repo stringclasses 1
value | github_id int64 1.27B 4.95B | github_node_id stringlengths 18 24 | number int64 8 14.3k | 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 2026-07-22 10:32:58 | updated_at stringdate 2022-06-12 22:18:01 2026-07-22 10:32:58 | closed_at stringdate 2022-06-12 22:18:01 2026-07-21 13:17:28 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 149
values | extracted_at stringdate 2026-04-07 13:34:13 2026-07-22 10:35:16 | author_login stringlengths 3 28 | author_id int64 1.54k 296M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 2,023,098,933 | I_kwDOHa8MBc54lgo1 | 6,037 | https://github.com/huggingface/diffusers/issues/6037 | https://api.github.com/repos/huggingface/diffusers/issues/6037 | [New Pipeline] InstaFlow! One-Step Stable Diffusion with Rectified Flow | ### Model/Pipeline/Scheduler description
InstaFlow is an ultra-fast, one-step image generator that achieves image quality close to Stable Diffusion, significantly reducing the demand of computational resources. This efficiency is made possible through a recent [Rectified Flow](https://github.com/gnobitab/RectifiedFlow... | closed | completed | false | 8 | [
"stale"
] | [] | 2023-12-04T06:06:41Z | 2024-01-15T15:06:24Z | 2024-01-15T15:06:24Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ayushtues | 43,698,245 | MDQ6VXNlcjQzNjk4MjQ1 | User | false |
huggingface/diffusers | 2,023,161,740 | I_kwDOHa8MBc54lv-M | 6,039 | https://github.com/huggingface/diffusers/issues/6039 | https://api.github.com/repos/huggingface/diffusers/issues/6039 | DatasetGenerationError: An error occurred while generating the dataset | I tried to run train_instruct_pix2pix_sdxl.py with my local data that looks like this:
**online_data/ 001input.jpg, 001edited.jpg, etc**
with a metadata.jsonl file following instructions in [here](https://huggingface.co/docs/datasets/image_dataset#imagefolder), and its lines look like : **{"input_image": "001input.j... | closed | not_planned | false | 2 | [] | [] | 2023-12-04T06:58:54Z | 2023-12-07T12:21:45Z | 2023-12-07T12:21:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DoctorTar | 147,801,017 | U_kgDOCM9DuQ | User | false |
huggingface/diffusers | 1,380,514,847 | I_kwDOHa8MBc5SSPwf | 604 | https://github.com/huggingface/diffusers/issues/604 | https://api.github.com/repos/huggingface/diffusers/issues/604 | [FLAX-PT conversion] test | Snippet from @patil-suraj that should become test for testing `from_pt=True` conversion
```py
from diffusers import UNet2DConditionModel, FlaxUNet2DConditionModel
import tempfile
import torch
import numpy as np
import jax.numpy as jnp
model = UNet2DConditionModel(
block_out_channels=(32, 64),
layer... | closed | completed | false | 3 | [
"stale"
] | [] | 2022-09-21T08:14:00Z | 2022-10-24T20:43:59Z | 2022-10-24T20:43:59Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mishig25 | 11,827,707 | MDQ6VXNlcjExODI3NzA3 | User | false |
huggingface/diffusers | 2,023,432,277 | I_kwDOHa8MBc54myBV | 6,041 | https://github.com/huggingface/diffusers/issues/6041 | https://api.github.com/repos/huggingface/diffusers/issues/6041 | Difference between origin EDM traning process and diffusers EDM training process | in the original EDM training the σ is sampled from a certain Gaussian distribution
[https://github.com/NVlabs/edm/blob/main/training/loss.py#L73C1-L73C9](https://github.com/NVlabs/edm/blob/main/training/loss.py#L73C1-L73C9)
![image](https://github.com/huggingface/diffusers/assets/26294676/85f44a73-329f-4267-97d0-940d... | closed | completed | false | 5 | [] | [] | 2023-12-04T09:47:22Z | 2023-12-06T08:28:43Z | 2023-12-05T09:24:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bendanzzc | 26,294,676 | MDQ6VXNlcjI2Mjk0Njc2 | User | false |
huggingface/diffusers | 2,023,452,777 | I_kwDOHa8MBc54m3Bp | 6,042 | https://github.com/huggingface/diffusers/issues/6042 | https://api.github.com/repos/huggingface/diffusers/issues/6042 | Sdxl-turbo distilling example? | ### Model/Pipeline/Scheduler description
Thanks for your dedication to the community, and any update would be greatly appreciated.
I'm excited about the sdxl-turbo algorithm and would like to inquire if there is any chance for an open-source diffusers version of the ADD distilling example.
### Open source stat... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-12-04T09:58:28Z | 2024-01-15T15:06:23Z | 2024-01-15T15:06:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | noway2beatme | 63,312,875 | MDQ6VXNlcjYzMzEyODc1 | User | false |
huggingface/diffusers | 2,023,730,729 | I_kwDOHa8MBc54n64p | 6,044 | https://github.com/huggingface/diffusers/issues/6044 | https://api.github.com/repos/huggingface/diffusers/issues/6044 | 'StableDiffusionImg2ImgPipeline' object has no attribute 'do_classifier_free_guidance' | ### Describe the bug
I try to use the "CompVis/ldm-celebahq-256" unet with the "runwayml/stable-diffusion-v1-5" clip to make img to img pipeline generate more human-like output as a part of my graduation project
### Reproduction
unet = UNet2DModel.from_pretrained("CompVis/ldm-celebahq-256", subfolder="unet")
vqvae... | closed | not_planned | false | 10 | [
"bug",
"stale"
] | [] | 2023-12-04T12:20:22Z | 2024-11-28T02:19:12Z | 2024-01-13T20:31:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Abdelrahman-Hanafy | 65,138,724 | MDQ6VXNlcjY1MTM4NzI0 | User | false |
huggingface/diffusers | 2,024,284,393 | I_kwDOHa8MBc54qCDp | 6,048 | https://github.com/huggingface/diffusers/issues/6048 | https://api.github.com/repos/huggingface/diffusers/issues/6048 | Pixart with Controllnet | ### Model/Pipeline/Scheduler description
Hello, thanks a lot for your great job.
However, I am wondering to use pixart with a extended constrain like controlnet's canny, but it seems that there is no such pipeline though shown on Pixart's github page. Will you plan to add this pipeline, or this will be proceeded af... | open | reopened | false | 7 | [
"stale",
"contributions-welcome"
] | [] | 2023-12-04T16:42:55Z | 2024-03-17T15:04:28Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Aziily | 101,855,289 | U_kgDOBhIwOQ | User | false |
huggingface/diffusers | 2,024,835,545 | I_kwDOHa8MBc54sInZ | 6,050 | https://github.com/huggingface/diffusers/issues/6050 | https://api.github.com/repos/huggingface/diffusers/issues/6050 | pip install --upgrade diffusers[torch] fails with error: subprocess-exited-with-error | Specifications :
Windows 10
Python 3
See Image for details
| closed | completed | false | 7 | [] | [] | 2023-12-04T22:00:29Z | 2023-12-11T08:10:42Z | 2023-12-11T08:10:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shivam-maurya-git | 74,864,959 | MDQ6VXNlcjc0ODY0OTU5 | User | false |
huggingface/diffusers | 2,025,436,589 | I_kwDOHa8MBc54ubWt | 6,054 | https://github.com/huggingface/diffusers/issues/6054 | https://api.github.com/repos/huggingface/diffusers/issues/6054 | EulerDiscreteScheduler.set_timesteps() torch.from_numpy misuse error. | ### Describe the bug
In version 0.24.0, on line 283 of ``schedulers\scheduling_euler_discrete.py``
https://github.com/huggingface/diffusers/blob/76c645d3a641c879384afcb43496f0b7db8cc5cb/src/diffusers/schedulers/scheduling_euler_discrete.py#L283
An exception occurs when ``sigmas`` is a Tensor object and not a num... | closed | completed | false | 11 | [
"bug",
"good first issue",
"contributions-welcome"
] | [] | 2023-12-05T07:15:54Z | 2024-11-28T05:34:05Z | 2024-11-28T05:34:05Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Teriks | 14,919,098 | MDQ6VXNlcjE0OTE5MDk4 | User | false |
huggingface/diffusers | 2,025,592,232 | I_kwDOHa8MBc54vBWo | 6,055 | https://github.com/huggingface/diffusers/issues/6055 | https://api.github.com/repos/huggingface/diffusers/issues/6055 | Error with accelerate | ### Describe the bug
can't open file 'D:\\Work\\diffusers\\examples\\dreambooth\\train_dream_booth.py': [Errno 2] No such file or directory
Command '['C:\\Users\\admin\\scoop\\apps\\python\\current\\python.exe', 'train_dream_booth.py']' returned non-zero exit status 2.
when i run accelerate launch train_dream_booth.... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-12-05T08:29:46Z | 2024-01-19T15:05:24Z | 2024-01-19T15:05:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thainh11 | 127,828,714 | U_kgDOB56C6g | User | false |
huggingface/diffusers | 1,380,576,182 | I_kwDOHa8MBc5SSeu2 | 606 | https://github.com/huggingface/diffusers/issues/606 | https://api.github.com/repos/huggingface/diffusers/issues/606 | [Community] Flax: load scheduler using same dtype as pipeline | Follow up to https://github.com/huggingface/diffusers/pull/600#discussion_r976064534 | closed | completed | false | 3 | [
"Good second issue"
] | [
"pcuenca"
] | 2022-09-21T08:56:34Z | 2024-11-24T03:24:21Z | 2024-11-24T03:24:21Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 2,026,239,309 | I_kwDOHa8MBc54xfVN | 6,062 | https://github.com/huggingface/diffusers/issues/6062 | https://api.github.com/repos/huggingface/diffusers/issues/6062 | Preprocessor: reference_only for ContolNet | ### Describe the bug
Hello,
I hope you are doing well. I have an issue with the preprocessor for Control Net. I am trying to transition from automatic to code-based workflow. I was attempting to create a ControlNet with the control type 'reference_only', but I haven't found a way to do it.
Thank you in advance f... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-12-05T13:47:36Z | 2024-02-05T09:50:52Z | 2023-12-08T07:10:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhuraromdev | 78,348,856 | MDQ6VXNlcjc4MzQ4ODU2 | User | false |
huggingface/diffusers | 2,026,512,826 | I_kwDOHa8MBc54yiG6 | 6,063 | https://github.com/huggingface/diffusers/issues/6063 | https://api.github.com/repos/huggingface/diffusers/issues/6063 | Adding StyleAligned | ### Model/Pipeline/Scheduler description
StyleAligned allows images to be created that follow the style of a reference image, without fine-tuning the pre-trained diffusion model! The method uses adaptive instance normalization layers and with the exception of the reference image, all other images to be generated per... | closed | completed | false | 3 | [] | [] | 2023-12-05T15:45:21Z | 2024-01-11T13:35:57Z | 2024-01-11T13:35:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shermansiu | 12,627,125 | MDQ6VXNlcjEyNjI3MTI1 | User | false |
huggingface/diffusers | 2,027,429,360 | I_kwDOHa8MBc542B3w | 6,067 | https://github.com/huggingface/diffusers/issues/6067 | https://api.github.com/repos/huggingface/diffusers/issues/6067 | How to run the fine_tuned model? | Hi all,
I used the instructions given [here](https://github.com/huggingface/diffusers/tree/main/examples/dreambooth) to fine_tune the model on dog pictures (as explained in the link).
The fine_tuning has finished, and a folder called path-to-save-model has been created (that has the weights of the model). Now how d... | closed | completed | false | 2 | [] | [] | 2023-12-06T01:01:56Z | 2025-04-28T10:32:33Z | 2023-12-07T18:48:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alireza18878 | 82,774,346 | MDQ6VXNlcjgyNzc0MzQ2 | User | false |
huggingface/diffusers | 2,027,431,392 | I_kwDOHa8MBc542CXg | 6,068 | https://github.com/huggingface/diffusers/issues/6068 | https://api.github.com/repos/huggingface/diffusers/issues/6068 | DDIM produces incorrect samples with SDXL (epsilon or v-prediction) | ### Describe the bug
When generating images with SDXL and DDIM, there is some residual noise in the outputs.
This leads to a "smudgy" look, and in cases where fewer steps are used, DDIM and Euler diverge a lot more than they should because of the cumulative impact of not having the timesteps aligned properly.
In... | closed | completed | false | 40 | [
"bug",
"scheduler"
] | [
"yiyixuxu"
] | 2023-12-06T01:04:08Z | 2024-04-03T15:16:37Z | 2024-04-03T15:16:37Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,027,621,793 | I_kwDOHa8MBc542w2h | 6,069 | https://github.com/huggingface/diffusers/issues/6069 | https://api.github.com/repos/huggingface/diffusers/issues/6069 | DPM ++ karras scheduler.add_noise fails when t = 0 | ### Describe the bug
When I calling the add_noise function of dpm++ karras when t = 0, it fails
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/xxx/miniconda3/envs/controlnet/lib/python3.8/site-packages/diffusers/schedulers/scheduling_dpmsolver_multistep.py", line 889, in... | closed | completed | false | 0 | [
"bug"
] | [
"yiyixuxu"
] | 2023-12-06T04:23:24Z | 2023-12-07T21:35:30Z | 2023-12-07T21:35:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | garychan22 | 108,175,311 | U_kgDOBnKfzw | User | false |
huggingface/diffusers | 2,027,775,653 | I_kwDOHa8MBc543Wal | 6,070 | https://github.com/huggingface/diffusers/issues/6070 | https://api.github.com/repos/huggingface/diffusers/issues/6070 | How to overload existing class in diffusers | That's just for personal development. I want to write a new class inherited from existing class (e.g. `ControlNetModel`) and I added some new parameters to `__init__` function, but found that the `__init__` function is still the parent's implementation, whether to add the decorator `register_to_config` or not.
Hope ... | closed | completed | false | 7 | [] | [] | 2023-12-06T06:41:44Z | 2024-09-25T14:44:04Z | 2023-12-10T06:44:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | OrangeSodahub | 54,439,582 | MDQ6VXNlcjU0NDM5NTgy | User | false |
huggingface/diffusers | 2,028,183,284 | I_kwDOHa8MBc544570 | 6,073 | https://github.com/huggingface/diffusers/issues/6073 | https://api.github.com/repos/huggingface/diffusers/issues/6073 | I am A beginner. When I practice according to diffusers/examples/dreambooth /README_sdxl.md, I encounter an error. Please help me | ### Describe the bug
A matching Triton is not available, some optimizations will not be enabled.
Error caught was: No module named 'triton'
12/06/2023 17:54:53 - INFO - __main__ - Distributed environment: NO
Num processes: 1
Process index: 0
Local process index: 0
Device: cuda
Mixed precision type: fp16
Tr... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-12-06T10:04:43Z | 2024-01-26T15:06:30Z | 2024-01-26T15:06:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | julyao-yph | 59,270,488 | MDQ6VXNlcjU5MjcwNDg4 | User | false |
huggingface/diffusers | 2,028,596,689 | I_kwDOHa8MBc546e3R | 6,077 | https://github.com/huggingface/diffusers/issues/6077 | https://api.github.com/repos/huggingface/diffusers/issues/6077 | Dataset Structure and Metadata Issues in ControlNet Training | I want to train a ControlNet using my custom dataset with the following structure:
```bash
train
├── conditioning_images
│ ├── 00001.png
│ └── 00002.png
└── images
├── 00001.jpg
└── 00002.jpg
```
For the `metadata.jsonl` file, I understand it needs a `file_name` column linking to image files. ... | closed | completed | false | 8 | [] | [] | 2023-12-06T13:45:27Z | 2025-12-04T13:49:33Z | 2024-01-04T23:28:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hanweikung | 643,825 | MDQ6VXNlcjY0MzgyNQ== | User | false |
huggingface/diffusers | 2,029,526,704 | I_kwDOHa8MBc54-B6w | 6,079 | https://github.com/huggingface/diffusers/issues/6079 | https://api.github.com/repos/huggingface/diffusers/issues/6079 | error: Expected scalar typeHalf but found Float | ### Describe the bug
Facing this issue with SDXL sporadically,Time and again.
### Reproduction
Normally running SDXL pipeline.
Also have,
-> base_pipe.enable_xformers_memory_efficient_attention()
### Logs
```shell
"error": "{\"error_type\": \"<class 'RuntimeError'>\", \"error_message\": \"expected scalar t... | closed | not_planned | false | 0 | [
"bug"
] | [] | 2023-12-06T22:47:08Z | 2023-12-06T23:15:17Z | 2023-12-06T23:15:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pandyamarut | 28,846,178 | MDQ6VXNlcjI4ODQ2MTc4 | User | false |
huggingface/diffusers | 2,029,783,185 | I_kwDOHa8MBc54_AiR | 6,080 | https://github.com/huggingface/diffusers/issues/6080 | https://api.github.com/repos/huggingface/diffusers/issues/6080 | is there any pipeline for outpainting? | outpainting is very interesting
does diffusers support it?
| closed | completed | false | 25 | [
"stale",
"contributions-welcome"
] | [] | 2023-12-07T02:58:34Z | 2024-09-17T21:52:03Z | 2024-09-17T21:52:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyun0071 | 35,762,050 | MDQ6VXNlcjM1NzYyMDUw | User | false |
huggingface/diffusers | 2,029,985,674 | I_kwDOHa8MBc54_x-K | 6,082 | https://github.com/huggingface/diffusers/issues/6082 | https://api.github.com/repos/huggingface/diffusers/issues/6082 | Err: expected scalar type Half but found Float\ | ### Describe the bug
My dependencies are upto date, But I face this issue sporadically, while loading the model.
### Reproduction
class ModelHandler:
def __init__(self):
self.base = None
self.refiner = None
self.load_models()
def load_base(self):
vae = AutoencoderKL.f... | closed | completed | false | 11 | [
"bug",
"stale"
] | [] | 2023-12-07T06:14:19Z | 2024-01-18T15:05:25Z | 2024-01-18T15:05:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pandyamarut | 28,846,178 | MDQ6VXNlcjI4ODQ2MTc4 | User | false |
huggingface/diffusers | 2,030,100,710 | I_kwDOHa8MBc55AODm | 6,084 | https://github.com/huggingface/diffusers/issues/6084 | https://api.github.com/repos/huggingface/diffusers/issues/6084 | LCM Lora training | Hi there,
I wanna train LCM Lora
But i faced issue like this "AttributeError: Can't pickle local object 'filter_keys.<locals>._f'"
is this problem with dataset? the problem occurs this line
`consistency_distillation\train_lcm_distill_lora_sd_wds.py", line 1124, in main
for step, batch in enumerate(train_da... | closed | completed | false | 2 | [] | [] | 2023-12-07T07:45:42Z | 2023-12-15T05:01:26Z | 2023-12-15T05:01:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anomySz | 128,664,175 | U_kgDOB6tCbw | User | false |
huggingface/diffusers | 2,030,115,700 | I_kwDOHa8MBc55ARt0 | 6,086 | https://github.com/huggingface/diffusers/issues/6086 | https://api.github.com/repos/huggingface/diffusers/issues/6086 | In training the script train_text_to_image_lora.py on Colab with a V100 GPU, the error ValueError: Attempting to unscale FP16 gradients occurred. | ### Describe the bug
12/07/2023 07:37:24 - INFO - __main__ - ***** Running training *****
12/07/2023 07:37:24 - INFO - __main__ - Num examples = 833
12/07/2023 07:37:24 - INFO - __main__ - Num Epochs = 72
12/07/2023 07:37:24 - INFO - __main__ - Instantaneous batch size per device = 1
12/07/2023 07:37:24 - IN... | closed | completed | false | 21 | [
"bug"
] | [] | 2023-12-07T07:56:39Z | 2024-01-06T21:08:59Z | 2023-12-19T04:24:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shangvo | 52,747,388 | MDQ6VXNlcjUyNzQ3Mzg4 | User | false |
huggingface/diffusers | 2,030,302,226 | I_kwDOHa8MBc55A_QS | 6,087 | https://github.com/huggingface/diffusers/issues/6087 | https://api.github.com/repos/huggingface/diffusers/issues/6087 | diffusers doesn't save and load the LoraConfig, resulting wrong `lora_alpha` during inference. | ### Describe the bug
When I run `train_dreambooth_lora_sdxl.py`, I find the StableDiffusionXLLoraLoaderMixin (and LoraLoaderMixin) doesn't save & load the LoraConfig.
The default `lora_alpha` is 8 in LoraConfig but is 1 in `peft` Linear. The lora_alpha is 8 during training but is 1 when I load the weights from a ... | closed | completed | false | 12 | [
"bug"
] | [] | 2023-12-07T09:37:55Z | 2023-12-24T04:29:42Z | 2023-12-24T04:29:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YiqunChen1999 | 47,789,930 | MDQ6VXNlcjQ3Nzg5OTMw | User | false |
huggingface/diffusers | 2,030,518,284 | I_kwDOHa8MBc55B0AM | 6,088 | https://github.com/huggingface/diffusers/issues/6088 | https://api.github.com/repos/huggingface/diffusers/issues/6088 | UNet Reusability Issue with IPAdapter in Pipeline | ### Issue description
In my workflow, there are two pipelines: both are instances of `StableDiffusionPipeline`. One of them utilizes the `load_ip_adapter` method.
To conserve resources, I am trying to reuse as many components as possible. Specifically, I want these pipelines to share the UNet.
I am encountering an i... | closed | completed | false | 9 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2023-12-07T11:11:30Z | 2024-10-23T06:39:43Z | 2024-02-11T15:05:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | arlegotin | 1,470,560 | MDQ6VXNlcjE0NzA1NjA= | User | false |
huggingface/diffusers | 2,030,531,253 | I_kwDOHa8MBc55B3K1 | 6,089 | https://github.com/huggingface/diffusers/issues/6089 | https://api.github.com/repos/huggingface/diffusers/issues/6089 | Loading LORA weights with `load_lora_weights` do not respect `HF_HUB_OFFLINE` / `local_file_only` | First reported by @rvorias in https://github.com/huggingface/huggingface_hub/issues/1893#issuecomment-1845134191.
**TL;DR:** Loading lora weights on a pipeline makes a call to the Hub even when `HF_HUB_OFFLINE=1` is set as environment variable (or `local_file_only=True` is passed as parameter), which is not expected... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-12-07T11:16:24Z | 2023-12-16T03:06:10Z | 2023-12-16T03:06:10Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Wauplin | 11,801,849 | MDQ6VXNlcjExODAxODQ5 | User | false |
huggingface/diffusers | 2,030,643,243 | I_kwDOHa8MBc55CSgr | 6,090 | https://github.com/huggingface/diffusers/issues/6090 | https://api.github.com/repos/huggingface/diffusers/issues/6090 | RuntimeError: Given groups=1, weight of size [320, 8, 3, 3], expected input[2, 4, 64, 64] to have 8 channels, but got 4 channels instead | ### Describe the bug
getting an error while try to infer during instruct_pix2pix lora training
### Reproduction
[img2img_lora.zip](https://github.com/huggingface/diffusers/files/13599694/img2img_lora.zip)
run above py file ( basicly edited from **train_instruct_pix2pix_sdxl.py** by added lora training part on it)... | closed | not_planned | false | 3 | [
"bug"
] | [] | 2023-12-07T12:19:16Z | 2023-12-08T01:57:18Z | 2023-12-08T01:57:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DoctorTar | 147,801,017 | U_kgDOCM9DuQ | User | false |
huggingface/diffusers | 2,030,786,410 | I_kwDOHa8MBc55C1dq | 6,091 | https://github.com/huggingface/diffusers/issues/6091 | https://api.github.com/repos/huggingface/diffusers/issues/6091 | Consistency Decoder memory issue with size argument | ### Describe the bug
I have an issue with ` ConsistencyDecoderVAE`. When i pass size argument, the release time is multiplied by 10 and use 99% of my RAM.
Too simple to be true :). Maybe something wrong with my configuration?
Thx,
### Reproduction
```python
import torch
from diffusers import StableDiffusionP... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-12-07T13:40:16Z | 2024-01-18T15:05:23Z | 2024-01-18T15:05:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DamienGaullier | 148,432,890 | U_kgDOCNjn-g | User | false |
huggingface/diffusers | 2,031,003,309 | I_kwDOHa8MBc55Dqat | 6,092 | https://github.com/huggingface/diffusers/issues/6092 | https://api.github.com/repos/huggingface/diffusers/issues/6092 | Lora example does not seem to train anyting | ### Describe the bug
I ran into this bug: https://github.com/huggingface/diffusers/issues/5897
so I used this for the training: https://github.com/huggingface/diffusers/blob/1477865e4838d887bb93750dc325e10f1e6ae534/examples/text_to_image/train_text_to_image_lora.py
### Reproduction
Install:
```
!git clone https:/... | closed | completed | false | 13 | [
"bug",
"stale"
] | [] | 2023-12-07T15:29:49Z | 2024-03-04T10:52:04Z | 2024-02-12T15:06:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kopyl | 17,604,849 | MDQ6VXNlcjE3NjA0ODQ5 | User | false |
huggingface/diffusers | 2,031,053,415 | I_kwDOHa8MBc55D2pn | 6,093 | https://github.com/huggingface/diffusers/issues/6093 | https://api.github.com/repos/huggingface/diffusers/issues/6093 | push_to_hub only create repo in consistency model lora SDXL training script | ### Describe the bug
https://github.com/huggingface/diffusers/blob/main/examples/consistency_distillation/train_lcm_distill_lora_sdxl_wds.py has a push_to_hub argument but all it does currently is create the repo.
### Reproduction
Run the lora example as decribed here: https://github.com/huggingface/diffusers/blob/m... | closed | completed | false | 5 | [
"stale"
] | [
"patil-suraj"
] | 2023-12-07T15:54:56Z | 2024-02-11T04:22:46Z | 2024-02-11T04:22:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | johnowhitaker | 6,575,163 | MDQ6VXNlcjY1NzUxNjM= | User | false |
huggingface/diffusers | 2,031,215,593 | I_kwDOHa8MBc55EePp | 6,094 | https://github.com/huggingface/diffusers/issues/6094 | https://api.github.com/repos/huggingface/diffusers/issues/6094 | TextToVideoZeroPipeline unusable after diffusers 0.24 update | ### Describe the bug
Hello diffusers team,
First of all, thanks for your awesome work !
Here's my problem : after upgrading to diffusers 0.24, i can't succeed loading a model into the **TextToVideoZeroPipeline**.
.from_pretrained fails with **AttributeError: 'bool' object has no attribute '__module__'. Di... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-12-07T17:34:19Z | 2023-12-18T22:53:51Z | 2023-12-18T22:53:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Woolverine94 | 146,110,286 | U_kgDOCLV3Tg | User | false |
huggingface/diffusers | 2,031,342,583 | I_kwDOHa8MBc55E9P3 | 6,095 | https://github.com/huggingface/diffusers/issues/6095 | https://api.github.com/repos/huggingface/diffusers/issues/6095 | SVD Pipeline IndexError in scheduling_euler_discrete | ### Describe the bug
I can consistently reproduce a crash on a asset when generating a video.
The crash happens across a bunch of different random seeds. Asset is attached.
Pipeline is running on an A100 40GB
![drshiba_6](https://github.com/huggingface/diffusers/assets/1213060/a1707082-91cf-4b34-8198-8f64f4af... | closed | completed | false | 12 | [
"bug",
"scheduler"
] | [
"yiyixuxu",
"patil-suraj"
] | 2023-12-07T18:52:27Z | 2024-04-01T09:15:02Z | 2024-04-01T09:15:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | akshetpandey | 1,213,060 | MDQ6VXNlcjEyMTMwNjA= | User | false |
huggingface/diffusers | 2,031,361,638 | I_kwDOHa8MBc55FB5m | 6,096 | https://github.com/huggingface/diffusers/issues/6096 | https://api.github.com/repos/huggingface/diffusers/issues/6096 | SVD torch.compile causes DynamicOutputShapeException | ### Describe the bug
Using torch compile causes SVD pipeline to crash. Without the compile code, it works fine.
Similar to #3915
### Reproduction
```python
import torch
from diffusers import StableVideoDiffusionPipeline
from diffusers.utils.loading_utils import load_image
image = load_image(filenam... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-12-07T19:06:39Z | 2024-02-02T09:16:35Z | 2024-01-27T15:04:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | akshetpandey | 1,213,060 | MDQ6VXNlcjEyMTMwNjA= | User | false |
huggingface/diffusers | 2,031,732,439 | I_kwDOHa8MBc55GcbX | 6,098 | https://github.com/huggingface/diffusers/issues/6098 | https://api.github.com/repos/huggingface/diffusers/issues/6098 | ValueError: Attempting to unscale FP16 gradients | ### Describe the bug
This bug seems to effectively make the script unusable on v100 or lower with the latest update to the dreambooth_lora_sdxl script, I tried a lot with every compromise I could think of (even lowering resolution) but still can't get it to run without running out of memory, using full precision.
##... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-12-08T00:34:08Z | 2024-02-12T15:05:59Z | 2024-02-12T15:05:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rmac85 | 90,567,763 | MDQ6VXNlcjkwNTY3NzYz | User | false |
huggingface/diffusers | 2,032,389,362 | I_kwDOHa8MBc55I8zy | 6,101 | https://github.com/huggingface/diffusers/issues/6101 | https://api.github.com/repos/huggingface/diffusers/issues/6101 | Much worse performance from StableDiffusionControlNetInpaintPipeline than sd-webui-controlnet | Hey folks, I'm getting much worse behavior with Diffusers than A1111 when using ControlNet Inpainting. I'm using the exact same model, seed, inputs, etc. but it's clear the inpainting behavior is very different. Below is one example but I have more if it's helpful.
Lots of artifacts from Diffusers, A1111 essentiall... | closed | completed | false | 26 | [
"stale",
"inpainting"
] | [
"yiyixuxu"
] | 2023-12-08T10:25:25Z | 2024-02-19T15:05:30Z | 2024-02-19T15:05:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | brandonwsaw | 19,142,171 | MDQ6VXNlcjE5MTQyMTcx | User | false |
huggingface/diffusers | 2,033,272,365 | I_kwDOHa8MBc55MUYt | 6,107 | https://github.com/huggingface/diffusers/issues/6107 | https://api.github.com/repos/huggingface/diffusers/issues/6107 | IP adapter - stable diffusion app works in colab but not on AWS EC2 | ### Describe the bug
I am building an app using diffusers, the python backend when I run on colab works great, but when I run it on EC2 I am getting header too large error.
The App uses IP adapter, & the iss... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-12-08T20:26:27Z | 2024-03-19T03:12:46Z | 2023-12-10T06:58:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AyushUnleashed | 77,686,324 | MDQ6VXNlcjc3Njg2MzI0 | User | false |
huggingface/diffusers | 2,033,420,372 | I_kwDOHa8MBc55M4hU | 6,108 | https://github.com/huggingface/diffusers/issues/6108 | https://api.github.com/repos/huggingface/diffusers/issues/6108 | Feature Request: Update Dreambooth training scripts to support loading base model from single safetensors file | **Is your feature request related to a problem? Please describe.**
Current SD and SDXL dreambooth training scripts use the traditional `from_pretrained` method to load the `pretrained_model_name_or_path`. Since diffusers now support saving a fine tune as a single safetensors file, these scripts need to be updated to... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-12-08T23:13:09Z | 2024-01-18T15:05:15Z | 2024-01-18T15:05:15Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | levi | 8,493 | MDQ6VXNlcjg0OTM= | User | false |
huggingface/diffusers | 2,033,522,239 | I_kwDOHa8MBc55NRY_ | 6,109 | https://github.com/huggingface/diffusers/issues/6109 | https://api.github.com/repos/huggingface/diffusers/issues/6109 | ControlNet-xs with depth Control | ### Describe the bug
I have tried to use controlnet-xs pipeline with depth control, but there are some bugs here. I cannot find any instructions on how to use it on the Depth map in diffusers (only a canny image). It would be great if the author can provide some instructions on Control-XS on depth map @sayakpaul
My... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-12-09T01:27:47Z | 2023-12-09T01:43:52Z | 2023-12-09T01:43:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | George0726 | 38,740,075 | MDQ6VXNlcjM4NzQwMDc1 | User | false |
huggingface/diffusers | 2,033,548,709 | I_kwDOHa8MBc55NX2l | 6,110 | https://github.com/huggingface/diffusers/issues/6110 | https://api.github.com/repos/huggingface/diffusers/issues/6110 | The load_lora_weights is not working offline | ### Describe the bug
The `load_lora_weights ` is not working offline.
### Reproduction
Sample code to reproduce this issue. (Turn off the internet)
```py
from diffusers import DiffusionPipeline, LCMScheduler
pipeline = DiffusionPipeline.from_pretrained(
"Lykon/dreamshaper-8",
local_files_... | closed | completed | false | 20 | [
"bug"
] | [] | 2023-12-09T02:32:22Z | 2023-12-16T04:37:54Z | 2023-12-16T04:37:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rupeshs | 3,255,994 | MDQ6VXNlcjMyNTU5OTQ= | User | false |
huggingface/diffusers | 2,033,726,711 | I_kwDOHa8MBc55ODT3 | 6,112 | https://github.com/huggingface/diffusers/issues/6112 | https://api.github.com/repos/huggingface/diffusers/issues/6112 | The picture is full of noise use sdxl-turbo | ### Describe the bug
from diffusers import StableDiffusionXLPipeline
import torch
pipeline = StableDiffusionXLPipeline.from_single_file(
r"sd_xl_turbo_1.0_fp16.safetensors", torch_dtype=torch.float16, variant="fp16")
pipeline = pipeline.to("cuda")
prompt = "A cinematic shot of a baby racoon wearing an int... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2023-12-09T08:27:08Z | 2024-02-28T12:04:48Z | 2024-01-18T15:05:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyun0071 | 35,762,050 | MDQ6VXNlcjM1NzYyMDUw | User | false |
huggingface/diffusers | 2,033,866,664 | I_kwDOHa8MBc55Oleo | 6,113 | https://github.com/huggingface/diffusers/issues/6113 | https://api.github.com/repos/huggingface/diffusers/issues/6113 | How to use the models from sd_control_collection hf repo in diffusers | How to load/convert the models at https://huggingface.co/lllyasviel/sd_control_collection/tree/main with diffusers?
```
>>> pipe = diffusers.StableDiffusionPipeline.from_single_file("diffusers_xl_canny_full.safetensors")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/ubunt... | closed | completed | false | 20 | [] | [] | 2023-12-09T14:11:26Z | 2024-06-11T18:22:03Z | 2024-02-24T07:14:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anilsathyan7 | 1,130,185 | MDQ6VXNlcjExMzAxODU= | User | false |
huggingface/diffusers | 2,034,155,918 | I_kwDOHa8MBc55PsGO | 6,117 | https://github.com/huggingface/diffusers/issues/6117 | https://api.github.com/repos/huggingface/diffusers/issues/6117 | Scheduler `CMStochasticIterativeScheduler` `add_noise` bug | ### Describe the bug
The `add_noise` function of the consistency model scheduler, `CMStochasticIterativeScheduler`, raises a runtime error in training due to inconsistency between scheduler timestep and timestep in the function argument.
`step_indices = [(schedule_timesteps == t).nonzero().item() for t in timesteps... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-12-10T00:55:47Z | 2024-01-18T15:05:09Z | 2024-01-18T15:05:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | matinmoezzi | 15,571,593 | MDQ6VXNlcjE1NTcxNTkz | User | false |
huggingface/diffusers | 2,034,218,066 | I_kwDOHa8MBc55P7RS | 6,118 | https://github.com/huggingface/diffusers/issues/6118 | https://api.github.com/repos/huggingface/diffusers/issues/6118 | [Advanced Diffusion Training] Adapt `train_dreambooth_lora_sdxl_advanced.py` to use `peft` | With #5388 we adapted our training examples that make use of LoRA to use `peft`. It'd be very nice to do the same for `advaced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py`.
Unable to assign @linoytsaban for some reason.
Cc: @younesbelkada for visibility. | closed | completed | false | 0 | [] | [
"apolinario"
] | 2023-12-10T04:45:21Z | 2023-12-27T13:00:34Z | 2023-12-27T13:00:34Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,381,564,186 | I_kwDOHa8MBc5SWP8a | 612 | https://github.com/huggingface/diffusers/issues/612 | https://api.github.com/repos/huggingface/diffusers/issues/612 | Latents / seeds are a mess. Make it easier to replicate a generated image using a seed. | **Problem:**
We often generate images with a batch_size >1.
However, images in the batch (after the first image) by default have an **seed that is unknown to the user**, so all but the first image in a batch can't be directly replicated.
To get around this, the [docs ](https://github.com/huggingface/diffusers/... | closed | completed | false | 20 | [
"stale"
] | [] | 2022-09-21T22:06:48Z | 2022-12-17T10:13:17Z | 2022-12-17T10:13:17Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | exo-pla-net | 47,930,742 | MDQ6VXNlcjQ3OTMwNzQy | User | false |
huggingface/diffusers | 2,034,233,908 | I_kwDOHa8MBc55P_I0 | 6,120 | https://github.com/huggingface/diffusers/issues/6120 | https://api.github.com/repos/huggingface/diffusers/issues/6120 | Hi, I'm ComfyUI newb user, i've got Error occurred when executing IPAdapterApply | Sorry wrong forums thanks | closed | completed | false | 0 | [] | [] | 2023-12-10T06:05:35Z | 2023-12-11T09:16:05Z | 2023-12-11T09:16:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | IMNASHR | 118,431,017 | U_kgDOBw8dKQ | User | false |
huggingface/diffusers | 2,034,436,404 | I_kwDOHa8MBc55Qwk0 | 6,123 | https://github.com/huggingface/diffusers/issues/6123 | https://api.github.com/repos/huggingface/diffusers/issues/6123 | AnimateDiffPipeline: add ImageToVideo and VideoToVideo | Can adapt the work in https://github.com/Kosinkadink/ComfyUI-AnimateDiff-Evolved | closed | completed | false | 7 | [
"stale"
] | [] | 2023-12-10T16:18:32Z | 2024-01-19T10:05:43Z | 2024-01-19T10:05:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jon-chuang | 9,093,549 | MDQ6VXNlcjkwOTM1NDk= | User | false |
huggingface/diffusers | 2,034,468,261 | I_kwDOHa8MBc55Q4Wl | 6,124 | https://github.com/huggingface/diffusers/issues/6124 | https://api.github.com/repos/huggingface/diffusers/issues/6124 | Issue "ValueError: Attempting to unscale FP16 gradients." when running SDXL_DreamBooth_LoRA_.ipynb | ### Describe the bug
I am trying to run the famous colab notebook SDXL_DreamBooth_LoRA_.ipynb to build a dreambooth model out of sdxl + vae using accelerate launch train_dreambooth_lora_sdxl.py.
Yet, i am ending with this error which is frustrating.
### Reproduction
Running only this notebook without changes : ht... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-10T17:48:31Z | 2024-02-12T15:05:55Z | 2024-02-12T15:05:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yleoed | 120,102,914 | U_kgDOByigAg | User | false |
huggingface/diffusers | 2,034,503,739 | I_kwDOHa8MBc55RBA7 | 6,125 | https://github.com/huggingface/diffusers/issues/6125 | https://api.github.com/repos/huggingface/diffusers/issues/6125 | UnetMotionModel ignores important parameters | ### Describe the bug
```
{'center_input_sample': False, 'flip_sin_to_cos': True, 'freq_shift': 0, 'mid_block_type': 'UNetMidBlock2DCrossAttn', 'only_cross_attention': False, 'dropout': 0.0, 'transformer_layers_per_block': 1, 'attention_head_dim': 8, 'dual_cross_attention': False, 'class_embed_type': None, 'addition_e... | closed | completed | false | 13 | [
"bug",
"stale",
"contributions-welcome"
] | [
"DN6"
] | 2023-12-10T19:33:50Z | 2024-03-17T15:26:26Z | 2024-03-17T15:26:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jon-chuang | 9,093,549 | MDQ6VXNlcjkwOTM1NDk= | User | false |
huggingface/diffusers | 2,034,527,027 | I_kwDOHa8MBc55RGsz | 6,126 | https://github.com/huggingface/diffusers/issues/6126 | https://api.github.com/repos/huggingface/diffusers/issues/6126 | AttributeError: 'ConsistencyDecoderVAE' object has no attribute 'tile_sample_min_size' | ### Describe the bug
The new decoder has enable_tiling and disable_tiling methods but has not correctly implemented the support attributes.
### Reproduction
```python
import torch
from diffusers import ConsistencyDecoderVAE
vae = ConsistencyDecoderVAE.from_pretrained("openai/consistency-decoder")
vae.enable_tili... | closed | completed | false | 7 | [
"bug",
"contributions-welcome"
] | [] | 2023-12-10T20:45:54Z | 2024-03-26T12:29:09Z | 2024-03-26T12:29:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | eliphatfs | 23,738,781 | MDQ6VXNlcjIzNzM4Nzgx | User | false |
huggingface/diffusers | 2,035,056,830 | I_kwDOHa8MBc55TIC- | 6,130 | https://github.com/huggingface/diffusers/issues/6130 | https://api.github.com/repos/huggingface/diffusers/issues/6130 | Error while loading Lora | ### Describe the bug
According to the [guidance ](https://huggingface.co/docs/diffusers/main/en/training/lora), I try to finetune the SD with lora, and use the lora weight, but the issue appears.
```
Traceback (most recent call last):
File "AutoPipelineForTex.py", line 17, in <module>
pipe.unet.load_attn_p... | closed | completed | false | 19 | [
"bug"
] | [] | 2023-12-11T08:03:46Z | 2023-12-14T22:49:42Z | 2023-12-12T02:02:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | istoneyou | 5,011,538 | MDQ6VXNlcjUwMTE1Mzg= | User | false |
huggingface/diffusers | 2,035,357,836 | I_kwDOHa8MBc55URiM | 6,131 | https://github.com/huggingface/diffusers/issues/6131 | https://api.github.com/repos/huggingface/diffusers/issues/6131 | IP adapter plus not working | ### Describe the bug
```
clip_embeddings_dim = state_dict["image_proj"]["proj.weight"].shape[-1]
KeyError: 'proj.weight'
```
### Reproduction
image1 = Image.open('a.jpg').convert('RGB')
pipe = StableDiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
)
pipe.to(torch_dev... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-12-11T10:38:52Z | 2024-01-19T15:05:00Z | 2024-01-19T15:05:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 2,035,404,135 | I_kwDOHa8MBc55Uc1n | 6,132 | https://github.com/huggingface/diffusers/issues/6132 | https://api.github.com/repos/huggingface/diffusers/issues/6132 | Add Pytorch's nn.Module LOADABLE_CLASSES | ### Model/Pipeline/Scheduler description
In class [diffusers.DiffusionPipeline](https://github.com/huggingface/diffusers/blob/v0.24.0/src/diffusers/pipelines/pipeline_utils.py#L528) there is a method [`save_pretrained`](https://github.com/huggingface/diffusers/blob/76c645d3a641c879384afcb43496f0b7db8cc5cb/src/diffuser... | closed | completed | false | 9 | [
"stale"
] | [] | 2023-12-11T10:58:33Z | 2024-01-29T15:48:28Z | 2024-01-29T15:48:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | luisaforozco | 99,738,896 | U_kgDOBfHlEA | User | false |
huggingface/diffusers | 2,035,623,996 | I_kwDOHa8MBc55VSg8 | 6,134 | https://github.com/huggingface/diffusers/issues/6134 | https://api.github.com/repos/huggingface/diffusers/issues/6134 | Irrelevant and repetitive outputs in stable diffusion | I trained a model with over 350 images on Colab dreambooth but the outputs weren't upto the mark, I did the same with limited images, the outputs were better. What can be the reasons? What are the things to consider while training images using Colab? | closed | completed | false | 3 | [
"stale"
] | [] | 2023-12-11T12:56:57Z | 2024-01-18T15:04:54Z | 2024-01-18T15:04:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dnyaneshthombre | 153,517,673 | U_kgDOCSZ-aQ | User | false |
huggingface/diffusers | 2,035,687,211 | I_kwDOHa8MBc55Vh8r | 6,136 | https://github.com/huggingface/diffusers/issues/6136 | https://api.github.com/repos/huggingface/diffusers/issues/6136 | `StableDiffusionPipeline` produces `nans` for SD2.1 (Mac, cpu device) | ### Describe the bug
`StableDiffusionPipeline` produces `nans` for SD2.1.
The issue seems to be in the `pipe.unet.conv_in`. Printing the mean / std for `sample` before and after `... | open | reopened | false | 15 | [
"bug",
"help wanted",
"stale",
"contributions-welcome"
] | [] | 2023-12-11T13:30:41Z | 2024-12-13T15:08:22Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | UmerHA | 40,663,591 | MDQ6VXNlcjQwNjYzNTkx | User | false |
huggingface/diffusers | 2,035,689,920 | I_kwDOHa8MBc55VinA | 6,137 | https://github.com/huggingface/diffusers/issues/6137 | https://api.github.com/repos/huggingface/diffusers/issues/6137 | Inaccurate `num_train_steps` in the LR scheduler during multi-GPU parallelism. | ### Describe the bug
When I was building my own LoRa training code, I encountered an issue with the `num_training_steps` parameter in the LR scheduler. I attempted to follow the official training script from [diffusers](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py#... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-11T13:32:02Z | 2024-01-18T15:04:50Z | 2024-01-18T15:04:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dai-Wenxun | 58,767,402 | MDQ6VXNlcjU4NzY3NDAy | User | false |
huggingface/diffusers | 2,035,926,982 | I_kwDOHa8MBc55WcfG | 6,139 | https://github.com/huggingface/diffusers/issues/6139 | https://api.github.com/repos/huggingface/diffusers/issues/6139 | AttributeError: 'unet2dconditionmodel' object has no attribute 'attn_processors'. | ### Describe the bug
This error happens with a custom attention processor after setting lora layers
### Reproduction
```
from diffusers import UNet2DConditionModel
from diffusers.models.lora import LoRALinearLayer
class AttnProcessorCustom:
def __call__(
self,
attn,
... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-12-11T15:24:48Z | 2024-02-19T15:05:23Z | 2024-02-19T15:05:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kamwoh | 17,477,159 | MDQ6VXNlcjE3NDc3MTU5 | User | false |
huggingface/diffusers | 2,036,798,069 | I_kwDOHa8MBc55ZxJ1 | 6,141 | https://github.com/huggingface/diffusers/issues/6141 | https://api.github.com/repos/huggingface/diffusers/issues/6141 | Diffuser ControlNet support the two parameters control_mode and resize_mode | When will Diffuser ControlNet support the two parameters control_mode and resize_mode like sd-webui-controlnet?
Is your feature request related to a problem? Please describe.
I'd like more options for controlnet usage
Describe the solution you'd like
implement this: https://github.com/Mikubill/sd-webui-controln... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-12-12T01:12:15Z | 2024-01-27T15:04:19Z | 2024-01-27T15:04:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yan-Ops | 29,395,050 | MDQ6VXNlcjI5Mzk1MDUw | User | false |
huggingface/diffusers | 2,036,858,184 | I_kwDOHa8MBc55Z_1I | 6,142 | https://github.com/huggingface/diffusers/issues/6142 | https://api.github.com/repos/huggingface/diffusers/issues/6142 | NameError: name 'start_training' is not defined | ### Describe the bug
thank you for your great job!
Run dreambooth_train.py, I meet this wrong.
### Reproduction
python3 dreambooth_train.py
### Logs
```shell
Traceback (most recent call last):
File "/home/masuxin/Dsd1.5/sd_dreambooth_extension/scripts/api.py", line 28, in <module>
from dreambooth.diff_to_... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-12T02:29:11Z | 2024-01-26T15:06:25Z | 2024-01-26T15:06:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | masuxin | 66,507,490 | MDQ6VXNlcjY2NTA3NDkw | User | false |
huggingface/diffusers | 2,037,574,093 | I_kwDOHa8MBc55cunN | 6,146 | https://github.com/huggingface/diffusers/issues/6146 | https://api.github.com/repos/huggingface/diffusers/issues/6146 | Training "train_dreambooth_lora_sdxl.py" on multi-gpu using accelerate | ### Describe the bug
HI, I'm trying to train the train_dreambooth_lora_sdxl.py on multi-gpu. This is my accelerate config, and the error that I got is- 'DistributedDataParallel' object has no attribute 'text_model'. I found out that using multi-gpu create DistributedDataParallel object instade of CLIPTextModel but I w... | closed | completed | false | 7 | [
"bug"
] | [
"sayakpaul"
] | 2023-12-12T11:41:15Z | 2024-02-03T04:16:52Z | 2024-02-03T04:16:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,037,887,392 | I_kwDOHa8MBc55d7Gg | 6,149 | https://github.com/huggingface/diffusers/issues/6149 | https://api.github.com/repos/huggingface/diffusers/issues/6149 | ip-adapter-full-face_sd15 load error | ### Describe the bug
`pipeline.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter-full-face_sd15.bin")`
diffusers\loaders\unet.py", line 711, in _load_ip_adapter_weights
clip_embeddings_dim = state_dict["image_proj"]["proj.weight"].shape[-1]
KeyError: 'proj.weight'
### Reproduc... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-12-12T14:35:58Z | 2024-01-26T15:06:21Z | 2024-01-26T15:06:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyun0071 | 35,762,050 | MDQ6VXNlcjM1NzYyMDUw | User | false |
huggingface/diffusers | 1,382,412,107 | I_kwDOHa8MBc5SZe9L | 615 | https://github.com/huggingface/diffusers/issues/615 | https://api.github.com/repos/huggingface/diffusers/issues/615 | Is there any outpainting pipe demo? | I would like to use it to outpainting images, can you help me | closed | completed | false | 8 | [
"stale"
] | [] | 2022-09-22T12:59:12Z | 2023-08-07T04:07:54Z | 2022-11-05T15:02:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CrazyBoyM | 35,400,185 | MDQ6VXNlcjM1NDAwMTg1 | User | false |
huggingface/diffusers | 2,038,584,015 | I_kwDOHa8MBc55glLP | 6,152 | https://github.com/huggingface/diffusers/issues/6152 | https://api.github.com/repos/huggingface/diffusers/issues/6152 | sdxl + controlnet out of memory in free colab | on a free colab instance comfyui loads sdxl and controlnet without problems, but diffusers can't seem to handle this and causes an out of memory.
Is there any option or parameter in diffusers to make sdxl and controlnet work in colab for free?
It seems strange to me that comnfyui can handle this and diffusers can't.
... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-12-12T21:35:34Z | 2024-01-26T15:06:20Z | 2024-01-26T15:06:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | loboere | 15,265,895 | MDQ6VXNlcjE1MjY1ODk1 | User | false |
huggingface/diffusers | 2,038,587,838 | I_kwDOHa8MBc55gmG- | 6,153 | https://github.com/huggingface/diffusers/issues/6153 | https://api.github.com/repos/huggingface/diffusers/issues/6153 | Reduce huge amounts of code duplication | **Is your feature request related to a problem? Please describe.**
There is a _ridiculous_ amount of code duplication going on in this library. Could anyone please tell me why the schedulers function `betas_for_alpha_bar` is duplicated - identically - **23 times**, appearing in about half of the schedulers files? Ther... | closed | completed | false | 4 | [] | [] | 2023-12-12T21:38:59Z | 2024-03-30T02:03:36Z | 2023-12-14T22:32:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zzbuzzard | 40,760,215 | MDQ6VXNlcjQwNzYwMjE1 | User | false |
huggingface/diffusers | 2,038,666,905 | I_kwDOHa8MBc55g5aZ | 6,154 | https://github.com/huggingface/diffusers/issues/6154 | https://api.github.com/repos/huggingface/diffusers/issues/6154 | Can't Train SDXL with LoRA on M2 Mac Mini Using Script | ### Describe the bug
Using the instructions in the ```examples/text_to_image``` folder's ```readme_sdxl.md``` file, I tried to train a Stable Diffusion XL model on a dataset with mixed precision, but failed because Autocast is unsupported on MPS. I cannot train in full precision because then the MPS backend OOMs.
###... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-12T22:54:45Z | 2024-01-26T15:06:18Z | 2024-01-26T15:06:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sr5434 | 118,690,585 | U_kgDOBxMTGQ | User | false |
huggingface/diffusers | 2,038,886,175 | I_kwDOHa8MBc55hu8f | 6,156 | https://github.com/huggingface/diffusers/issues/6156 | https://api.github.com/repos/huggingface/diffusers/issues/6156 | it seems that the combination of lcm-lora and style lora on finetuned model will decrease the effect of style lora? | ### Describe the bug
I fused the lcm-lora into the model in Civitai, then fused my own trained LoRa. There is a noticeable difference in the effect of fusing my own LoRa directly on Civitai compared to fusing it after lcm-lora. The original direct fusion of my own LoRa has a more distinct style, while the difference i... | closed | completed | false | 6 | [
"bug",
"stale"
] | [
"patil-suraj"
] | 2023-12-13T03:29:46Z | 2024-01-27T03:54:18Z | 2024-01-27T03:54:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhangvia | 38,352,569 | MDQ6VXNlcjM4MzUyNTY5 | User | false |
huggingface/diffusers | 2,038,909,434 | I_kwDOHa8MBc55h0n6 | 6,157 | https://github.com/huggingface/diffusers/issues/6157 | https://api.github.com/repos/huggingface/diffusers/issues/6157 | why ip_adapter not work in controlnetimg2imgpipeline? | ### Describe the bug
When using the controlnetimg2img pipe, id_adapter is not work.
### Reproduction
from diffusers import StableDiffusionControlNetImg2ImgPipeline, ControlNetModel
from diffusers import DDIMScheduler
import torch
controlnet = ControlNetModel.from_pretrained("./controlnet/", torch_dtype=torch.fl... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-12-13T04:02:56Z | 2024-01-26T15:06:09Z | 2024-01-26T15:06:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | htwdls | 35,591,552 | MDQ6VXNlcjM1NTkxNTUy | User | false |
huggingface/diffusers | 2,039,121,820 | I_kwDOHa8MBc55ioec | 6,158 | https://github.com/huggingface/diffusers/issues/6158 | https://api.github.com/repos/huggingface/diffusers/issues/6158 | Feature Request: add SDXL support to animatediff pipeline | **Is your feature request related to a problem? Please describe.**
Currently the AnimateDiff pipeline only supports SD 1.4 and SD 1.5. AnimateDiff with SDXL seems like a great addition as the quality is greater.
**Describe the solution you'd like.**
Currently the MotionAdapter cannot load the SDXL motion module : ... | closed | completed | false | 3 | [] | [] | 2023-12-13T07:31:43Z | 2024-02-05T15:58:16Z | 2024-02-04T15:05:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | madriss | 13,274,661 | MDQ6VXNlcjEzMjc0NjYx | User | false |
huggingface/diffusers | 2,039,496,062 | I_kwDOHa8MBc55kD1- | 6,159 | https://github.com/huggingface/diffusers/issues/6159 | https://api.github.com/repos/huggingface/diffusers/issues/6159 | StableDiffusionXLControlNetXSPipeline does not support keyword argument 'denoising_end' | ### Describe the bug
StableDiffusionXLControlNetXSPipeline does not support 'denoising_end' for use in combination with the SDXL refiner pipeline.
```
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
[<ipyth... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-12-13T11:18:32Z | 2024-03-09T10:46:02Z | 2024-03-09T02:42:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vargol | 62,868 | MDQ6VXNlcjYyODY4 | User | false |
huggingface/diffusers | 2,039,575,365 | I_kwDOHa8MBc55kXNF | 6,160 | https://github.com/huggingface/diffusers/issues/6160 | https://api.github.com/repos/huggingface/diffusers/issues/6160 | Textual Inversion Colab trainer Error | ### Describe the bug
When launching the import accelerate code on the google colab textual inversion trainer I keep getting the error: RuntimeError: Expected is_sm80 || is_sm90 to be true, but got false. (Could this error message be improved? If so, please report an enhancement request to PyTorch.)
I haven't be... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-12-13T12:06:46Z | 2024-01-05T02:20:35Z | 2024-01-05T02:20:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | domochevisk | 114,751,551 | U_kgDOBtb4Pw | User | false |
huggingface/diffusers | 2,039,595,332 | I_kwDOHa8MBc55kcFE | 6,161 | https://github.com/huggingface/diffusers/issues/6161 | https://api.github.com/repos/huggingface/diffusers/issues/6161 | always discard next-to-last sigma for diffusers | **Is your feature request related to a problem? Please describe.**
There's a feature from A1111 that I want to use in diffusers
**Describe the solution you'd like.**
I'd like it implemented or instructions on how to implement it
**Describe alternatives you've considered.**
figuring it out myself
**Additiona... | closed | completed | false | 14 | [] | [] | 2023-12-13T12:18:57Z | 2024-11-24T09:53:04Z | 2024-11-24T09:53:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 2,039,625,473 | I_kwDOHa8MBc55kjcB | 6,162 | https://github.com/huggingface/diffusers/issues/6162 | https://api.github.com/repos/huggingface/diffusers/issues/6162 | IP_Adapters shape mismatch when generating images on v0.25.0_dev using SDXL? | ### Describe the bug
When using ip_adapters with controlnets and sdxl (whether sdxl-turbo or sdxl1.0) you get a shape mismatch when generating images. If you remove the ip_adapter things start working again. Not sure what the problem might be here?
### Reproduction
Here's what I'm doing:
```
from diffusers imp... | closed | completed | false | 12 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2023-12-13T12:37:11Z | 2024-02-07T15:05:20Z | 2024-02-07T15:05:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | salahzoubi | 41,164,256 | MDQ6VXNlcjQxMTY0MjU2 | User | false |
huggingface/diffusers | 2,040,255,023 | I_kwDOHa8MBc55m9Iv | 6,164 | https://github.com/huggingface/diffusers/issues/6164 | https://api.github.com/repos/huggingface/diffusers/issues/6164 | Output type "latent" not correctly handled by StableDiffusionPipeline | ### Describe the bug
The issue is similar to what is described here: https://github.com/huggingface/diffusers/issues/2871. Namely, when using `output_type="latent"`, the returned latent cannot be successfully passed back in to the `__call__` method of `StableDiffusionPipeline` (perhaps a result of [this line?](https... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-12-13T18:27:47Z | 2023-12-15T01:05:46Z | 2023-12-15T01:05:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 2014mchidamb | 8,584,333 | MDQ6VXNlcjg1ODQzMzM= | User | false |
huggingface/diffusers | 2,040,322,316 | I_kwDOHa8MBc55nNkM | 6,165 | https://github.com/huggingface/diffusers/issues/6165 | https://api.github.com/repos/huggingface/diffusers/issues/6165 | FreeInit | ### Model/Pipeline/Scheduler description
FreeInit allows for improved temporal consistency of diffusion-based video models (for example, AnimateDiff and SVD). With no additional training requirement, it produces much higher quality videos.
Going through the code, it seems like an easy addition to current diffuser... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-12-13T19:17:46Z | 2024-01-17T11:47:09Z | 2024-01-17T11:47:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | a-r-r-o-w | 72,266,394 | MDQ6VXNlcjcyMjY2Mzk0 | User | false |
huggingface/diffusers | 2,040,883,760 | I_kwDOHa8MBc55pWow | 6,167 | https://github.com/huggingface/diffusers/issues/6167 | https://api.github.com/repos/huggingface/diffusers/issues/6167 | The addition of 'load_ip_adapter' to the StableDiffusionXLAdapterPipeline | **What API design would you like to have changed or added to the library? Why?**
It would be great to have IP-adapters work with t2i-adapter as they're usually faster and less heavy on the image generation process than controlnet.
**What use case would this enable or better enable? Can you give us a code example?*... | closed | completed | false | 2 | [] | [] | 2023-12-14T04:31:43Z | 2024-01-11T09:01:39Z | 2024-01-11T09:01:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jquintanilla4 | 32,947,277 | MDQ6VXNlcjMyOTQ3Mjc3 | User | false |
huggingface/diffusers | 2,040,890,197 | I_kwDOHa8MBc55pYNV | 6,168 | https://github.com/huggingface/diffusers/issues/6168 | https://api.github.com/repos/huggingface/diffusers/issues/6168 | Working offline automatically after model has downloaded | I have a computer that has an unstable internet connection. When I run my diffusers script, I would like the model to download if it is not already downloaded. However, after the first time when it downloads, I would like the script to run even if the computer is offline.
I found that using [local_files_only](https:... | closed | completed | false | 16 | [] | [] | 2023-12-14T04:40:09Z | 2024-10-25T14:47:44Z | 2024-01-04T16:05:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kylemcdonald | 157,106 | MDQ6VXNlcjE1NzEwNg== | User | false |
huggingface/diffusers | 1,382,589,211 | I_kwDOHa8MBc5SaKMb | 617 | https://github.com/huggingface/diffusers/issues/617 | https://api.github.com/repos/huggingface/diffusers/issues/617 | AutoencoderKL: [W NNPACK.cpp:51] Could not initialize NNPACK! Reason: Unsupported hardware. | ### Describe the bug
When on macOS Apple M1 Silicon,
```python
import torch
import numpy as np
from torchvision.io import read_image
from diffusers import AutoencoderKL
vae = AutoencoderKL.from_pretrained(
"CompVis/stable-diffusion-v1-4",
subfolder="vae",
torch_dtype=torch.float32,
revisi... | closed | completed | false | 1 | [
"bug"
] | [
"pcuenca"
] | 2022-09-22T14:53:43Z | 2022-10-20T13:36:49Z | 2022-10-20T13:36:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | loretoparisi | 163,333 | MDQ6VXNlcjE2MzMzMw== | User | false |
huggingface/diffusers | 2,041,432,134 | I_kwDOHa8MBc55rchG | 6,170 | https://github.com/huggingface/diffusers/issues/6170 | https://api.github.com/repos/huggingface/diffusers/issues/6170 | Inquiry on diffusion model training using paired data | **Is your feature request related to a problem? Please describe.**
The training data of existing diffusion models all use pictures of the same style and then train. Isn't this essentially style transfer?
I would like to ask if there is any training data using paired data to train the diffusion model. One round of t... | closed | completed | false | 2 | [] | [] | 2023-12-14T10:46:07Z | 2023-12-15T10:09:21Z | 2023-12-15T06:52:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | GuinanGuo | 115,399,025 | U_kgDOBuDZcQ | User | false |
huggingface/diffusers | 2,042,143,650 | I_kwDOHa8MBc55uKOi | 6,172 | https://github.com/huggingface/diffusers/issues/6172 | https://api.github.com/repos/huggingface/diffusers/issues/6172 | Text-to-image overfitting | Hi,
I have fine-tuned the [text_to_image](https://huggingface.co/docs/diffusers/training/text2image?installation=PyTorch) model on my dataset, but the images that are produced by the prompt are simply the same images that I trained the model with. In other words, the model is overfitting to my data.
My dataset is s... | closed | completed | false | 1 | [] | [] | 2023-12-14T17:22:07Z | 2023-12-15T06:52:10Z | 2023-12-15T06:52:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alireza18878 | 82,774,346 | MDQ6VXNlcjgyNzc0MzQ2 | User | false |
huggingface/diffusers | 2,042,514,221 | I_kwDOHa8MBc55vkst | 6,176 | https://github.com/huggingface/diffusers/issues/6176 | https://api.github.com/repos/huggingface/diffusers/issues/6176 | Error when load LoRA weight after training dreambooth_lora | ### Describe the bug
Dear All,
I have use the script and md file to train dreambooth_LoRA.py and I try to use the method you inference. https://github.com/huggingface/diffusers/tree/main/examples/dreambooth
### Reproduction
This is the repo I use after the update of [train_dreambooth_lora.py](https://github.c... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-12-14T21:15:15Z | 2024-01-26T15:06:00Z | 2024-01-26T15:06:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CaffreyR | 84,232,793 | MDQ6VXNlcjg0MjMyNzkz | User | false |
huggingface/diffusers | 2,042,794,945 | I_kwDOHa8MBc55wpPB | 6,178 | https://github.com/huggingface/diffusers/issues/6178 | https://api.github.com/repos/huggingface/diffusers/issues/6178 | How to train Stable Diffusion with DDPM? | I want to train Stable Diffusion with DDPM, but I can't find the code in this project. I found a lot of training code elsewhere on the internet, but most of it is distillation code on pre-trained models, not the original DDPM training code. I also tried to implement the original training code myself, but I couldn't get... | closed | completed | false | 0 | [] | [] | 2023-12-15T02:43:07Z | 2023-12-15T02:54:06Z | 2023-12-15T02:54:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MenSanYan | 24,693,012 | MDQ6VXNlcjI0NjkzMDEy | User | false |
huggingface/diffusers | 2,042,889,861 | I_kwDOHa8MBc55xAaF | 6,180 | https://github.com/huggingface/diffusers/issues/6180 | https://api.github.com/repos/huggingface/diffusers/issues/6180 | [IP-Adapter] add a 'unload_ip_adapter' method | fix https://github.com/huggingface/diffusers/issues/6088
Is anyone interested in adding a `unload_ip_adapter` method? it needs to undo everything we did with `load_ip_adapter` method:
https://github.com/huggingface/diffusers/blob/93ea26f272f69a0e27afaebc96b68a2221a7eda0/src/diffusers/loaders/ip_adapter.py#L46
It... | closed | completed | false | 4 | [
"good first issue",
"help wanted",
"Good second issue",
"contributions-welcome"
] | [] | 2023-12-15T04:44:30Z | 2024-01-02T13:40:48Z | 2024-01-02T13:40:48Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 2,043,073,970 | I_kwDOHa8MBc55xtWy | 6,182 | https://github.com/huggingface/diffusers/issues/6182 | https://api.github.com/repos/huggingface/diffusers/issues/6182 | Unexpected Behavior of the UNet1DModel | ### Describe the bug
The UNet1DModel class is not functioning correctly.
Upon investigation, the problem is as follows: Within the UNet1DModel class, there is a call to the get_down_block function.
When the input parameter down_block_type is set to "DownBlock1DNoSkip," the function returns an object instantiate... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-15T07:44:43Z | 2024-01-26T15:05:54Z | 2024-01-26T15:05:54Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jinc7461 | 126,931,906 | U_kgDOB5DTwg | User | false |
huggingface/diffusers | 2,043,111,009 | I_kwDOHa8MBc55x2Zh | 6,183 | https://github.com/huggingface/diffusers/issues/6183 | https://api.github.com/repos/huggingface/diffusers/issues/6183 | Incorrect shape when using StableDiffusionXLPipeline.from_single_file | ### Describe the bug
I was trying to load the single `safetensors` from [playground_v2](https://huggingface.co/playgroundai/playground-v2-1024px-aesthetic) by calling `StableDiffusionXLPipeline.from_single_file`. The function failed with error message:
```
Traceback (most recent call last):
File "workspace/sdxl/p... | closed | completed | false | 11 | [
"bug"
] | [
"DN6"
] | 2023-12-15T08:09:16Z | 2024-01-04T04:04:33Z | 2024-01-04T04:04:32Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dalanke | 25,289,727 | MDQ6VXNlcjI1Mjg5NzI3 | User | false |
huggingface/diffusers | 2,043,258,390 | I_kwDOHa8MBc55yaYW | 6,185 | https://github.com/huggingface/diffusers/issues/6185 | https://api.github.com/repos/huggingface/diffusers/issues/6185 | [LoRA fusion tests] SDXL LoRA tests with fusion when using `peft` | Currently, the following test is failing:
```bash
RUN_SLOW=1 pytest tests/lora/test_lora_layers_peft.py::LoraSDXLIntegrationTests::test_sdxl_1_0_fuse_unfuse_all
```
Error:
```bash
self = <test_lora_layers_peft.LoraSDXLIntegrationTests testMethod=test_sdxl_1_0_fuse_unfuse_all>
def test_sdxl_1_0_fuse... | closed | completed | false | 9 | [] | [] | 2023-12-15T09:17:34Z | 2023-12-21T10:50:07Z | 2023-12-21T10:50:07Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,043,346,377 | I_kwDOHa8MBc55yv3J | 6,186 | https://github.com/huggingface/diffusers/issues/6186 | https://api.github.com/repos/huggingface/diffusers/issues/6186 | i2vgen-xl implement | ### Model/Pipeline/Scheduler description
I2VGEN-XL recently disclosed their training and testing codes for Wensheng videos and Tusheng videos. Does diffusers have any plans to integrate these models and codes?
Code: [https://github.com/damo-vilab/i2vgen-xl?tab=readme-ov-file](url)
### Open source status
- [X] Th... | closed | completed | false | 6 | [
"contributions-welcome"
] | [] | 2023-12-15T09:57:31Z | 2024-01-31T20:38:52Z | 2024-01-31T20:38:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhw-zhang | 135,094,935 | U_kgDOCA1ilw | User | false |
huggingface/diffusers | 2,044,135,361 | I_kwDOHa8MBc551wfB | 6,193 | https://github.com/huggingface/diffusers/issues/6193 | https://api.github.com/repos/huggingface/diffusers/issues/6193 | [New Pipeline]: DeepCache: Accelerating Diffusion Models for Free | Repo: https://github.com/horseee/DeepCache
Project Page: https://horseee.github.io/Diffusion_DeepCache/
@patrickvonplaten @sayakpaul
I tested it locally and it is a very fast pipeline. But I need help adding it to the diffusers library. Should I add this to the community pipeline section?
I just need to updat... | closed | completed | false | 16 | [
"stale"
] | [] | 2023-12-15T17:53:53Z | 2024-11-18T05:47:20Z | 2024-01-27T03:44:17Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kadirnar | 36,204,372 | MDQ6VXNlcjM2MjA0Mzcy | User | false |
huggingface/diffusers | 2,044,842,453 | I_kwDOHa8MBc554dHV | 6,196 | https://github.com/huggingface/diffusers/issues/6196 | https://api.github.com/repos/huggingface/diffusers/issues/6196 | Error loading ControlNetXS Depth SDXL | ### Describe the bug
```
Traceback (most recent call last):
File "/mnt/d/OpenPrograms/daedalus/perf_tests/sdxl_controlnet-depth_test.py", line 18, in <module>
controlnet = ControlNetXSModel.from_pretrained(
File "/mnt/d/OpenPrograms/daedalus/wsl_venv/lib/python3.9/site-packages/huggingface_hub/utils/_valid... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-16T16:16:53Z | 2024-01-26T15:05:44Z | 2024-01-26T15:05:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aaronsantiago | 47,049,060 | MDQ6VXNlcjQ3MDQ5MDYw | User | false |
huggingface/diffusers | 2,045,379,468 | I_kwDOHa8MBc556gOM | 6,201 | https://github.com/huggingface/diffusers/issues/6201 | https://api.github.com/repos/huggingface/diffusers/issues/6201 | pipeline_stable_diffusion.py trys to encode the prompt when there is no prompt and `prompt_embeds` and `negative_prompt_embeds` are given | ### Describe the bug
`pipeline_stable_diffusion.py` trys to encode the prompt when there is no prompt and `prompt_embeds` and `negative_prompt_embeds` are given.
### Reproduction
Use `--pre_compute_text_embeddings` when train a dreambooth scripts. Use #6135 by @sayakpaul updating the `train_dreambooth_lora.py`. Eve... | open | null | false | 9 | [
"bug",
"stale"
] | [] | 2023-12-17T21:14:55Z | 2024-02-24T15:04:54Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CaffreyR | 84,232,793 | MDQ6VXNlcjg0MjMyNzkz | User | false |
huggingface/diffusers | 2,045,551,858 | I_kwDOHa8MBc557KTy | 6,203 | https://github.com/huggingface/diffusers/issues/6203 | https://api.github.com/repos/huggingface/diffusers/issues/6203 | Question about SDXL controlnet image size condition | The original SDXL is using original image size and cropped image size for condition. Why does controlnet SDXL directly use the resized image resolution as the condition information as below?
https://github.com/huggingface/diffusers/blob/9cef07da5a1f28ddfeadd232ede9a815c7945aec/examples/controlnet/train_controlnet_sd... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-12-18T02:01:47Z | 2024-01-26T15:05:36Z | 2024-01-26T15:05:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | GongXinyuu | 20,336,084 | MDQ6VXNlcjIwMzM2MDg0 | User | false |
huggingface/diffusers | 2,045,622,170 | I_kwDOHa8MBc557bea | 6,204 | https://github.com/huggingface/diffusers/issues/6204 | https://api.github.com/repos/huggingface/diffusers/issues/6204 | what's the difference between StableDiffusionInstructPix2PixPipeline and StableDiffusionImg2ImgPipeline | I don't know the difference between these two pipeline, since they are used to generate image by image, prompt. could you tell me the difference, thanks a lot!!
| closed | completed | false | 2 | [] | [] | 2023-12-18T02:52:41Z | 2023-12-18T07:00:12Z | 2023-12-18T03:59:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Emberss | 14,838,381 | MDQ6VXNlcjE0ODM4Mzgx | User | false |
huggingface/diffusers | 1,383,599,511 | I_kwDOHa8MBc5SeA2X | 621 | https://github.com/huggingface/diffusers/issues/621 | https://api.github.com/repos/huggingface/diffusers/issues/621 | Flax: trickle down `norm_num_groups` | `FlaxEncoder` receives it [here](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/vae_flax.py#L326), but then it's ignored.
`FlaxResnetBlock2D` uses a [hardcoded value](https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/vae_flax.py#L99).
Discovered by @younesbelkada in ... | closed | completed | false | 5 | [
"stale"
] | [] | 2022-09-23T10:27:55Z | 2022-11-04T12:14:11Z | 2022-11-04T12:14:11Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 2,046,007,047 | I_kwDOHa8MBc5585cH | 6,211 | https://github.com/huggingface/diffusers/issues/6211 | https://api.github.com/repos/huggingface/diffusers/issues/6211 | [Examples] How much time you support training scripts of text to video in diffusers? | I want to train svd in diffusers, can you support this feature in examples.
Thanks for your contributions. | closed | completed | false | 2 | [
"stale"
] | [] | 2023-12-18T08:26:57Z | 2024-01-26T15:05:32Z | 2024-01-26T15:05:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jiaxiangc | 76,088,379 | MDQ6VXNlcjc2MDg4Mzc5 | User | false |
huggingface/diffusers | 2,046,563,288 | I_kwDOHa8MBc55_BPY | 6,214 | https://github.com/huggingface/diffusers/issues/6214 | https://api.github.com/repos/huggingface/diffusers/issues/6214 | [Prompt Diffusion] In-Context Learning Unlocked for Diffusion Models | In short, prompt diffusion allows for in-context learning for diffusion models.
Code: https://github.com/Zhendong-Wang/Prompt-Diffusion
Models: https://huggingface.co/zhendongw/prompt-diffusion
One should be able to use https://github.com/huggingface/diffusers/blob/main/scripts/convert_original_controlnet_to_di... | closed | completed | false | 9 | [
"contributions-welcome"
] | [] | 2023-12-18T12:28:50Z | 2024-03-05T03:36:04Z | 2024-03-05T03:36:04Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,046,814,214 | I_kwDOHa8MBc55_-gG | 6,218 | https://github.com/huggingface/diffusers/issues/6218 | https://api.github.com/repos/huggingface/diffusers/issues/6218 | Batched seeds does not work for `StableDiffusionControlNetPipeline` (and Img2Img pipeline) | ### Describe the bug
When trying to create images with a consistent seed in a batch, the images turn out different.
### Reproduction
Load the model and pipeline
```py
import torch
from diffusers import StableDiffusionControlNetImg2ImgPipeline, ControlNetModel, UniPCMultistepScheduler
controlnet = ControlNetMod... | closed | completed | false | 12 | [
"bug",
"stale"
] | [] | 2023-12-18T14:39:31Z | 2024-01-18T16:22:31Z | 2024-01-18T16:22:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmitMY | 5,757,359 | MDQ6VXNlcjU3NTczNTk= | User | false |
huggingface/diffusers | 2,047,298,765 | I_kwDOHa8MBc56B0zN | 6,222 | https://github.com/huggingface/diffusers/issues/6222 | https://api.github.com/repos/huggingface/diffusers/issues/6222 | DDPM scheduler dependency on batch size | ### Describe the bug
I'm not sure if this is "officially" a bug, but it's unexpected enough to me that I wanted to flag it and check before submitting a PR.
When using the DDPM scheduler, the variance noise uses a single `randn_tensor` the same shape as `model_output` - this means that even with the same prompt an... | closed | completed | false | 18 | [
"bug",
"stale"
] | [] | 2023-12-18T19:19:32Z | 2024-02-18T09:32:04Z | 2024-02-18T09:32:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gtebbutt | 5,956,226 | MDQ6VXNlcjU5NTYyMjY= | User | false |
huggingface/diffusers | 2,047,457,695 | I_kwDOHa8MBc56Cbmf | 6,223 | https://github.com/huggingface/diffusers/issues/6223 | https://api.github.com/repos/huggingface/diffusers/issues/6223 | OSError: We couldn't connect to 'https://huggingface.co' | Hi all,
I have pushed the weights of my Dreambooth model to Huggingface (I used LoRa model). I confirm that pytorch_lora_weights.safetensors is indeed on Huggingface. When I want to get inference from the model, I run the following (as per the Readme file):
`from huggingface_hub.repocard import RepoCard
from dif... | closed | completed | false | 18 | [] | [] | 2023-12-18T21:08:17Z | 2025-12-12T20:49:50Z | 2023-12-26T17:27:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alirezaomneky | 152,901,224 | U_kgDOCR0WaA | User | false |
huggingface/diffusers | 2,047,664,531 | I_kwDOHa8MBc56DOGT | 6,224 | https://github.com/huggingface/diffusers/issues/6224 | https://api.github.com/repos/huggingface/diffusers/issues/6224 | TorchDynamo issue when fine-tuning text_to_image model | ### Describe the bug
I am trying to run the example, train_text_to_image.py, given by the authors, which is based on the Pokemon dataset. I just follow the instructions on huggingface and never revise the code. but I got this error:Python3 anaconda is now loaded in your environment.
For all commands you should now ... | closed | completed | false | 8 | [
"bug"
] | [] | 2023-12-18T23:50:52Z | 2023-12-29T00:49:58Z | 2023-12-29T00:47:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dahui-y | 93,740,259 | U_kgDOBZZc4w | User | false |
huggingface/diffusers | 2,047,871,693 | I_kwDOHa8MBc56EArN | 6,226 | https://github.com/huggingface/diffusers/issues/6226 | https://api.github.com/repos/huggingface/diffusers/issues/6226 | convert_diffusers_to_original_stable_diffusion.py causing loading errors in stable-diffusion-webui | ### Describe the bug
The latest [change](https://github.com/huggingface/diffusers/commit/6976cab7caa959bff46896c52ec03d9d414fc689) to scripts/convert_diffusers_to_original_stable_diffusion.py causes the following errors when loading the .safetensors file in stable-diffusion-webui:
```
RuntimeError: Error(s) in loa... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-12-19T03:53:11Z | 2024-02-12T15:05:33Z | 2024-02-12T15:05:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | georgesung | 16,344,756 | MDQ6VXNlcjE2MzQ0NzU2 | User | false |
huggingface/diffusers | 2,048,230,813 | I_kwDOHa8MBc56FYWd | 6,230 | https://github.com/huggingface/diffusers/issues/6230 | https://api.github.com/repos/huggingface/diffusers/issues/6230 | train_text_to_image_sdxl.py report cuda of out memory | I use the V100 to train, when I use the command`
accelerate launch train_text_to_image_sdxl.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--pretrained_vae_model_name_or_path=$VAE_NAME \
--dataset_name=$DATASET_NAME \
--enable_xformers_memory_efficient_attention \
--resolution=512 --center_crop --ra... | closed | completed | false | 2 | [] | [] | 2023-12-19T09:09:51Z | 2023-12-26T20:22:03Z | 2023-12-26T20:22:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chongxian | 18,145,013 | MDQ6VXNlcjE4MTQ1MDEz | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.