc6UwI16 | 9,322 | https://github.com/huggingface/diffusers/issues/9322 | https://api.github.com/repos/huggingface/diffusers/issues/9322 | RunwayML obsoletion causes issues | ### Describe the bug
RunwayML just removed their entire code from huggingface!
See note at <https://huggingface.co/runwayml>: *"We are no longer maintaining a HuggingFace organization"*
Which causes pretty big issues since `runwayml/stable-diffusion-v1-5` is referenced as default/fallback in many places in the... | closed | completed | false | 15 | [
"bug",
"stale",
"contributions-welcome"
] | [] | 2024-08-29T21:54:16Z | 2024-10-26T23:12:45Z | 2024-10-26T23:12:44Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vladmandic | 57,876,960 | MDQ6VXNlcjU3ODc2OTYw | User | false |
huggingface/diffusers | 2,496,246,399 | I_kwDOHa8MBc6UybJ_ | 9,325 | https://github.com/huggingface/diffusers/issues/9325 | https://api.github.com/repos/huggingface/diffusers/issues/9325 | Unload lora specific lora weight | I've had trouble loading a few loras, and then unloading only a select few to free up more memory, but we only support unload_lora_weights() currently, which unloads all the loras - so I have to reload the loras I want again which is time consuming.
I would like a solution where we could specify the adapter_name as... | open | null | false | 22 | [
"stale",
"lora"
] | [
"asomoza",
"sayakpaul"
] | 2024-08-30T05:05:18Z | 2025-02-01T15:04:53Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AMohamedAakhil | 78,485,654 | MDQ6VXNlcjc4NDg1NjU0 | User | false |
huggingface/diffusers | 2,496,918,515 | I_kwDOHa8MBc6U0_Pz | 9,326 | https://github.com/huggingface/diffusers/issues/9326 | https://api.github.com/repos/huggingface/diffusers/issues/9326 | [Pipeline] AnimateDiff + VideoToVideo + ControlNet | As the title suggests, we would like a pipeline that supports all three techniques. Currently, we have standalone pipelines for [VideoToVideo](https://github.com/huggingface/diffusers/blob/e417d028115e72b953a73e39d9687aa70ba3e37e/src/diffusers/pipelines/animatediff/pipeline_animatediff_video2video.py) and [ControlNet](... | closed | completed | false | 3 | [
"good first issue",
"community-examples",
"contributions-welcome",
"video"
] | [] | 2024-08-30T11:00:07Z | 2024-09-09T17:20:46Z | 2024-09-09T17:20:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | a-r-r-o-w | 72,266,394 | MDQ6VXNlcjcyMjY2Mzk0 | User | false |
huggingface/diffusers | 2,497,379,530 | I_kwDOHa8MBc6U2vzK | 9,328 | https://github.com/huggingface/diffusers/issues/9328 | https://api.github.com/repos/huggingface/diffusers/issues/9328 | Sample latents from VAE to generate close images | hi,
I would like to use a VAE (SDXL's VAE in the example below) in order to get close variations of a given image, in the spirit of what was demonstrated in [that paper](https://arxiv.org/abs/1609.04468).
There are examples out there using the MNIST or faces datasets.
I could achieve this with inpainting or mayb... | closed | completed | false | 6 | [
"stale"
] | [] | 2024-08-30T14:22:12Z | 2024-10-09T16:51:32Z | 2024-10-09T16:51:31Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Clement-Lelievre | 70,368,164 | MDQ6VXNlcjcwMzY4MTY0 | User | false |
huggingface/diffusers | 2,497,721,143 | I_kwDOHa8MBc6U4DM3 | 9,329 | https://github.com/huggingface/diffusers/issues/9329 | https://api.github.com/repos/huggingface/diffusers/issues/9329 | [Pipeline] AnimateDiff + SparseControl + ControlNet | As the title suggests, we would like a pipeline that supports all three techniques. Currently, we have standalone pipelines for [SparseCtrl](https://github.com/huggingface/diffusers/blob/e417d028115e72b953a73e39d9687aa70ba3e37e/src/diffusers/pipelines/animatediff/pipeline_animatediff_sparsectrl.py) and [ControlNet](htt... | open | null | false | 14 | [
"good first issue",
"community-examples",
"contributions-welcome",
"consider-for-modular-diffusers"
] | [] | 2024-08-30T16:28:27Z | 2025-09-10T10:34:06Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aihopper | 174,629,473 | U_kgDOCmiiYQ | User | false |
huggingface/diffusers | 2,498,082,096 | I_kwDOHa8MBc6U5bUw | 9,331 | https://github.com/huggingface/diffusers/issues/9331 | https://api.github.com/repos/huggingface/diffusers/issues/9331 | Problem in FlowMatchEulerDiscreteScheduler's index_for_timestep method | ### Describe the bug
In `index_for_timestep()` ([here](https://github.com/huggingface/diffusers/blob/d8a16635f47ac455abd61879bcc6be32dfeaa561/src/diffusers/schedulers/scheduling_flow_match_euler_discrete.py#L213)), the provided timestep is checked against the scheduler's timesteps. However, the scheduler timesteps are... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-30T19:57:17Z | 2024-09-30T20:52:19Z | 2024-09-30T20:52:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mashrurmorshed | 38,450,043 | MDQ6VXNlcjM4NDUwMDQz | User | false |
huggingface/diffusers | 2,498,724,956 | I_kwDOHa8MBc6U74Rc | 9,332 | https://github.com/huggingface/diffusers/issues/9332 | https://api.github.com/repos/huggingface/diffusers/issues/9332 | StableDiffusionPipeline.from_pretrained() fails to load specified models, resulting in identical weights | ### Describe the bug
## Issue Summary
The `from_pretrained()` method of StableDiffusionPipeline fails to correctly load the specified models on local directory. Instead, it appears to use the same (likely default) weights for different model paths. This results in identical models being used even when attempting ... | open | null | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-31T08:19:29Z | 2024-10-26T15:03:27Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fule1255 | 42,274,176 | MDQ6VXNlcjQyMjc0MTc2 | User | false |
huggingface/diffusers | 2,499,387,188 | I_kwDOHa8MBc6U-Z80 | 9,336 | https://github.com/huggingface/diffusers/issues/9336 | https://api.github.com/repos/huggingface/diffusers/issues/9336 | IPAdapter attention layers run with requires_grad=True for parameters | ### Describe the bug
When IP adapter is loaded, the IPAdapterAttnProcessor2_0 class is not set with training=False. Didnt see a huge difference in memory if its manually set to false on those processors, but it looks like a miss?
Same for image_projection_layers
### Reproduction
```
pipeline.load_ip_adapter("h94/I... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-09-01T10:50:25Z | 2024-10-02T05:16:25Z | 2024-10-02T05:16:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | darshats | 56,063,876 | MDQ6VXNlcjU2MDYzODc2 | User | false |
huggingface/diffusers | 2,500,001,486 | I_kwDOHa8MBc6VAv7O | 9,339 | https://github.com/huggingface/diffusers/issues/9339 | https://api.github.com/repos/huggingface/diffusers/issues/9339 | SD3 inpatinting | I found the StableDiffusion3InpaintPipeline, where can i found the weight of SD3 inpainting | closed | completed | false | 5 | [
"stale"
] | [] | 2024-09-02T05:00:19Z | 2024-10-02T15:43:24Z | 2024-10-02T15:43:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ucasyjz | 35,628,125 | MDQ6VXNlcjM1NjI4MTI1 | User | false |
huggingface/diffusers | 2,500,514,100 | I_kwDOHa8MBc6VCtE0 | 9,341 | https://github.com/huggingface/diffusers/issues/9341 | https://api.github.com/repos/huggingface/diffusers/issues/9341 | cannot load flux lora that contain text_model lora | ### Describe the bug
When I try to load a lora trained by sd-scripts that contain text_model, it failed on text_model part.
### Reproduction
```
from diffusers.pipelines.flux.pipeline_flux import FluxPipeline
pipeline = FluxPipeline.from_pretrained("FLUX.1-dev")
pipeline.load_lora_weights("lora.safetens... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-09-02T09:50:38Z | 2024-10-02T19:58:32Z | 2024-10-02T19:58:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jiashenggu | 32,376,856 | MDQ6VXNlcjMyMzc2ODU2 | User | false |
huggingface/diffusers | 2,500,719,304 | I_kwDOHa8MBc6VDfLI | 9,343 | https://github.com/huggingface/diffusers/issues/9343 | https://api.github.com/repos/huggingface/diffusers/issues/9343 | FLUX error when loading with low_cpu_mem_usage=False and ignore_mismatched_sizes=True | ### Describe the bug
I'd like to change the input layers of FLUX for training some img2img stuff, but got:
`TypeError: expected str, bytes or os.PathLike object, not NoneType`
when loading `FluxTransformer2DModel` with `low_cpu_mem_usage=False, ignore_mismatched_sizes=True`.
### Reproduction
```
from diffusers.mo... | open | null | false | 16 | [
"bug",
"stale"
] | [] | 2024-09-02T11:28:48Z | 2026-01-04T01:33:08Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | primecai | 43,519,981 | MDQ6VXNlcjQzNTE5OTgx | User | false |
huggingface/diffusers | 2,501,825,108 | I_kwDOHa8MBc6VHtJU | 9,350 | https://github.com/huggingface/diffusers/issues/9350 | https://api.github.com/repos/huggingface/diffusers/issues/9350 | Passing `txt_ids` 3d torch.Tensor is deprecated.Please remove the batch dimension and pass it as a 2d torch Tensor | ### Describe the bug
Passing `txt_ids` 3d torch.Tensor is deprecated.Please remove the batch dimension and pass it as a 2d torch Tensor
Is the bath_size shouldn't be 1?
### Reproduction
accelerate launch train_dreambooth_lora_flux.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--dataset_name=$DAT... | closed | completed | false | 7 | [
"bug",
"good first issue",
"contributions-welcome"
] | [] | 2024-09-03T02:54:41Z | 2024-10-15T11:48:14Z | 2024-10-15T11:48:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | arrowonstr | 145,024,485 | U_kgDOCKTl5Q | User | false |
huggingface/diffusers | 2,502,066,042 | I_kwDOHa8MBc6VIn96 | 9,352 | https://github.com/huggingface/diffusers/issues/9352 | https://api.github.com/repos/huggingface/diffusers/issues/9352 | Text generation? | Hi thanks for this great library!
There seems to be some diffusion models that generate text, instead of images. (For example, these two surveys: https://arxiv.org/abs/2303.06574, https://www.semanticscholar.org/paper/Diffusion-models-in-text-generation%3A-a-survey-Yi-Chen/41941f072db18972b610de9979e755afba35f11e). ... | open | null | false | 13 | [
"wip"
] | [
"asomoza"
] | 2024-09-03T06:54:38Z | 2024-11-23T04:57:37Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fzyzcjy | 5,236,035 | MDQ6VXNlcjUyMzYwMzU= | User | false |
huggingface/diffusers | 2,502,499,857 | I_kwDOHa8MBc6VKR4R | 9,356 | https://github.com/huggingface/diffusers/issues/9356 | https://api.github.com/repos/huggingface/diffusers/issues/9356 | pipeline_stable_diffusion_xl_adapter | ### Describe the bug
I want to rewrite the call function of the pipeline_stable_diffusion_xl_adapter. When I want to use the function prepare_ip_adapter_image_embeds, there is an error called "AttributeError: 'NoneType' object has no attribute 'image_projection_layers'". The error tells me that the attribution self.un... | open | null | false | 6 | [
"bug",
"stale"
] | [] | 2024-09-03T10:25:57Z | 2024-10-28T15:03:18Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Yuhan291 | 70,860,147 | MDQ6VXNlcjcwODYwMTQ3 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.