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 | 2,203,008,718 | I_kwDOHa8MBc6DTz7O | 7,440 | https://github.com/huggingface/diffusers/issues/7440 | https://api.github.com/repos/huggingface/diffusers/issues/7440 | Pokemon dataset no longer available for text-to-image | ### Describe the bug
The text to image Readme uses the pokemon dataset to finetune :
https://github.com/huggingface/diffusers/tree/main/examples/text_to_image#pokemon-example
However this dataset is no longer available:
https://huggingface.co/datasets/lambdalabs/pokemon-blip-captions
### Reproduction
from d... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-03-22T17:47:24Z | 2024-03-23T03:03:40Z | 2024-03-23T03:03:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ssarkar2 | 15,640,036 | MDQ6VXNlcjE1NjQwMDM2 | User | false |
huggingface/diffusers | 2,203,209,174 | I_kwDOHa8MBc6DUk3W | 7,441 | https://github.com/huggingface/diffusers/issues/7441 | https://api.github.com/repos/huggingface/diffusers/issues/7441 | pytest doesn't run on python3.12 | ### Describe the bug
https://peps.python.org/pep-0632/
### Reproduction
`./setup.py`
### Logs
_No response_
### System Info
diffusers 0.27.1, python 3.12, nixos 23.11
### Who can help?
@DN6 @sayakpaul @yiyixuxu | closed | completed | false | 13 | [
"bug"
] | [] | 2024-03-22T19:52:29Z | 2024-03-25T01:14:54Z | 2024-03-25T01:14:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pbsds | 140,964 | MDQ6VXNlcjE0MDk2NA== | User | false |
huggingface/diffusers | 2,203,797,165 | I_kwDOHa8MBc6DW0at | 7,443 | https://github.com/huggingface/diffusers/issues/7443 | https://api.github.com/repos/huggingface/diffusers/issues/7443 | torch.compile fails with StableDiffusionXL | ### Describe the bug
torch.compile is fialing recently with reduce-overhead mode using with StableDiffusionXL
### Reproduction
pip install torch diffusers transformers accelerate peft
from diffusers import StableDiffusionXLPipeline
import torch
pipe = StableDiffusionXLPipeline.from_pretrained(
"s... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-03-23T09:45:11Z | 2024-06-29T13:32:05Z | 2024-06-29T13:32:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | snehith-r | 35,597,622 | MDQ6VXNlcjM1NTk3NjIy | User | false |
huggingface/diffusers | 2,204,273,515 | I_kwDOHa8MBc6DYotr | 7,449 | https://github.com/huggingface/diffusers/issues/7449 | https://api.github.com/repos/huggingface/diffusers/issues/7449 | Clear finetuning example | I followed the tutorial for using a different scheduler and after loading all the other SD components, I was successfully able to use that pipeline for inference. However, I am having trouble finding a script that helps me to finetune the Unet module after having loaded the components like this. Particularly, I am stru... | closed | completed | true | 5 | [] | [] | 2024-03-24T09:59:09Z | 2024-03-25T10:58:17Z | 2024-03-25T10:58:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hamza13-12 | 109,283,952 | U_kgDOBoOKcA | User | false |
huggingface/diffusers | 1,398,876,107 | I_kwDOHa8MBc5TYSfL | 745 | https://github.com/huggingface/diffusers/issues/745 | https://api.github.com/repos/huggingface/diffusers/issues/745 | CLIP model support for pipelines | I am not sure if this has been considered, or already on the roadmap, but I'd love to be able to just throw a CLIP model ID at a pipe and have it download, and use said model.
I see HF has the capability to do this, and I have seen the clip guided diffusion example Colab, but it seems to be just implemented for tex... | closed | completed | false | 9 | [
"stale"
] | [] | 2022-10-06T06:10:33Z | 2022-11-21T15:03:09Z | 2022-11-21T15:03:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WASasquatch | 1,151,589 | MDQ6VXNlcjExNTE1ODk= | User | false |
huggingface/diffusers | 2,204,298,477 | I_kwDOHa8MBc6DYuzt | 7,450 | https://github.com/huggingface/diffusers/issues/7450 | https://api.github.com/repos/huggingface/diffusers/issues/7450 | AutoencoderKL multi thread issues | ### Describe the bug
Multi pipelines use same AutoencoderKL run concurrency lead to wired image like black.
### Reproduction
from diffusers import StableDiffusionPipeline
import torch
import threading
import copy
import numpy as np
import torch
from diffusers import (
... | closed | completed | false | 12 | [
"bug"
] | [] | 2024-03-24T11:11:08Z | 2024-03-28T04:07:06Z | 2024-03-28T04:07:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jasstionzyf | 5,241,459 | MDQ6VXNlcjUyNDE0NTk= | User | false |
huggingface/diffusers | 2,204,358,253 | I_kwDOHa8MBc6DY9Zt | 7,453 | https://github.com/huggingface/diffusers/issues/7453 | https://api.github.com/repos/huggingface/diffusers/issues/7453 | Controlnet processors do not respect local_files_only flag | ### Describe the bug
There is no way to pass the local_files_only flag to the controlnet Processor `load_processor` function (see `controlnet_aux/processor.py`)
**Code:**
```python
def load_processor(self, processor_id: str) -> 'Processor':
"""Load controlnet aux processors
Args:
... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-03-24T13:43:04Z | 2024-03-24T13:46:12Z | 2024-03-24T13:46:11Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | w4ffl35 | 25,737,761 | MDQ6VXNlcjI1NzM3NzYx | User | false |
huggingface/diffusers | 2,204,386,512 | I_kwDOHa8MBc6DZETQ | 7,454 | https://github.com/huggingface/diffusers/issues/7454 | https://api.github.com/repos/huggingface/diffusers/issues/7454 | LORA support | **Is your feature request related to a problem? Please describe.**
I'm trying to use LORA models from civitai and met the following issues:
- there are two ways to load weights: pipe.load_lora_weights and the conversion script in `/scripts`
- some LORAs bug out on loading (with both ways), some load fine
- for the ... | closed | completed | false | 1 | [] | [] | 2024-03-24T14:45:13Z | 2024-03-27T07:02:36Z | 2024-03-27T07:02:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mohamed--abdel-maksoud | 1,863,880 | MDQ6VXNlcjE4NjM4ODA= | User | false |
huggingface/diffusers | 2,204,477,013 | I_kwDOHa8MBc6DZaZV | 7,456 | https://github.com/huggingface/diffusers/issues/7456 | https://api.github.com/repos/huggingface/diffusers/issues/7456 | Missing raise statement | ### Describe the bug
https://github.com/huggingface/diffusers/blob/363699044e365ef977a7646b500402fa585e1b6b/src/diffusers/pipelines/stable_diffusion_gligen/pipeline_stable_diffusion_gligen.py#L471-L475
### Reproduction
There seems to be a missing raise statement in the gligen code (the `ValueError` is not raised). I... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-24T17:54:21Z | 2024-03-26T23:34:29Z | 2024-03-26T23:34:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TonyLianLong | 1,451,234 | MDQ6VXNlcjE0NTEyMzQ= | User | false |
huggingface/diffusers | 1,399,042,366 | I_kwDOHa8MBc5TY7E- | 746 | https://github.com/huggingface/diffusers/issues/746 | https://api.github.com/repos/huggingface/diffusers/issues/746 | StableDiffusionImg2ImgPipeline can't load by onnx revision | ### Describe the bug
when load StableDiffusionImg2ImgPipeline through from_pretrained by revision="onnx". an Exception raised:
```
File ~/venv/lib/python3.8/site-packages/diffusers/pipeline_utils.py:391, in DiffusionPipeline.from_pretrained(cls, pretrained_model_name_or_path, **kwargs)
388 init_kwargs[nam... | closed | completed | false | 2 | [
"bug"
] | [
"anton-l"
] | 2022-10-06T07:59:20Z | 2022-10-09T13:26:23Z | 2022-10-09T13:26:23Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LowinLi | 33,021,512 | MDQ6VXNlcjMzMDIxNTEy | User | false |
huggingface/diffusers | 2,204,795,811 | I_kwDOHa8MBc6DaoOj | 7,460 | https://github.com/huggingface/diffusers/issues/7460 | https://api.github.com/repos/huggingface/diffusers/issues/7460 | A bug in AdaLayerNorm | ### Describe the bug
[diffusers/blob/main/src/diffusers/models/normalization.py](https://github.com/huggingface/diffusers/blob/v0.26.3/src/diffusers/models/normalization.py#L46)
The default value of dim parameter of chunk function is 0. The shape of scale and the shape of x doesn't match.
### Reproduction
Th... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2024-03-25T03:14:48Z | 2024-07-25T11:23:12Z | 2024-06-29T13:32:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | akstt | 41,771,917 | MDQ6VXNlcjQxNzcxOTE3 | User | false |
huggingface/diffusers | 2,205,245,101 | I_kwDOHa8MBc6DcV6t | 7,462 | https://github.com/huggingface/diffusers/issues/7462 | https://api.github.com/repos/huggingface/diffusers/issues/7462 | [Pipeline] Champ: Controllable and Consistent Human Image Animation with 3D Parametric Guidance | ### Model/Pipeline/Scheduler description
Opening this issue for sharing the news but it should already be widely covered across social media or will be soon. The authors compare performance against the mythical [AnimateAnyone](https://github.com/HumanAIGC/AnimateAnyone) with 13k+ stars and 900 forks but no code, and... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-03-25T09:10:25Z | 2024-06-29T13:32:44Z | 2024-06-29T13:32:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | a-r-r-o-w | 72,266,394 | MDQ6VXNlcjcyMjY2Mzk0 | User | false |
huggingface/diffusers | 2,205,955,105 | I_kwDOHa8MBc6DfDQh | 7,466 | https://github.com/huggingface/diffusers/issues/7466 | https://api.github.com/repos/huggingface/diffusers/issues/7466 | torch.compile failed for multi-controlnet of sdxl inpaint | ### Describe the bug
torch.compile failed for multi-controlnet of sdxl inpaint
### Reproduction
controlnet_canny = ControlNetModel.from_pretrained(
"diffusers/controlnet-canny-sdxl-1.0", torch_dtype=torch.float16
)
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-03-25T14:55:13Z | 2024-03-26T12:48:56Z | 2024-03-26T12:48:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | putdanil | 75,547,456 | MDQ6VXNlcjc1NTQ3NDU2 | User | false |
huggingface/diffusers | 2,206,371,815 | I_kwDOHa8MBc6Dgo_n | 7,469 | https://github.com/huggingface/diffusers/issues/7469 | https://api.github.com/repos/huggingface/diffusers/issues/7469 | pretrained_model_or_path variable name in AutoPipeline | ### Describe the bug
In DiffusionPipeline.from_pretrained() the first parameter is called `pretrained_model_name_or_path` whereas in the Autopipelines, it is called `pretrained_model_or_path` (despite the class docs calling it `pretrained_model_name_or_path` here too).
Don't know if this was intended, just a bit o... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2024-03-25T18:05:23Z | 2024-04-26T17:09:38Z | 2024-04-26T17:09:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | PhilippeNguyen | 7,034,878 | MDQ6VXNlcjcwMzQ4Nzg= | User | false |
huggingface/diffusers | 2,208,831,664 | I_kwDOHa8MBc6DqBiw | 7,480 | https://github.com/huggingface/diffusers/issues/7480 | https://api.github.com/repos/huggingface/diffusers/issues/7480 | OOM appears using DeepFloyd_IF (I use 32G video memory) | ### Describe the bug
As in the title, OOM appears when I use DeepFloyd IF through diffuser (I use 32GB video memory).
### Reproduction
**I used the example code directly:**
from diffusers import Diffusio... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-26T17:11:19Z | 2024-03-27T07:02:10Z | 2024-03-27T07:02:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | qixuanwang-233 | 71,957,622 | MDQ6VXNlcjcxOTU3NjIy | User | false |
huggingface/diffusers | 2,209,613,367 | I_kwDOHa8MBc6DtAY3 | 7,483 | https://github.com/huggingface/diffusers/issues/7483 | https://api.github.com/repos/huggingface/diffusers/issues/7483 | LoRA to Diffusers conversion script not working with models using SDXL 1.0 as base model | ### Describe the bug
I am trying to convert .safetensors that are trained with LoRA using SDXL 1.0 as the base model to the Diffusers compatible format. I am running the conversion script convert_lora_safetensor_to_diffusers.py and the error I get is in the python logs below. The error is coming from line 55:
`curr_l... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-27T01:02:30Z | 2024-03-27T20:03:29Z | 2024-03-27T20:03:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | richiebailey74 | 57,724,969 | MDQ6VXNlcjU3NzI0OTY5 | User | false |
huggingface/diffusers | 2,209,794,745 | I_kwDOHa8MBc6Dtsq5 | 7,484 | https://github.com/huggingface/diffusers/issues/7484 | https://api.github.com/repos/huggingface/diffusers/issues/7484 | playground-v2.5-1024px-aesthetic does not support StableDiffusionXLInpaintPipeline | The generation effect is not good
vae = AutoencoderKL.from_pretrained("madebyollin/sdxl-vae-fp16-fix", torch_dtype=torch.float16)
pipe = StableDiffusionXLInpaintPipeline.from_pretrained(
"playgroundai/playground-v2.5-1024px-aesthetic",
# image_encoder=image_encoder,
# feature_extractor=feature_extr... | closed | completed | false | 1 | [] | [] | 2024-03-27T03:56:41Z | 2024-03-27T07:00:39Z | 2024-03-27T07:00:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yanchaoguo | 11,827,802 | MDQ6VXNlcjExODI3ODAy | User | false |
huggingface/diffusers | 2,209,889,554 | I_kwDOHa8MBc6DuD0S | 7,485 | https://github.com/huggingface/diffusers/issues/7485 | https://api.github.com/repos/huggingface/diffusers/issues/7485 | TypeError: 'NoneType' object is not callable when setting safety_checker to the pipeline | ### Describe the bug
I am using StableDiffusionControlNetInpaintPipeline for inference, the model I am using does not comes with a safety_checker, so I am setting the checker after creating pipeline from this pretrained model.
After the inferenc steps, I am getting this error:
TypeError: 'NoneType' object is not c... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-03-27T05:23:14Z | 2024-03-27T05:29:49Z | 2024-03-27T05:29:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nayan-dhabarde | 24,761,808 | MDQ6VXNlcjI0NzYxODA4 | User | false |
huggingface/diffusers | 2,210,254,811 | I_kwDOHa8MBc6Dvc_b | 7,488 | https://github.com/huggingface/diffusers/issues/7488 | https://api.github.com/repos/huggingface/diffusers/issues/7488 | [feat] Add option to use (Scheduled) Huber Loss in all diffusion training pipelines to improve resilience to data corruption and better image quality | **Is your feature request related to a problem? Please describe.**
Diffusion models are known to be vulnerable to outliers in training data. Therefore it's possible for a relatively small number of corrupting samples to "poison" the model, making it unable to produce desired output, which has been exploited by the p... | closed | completed | false | 8 | [
"stale"
] | [] | 2024-03-27T09:13:00Z | 2024-06-30T05:31:02Z | 2024-06-30T05:31:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,210,367,484 | I_kwDOHa8MBc6Dv4f8 | 7,490 | https://github.com/huggingface/diffusers/issues/7490 | https://api.github.com/repos/huggingface/diffusers/issues/7490 | Problem in backward | ### Describe the bug
Hi, I modified the U-Net forward function to use the "wrapped_forward" function. However, I encountered an issue when attempting to backpropagate the loss. I adapted the training code from "examples/train_dreambooth.py". The error log is provided below. It appears to be related to gradient checkpo... | closed | completed | true | 1 | [
"bug"
] | [] | 2024-03-27T10:05:32Z | 2024-03-27T13:22:31Z | 2024-03-27T13:22:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | haoweiz23 | 43,101,596 | MDQ6VXNlcjQzMTAxNTk2 | User | false |
huggingface/diffusers | 2,210,585,557 | I_kwDOHa8MBc6DwtvV | 7,492 | https://github.com/huggingface/diffusers/issues/7492 | https://api.github.com/repos/huggingface/diffusers/issues/7492 | Repository Not Found for url: https://huggingface.co/api/models/t2iadapter-model. | ### Describe the bug
when i am trying to run "train_t2i_adapter_sdxl.py", it's showing Repository Not Found for url : [](https://huggingface.co/api/models/t2iadapter-model).
I have already logged in on HF and use Token as well.
Here is the HF error:
![train_t2i_adapter_sdxl py - diffusers SSH_ 4090 - Visual St... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-27T11:43:51Z | 2024-03-28T03:06:16Z | 2024-03-28T03:06:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Shuvo001 | 61,898,803 | MDQ6VXNlcjYxODk4ODAz | User | false |
huggingface/diffusers | 2,211,384,579 | I_kwDOHa8MBc6Dzw0D | 7,497 | https://github.com/huggingface/diffusers/issues/7497 | https://api.github.com/repos/huggingface/diffusers/issues/7497 | Fine-Grained Subject-Specific Attribute Control | ### Model/Pipeline/Scheduler description
[This paper](https://github.com/CompVis/attribute-control) shows you can control attributes of subjects in a text-to-image model. E.g., for the prompt "A cat and dog", you can change the attributes of e.g. dog. For example, make it larger, smaller, darker, lighter etc while kee... | closed | completed | false | 10 | [
"community-examples",
"stale",
"contributions-welcome"
] | [] | 2024-03-27T17:33:50Z | 2024-11-30T02:36:56Z | 2024-11-30T02:36:55Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | UmerHA | 40,663,591 | MDQ6VXNlcjQwNjYzNTkx | User | false |
huggingface/diffusers | 2,211,565,892 | I_kwDOHa8MBc6D0dFE | 7,498 | https://github.com/huggingface/diffusers/issues/7498 | https://api.github.com/repos/huggingface/diffusers/issues/7498 | Issue running stable diffusion dreambooth on mac m3 max apple silicon. | ### Describe the bug
I am trying to run dreambooth stable diffusion on m3 max.
However I am running into an issue because of which whenever I am trying to generate the class images for the concepts, it fails.
### Reproduction
To reproduce the errors, try to setup dreambooth extension of m3 max apple silicon.
... | closed | completed | false | 35 | [
"bug",
"stale"
] | [] | 2024-03-27T18:28:39Z | 2024-10-21T15:14:38Z | 2024-10-21T15:14:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sagargulabani | 7,789,813 | MDQ6VXNlcjc3ODk4MTM= | User | false |
huggingface/diffusers | 2,213,249,156 | I_kwDOHa8MBc6D64CE | 7,506 | https://github.com/huggingface/diffusers/issues/7506 | https://api.github.com/repos/huggingface/diffusers/issues/7506 | NotImplementedError: Cannot copy out of meta tensor; no data! coz, Some weights of the model checkpoint were not used when initializing CLIPTextModelWithProjection | ### Describe the bug
I try to load a .safetensors file and save it as diffusers type model and I got **Some weights of the model checkpoint were not used when initializing CLIPTextModelWithProjection:
['text_model.embeddings.position_ids']** warning
```
import torch
from diffusers import StableDiffusionXLPipel... | closed | completed | false | 15 | [
"bug",
"stale",
"single_file"
] | [] | 2024-03-28T13:28:53Z | 2024-11-30T02:35:49Z | 2024-11-30T02:35:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shabri-arrahim | 54,322,914 | MDQ6VXNlcjU0MzIyOTE0 | User | false |
huggingface/diffusers | 2,213,749,946 | I_kwDOHa8MBc6D8yS6 | 7,509 | https://github.com/huggingface/diffusers/issues/7509 | https://api.github.com/repos/huggingface/diffusers/issues/7509 | [Pipeline] ReNoise: Real Image Inversion Through Iterative Noising | ### Model/Pipeline/Scheduler description
Achieving faithful image-to-noise inversion with Denoising Diffusion models remains a challenge, particularly for more recent models trained to generate images with a small number of denoising steps. This work introduces an inversion method with a high quality-to-operation rati... | open | null | false | 6 | [
"community-examples",
"stale",
"contributions-welcome"
] | [] | 2024-03-28T17:22:10Z | 2024-09-14T15:16:42Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clarencechen | 8,482,341 | MDQ6VXNlcjg0ODIzNDE= | User | false |
huggingface/diffusers | 2,214,468,572 | I_kwDOHa8MBc6D_hvc | 7,511 | https://github.com/huggingface/diffusers/issues/7511 | https://api.github.com/repos/huggingface/diffusers/issues/7511 | It is so confused to load textual-inversion sdxl embedding as inference prompt | ### Describe the bug
I trained a textual inversion model on SDXL pretrained model ( RealVisXL3.0 ), and then when I want to inference, I use this textual inversion model as a positive prompt to strengthen some aspect of the model.
But when I try to load the RealVisXL3.0 with `from_pretrained()` first and load the te... | open | null | false | 17 | [
"bug",
"stale"
] | [] | 2024-03-29T02:02:05Z | 2024-09-14T15:16:39Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xinli2008 | 71,311,087 | MDQ6VXNlcjcxMzExMDg3 | User | false |
huggingface/diffusers | 2,214,716,299 | I_kwDOHa8MBc6EAeOL | 7,513 | https://github.com/huggingface/diffusers/issues/7513 | https://api.github.com/repos/huggingface/diffusers/issues/7513 | Issue running SDXL Lora dreambooth on Nvidia 3060 12 GB RAM | ### Describe the bug
I am trying to run train_dreambooth_lora_sdxl on ubuntu WSL.
I have an Nvidia card with 3060.
I have about 19.5 GB of RAM allocated to ubuntu WSL.
I am running into the following error.
```
Query/Key/Value should either all have the same dtype, or (in the quantized case) Key/Value should... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-03-29T06:01:17Z | 2024-03-29T09:34:07Z | 2024-03-29T09:34:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sagargulabani | 7,789,813 | MDQ6VXNlcjc3ODk4MTM= | User | false |
huggingface/diffusers | 2,214,821,588 | I_kwDOHa8MBc6EA37U | 7,516 | https://github.com/huggingface/diffusers/issues/7516 | https://api.github.com/repos/huggingface/diffusers/issues/7516 | AttributeError: module diffusers has no attribute DepthNormalEstimationPipeline | ### Describe the bug
Hi, I'm trying to use the [GeoWizzard](https://huggingface.co/lemonaddie/Geowizard) model from the Huggingface Page, I just copied the codes from 'Use in Diffusers' as described on the webpage, but it raised the error `AttributeError: module diffusers has no attribute DepthNormalEstimationPipeline... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2024-03-29T07:22:01Z | 2024-04-29T10:21:03Z | 2024-04-29T10:18:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | csyhping | 38,368,516 | MDQ6VXNlcjM4MzY4NTE2 | User | false |
huggingface/diffusers | 2,215,157,157 | I_kwDOHa8MBc6ECJ2l | 7,519 | https://github.com/huggingface/diffusers/issues/7519 | https://api.github.com/repos/huggingface/diffusers/issues/7519 | cross_attention_kwargs ['scale'] are not expected by XFormersAttnProcessor and will be ignored. | ### Describe the bug
when i want to inference the trainde model, i use the "scale" params in cross_attention_kwargs, there has the info as "cross_attention_kwargs ['scale'] are not expected by XFormersAttnProcessor and will be ignored."。
I what to konw the "scale" is effective ?
Looking forward to your reply！
#... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-03-29T11:22:28Z | 2024-05-06T19:50:06Z | 2024-03-30T04:54:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wb315 | 39,976,683 | MDQ6VXNlcjM5OTc2Njgz | User | false |
huggingface/diffusers | 1,399,536,707 | I_kwDOHa8MBc5TazxD | 752 | https://github.com/huggingface/diffusers/issues/752 | https://api.github.com/repos/huggingface/diffusers/issues/752 | Support for multiple fine-tuned concepts & classes on one model with Dreambooth | **Is your feature request related to a problem? Please describe.**
I'd like to add multiple people and/or objects to a single model and finetune my model for that. For example, I'd like to add an entire family, together with their pets and their car, and this with as little degradation of my overall model as humanly p... | closed | completed | false | 12 | [
"stale",
"contributions-welcome"
] | [
"patil-suraj"
] | 2022-10-06T12:38:31Z | 2024-01-09T15:10:37Z | 2024-01-09T15:10:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jslegers | 883,647 | MDQ6VXNlcjg4MzY0Nw== | User | false |
huggingface/diffusers | 2,216,044,761 | I_kwDOHa8MBc6EFijZ | 7,522 | https://github.com/huggingface/diffusers/issues/7522 | https://api.github.com/repos/huggingface/diffusers/issues/7522 | move MarigoldPipeline to core | ### Model/Pipeline/Scheduler description
[Marigold depth pipeline](https://github.com/huggingface/diffusers/blob/main/examples/community/marigold_depth_estimation.py) is one of our most highly used community pipeline (it already has 40k downloads on the hub) - we are going to integrate it into core upon popular dema... | closed | completed | false | 5 | [] | [] | 2024-03-29T22:11:06Z | 2024-03-31T19:35:56Z | 2024-03-31T19:33:29Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 2,216,319,794 | I_kwDOHa8MBc6EGlsy | 7,525 | https://github.com/huggingface/diffusers/issues/7525 | https://api.github.com/repos/huggingface/diffusers/issues/7525 | Discrete sceduler | ### Model/Pipeline/Scheduler description
Hi, I was looking into the paper, [D3PM](https://openreview.net/forum?id=h7-XixPCAL) was wondering which scheduler to choose from particularly my problem deals with discrete action space and will be using Cross entropy to predictions and target
jax implementation of the sa... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-03-30T08:04:57Z | 2024-05-13T21:33:38Z | 2024-05-13T21:33:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Abc11c | 35,807,729 | MDQ6VXNlcjM1ODA3NzI5 | User | false |
huggingface/diffusers | 2,216,547,789 | I_kwDOHa8MBc6EHdXN | 7,529 | https://github.com/huggingface/diffusers/issues/7529 | https://api.github.com/repos/huggingface/diffusers/issues/7529 | ValueError: For the given accelerator, there seems to be an unexpected problem in type-casting. Please file an issue on the PyTorch GitHub repository. | ### Describe the bug
Running the script examples/text_to_image/train_text_to_image_lora_sdxl.py
Not working anymore on mixed_precision fp16, bf16
Traceback (most recent call last):0%|██████████████████████████████████████████████████████████████| 7/7 [00:00<00:00, 67.30it/s]
File "train_text_to_image_lora_sdxl.... | closed | completed | false | 13 | [
"bug"
] | [] | 2024-03-30T15:52:41Z | 2024-04-04T02:12:50Z | 2024-04-02T14:45:07Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,217,045,052 | I_kwDOHa8MBc6EJWw8 | 7,535 | https://github.com/huggingface/diffusers/issues/7535 | https://api.github.com/repos/huggingface/diffusers/issues/7535 | FloatTensor type hint is incompatible with Tensor | ### Describe the bug
At present, the `torch.FloatTensor` type hint is commonly employed in the codebase. However, if one were to pass a `torch.Tensor`, Pyright would raise complaints about the inconsistency between these two types. Furthermore, based on information from [this thread](https://discuss.pytorch.org/t/ty... | closed | completed | false | 3 | [
"bug",
"good first issue",
"contributions-welcome"
] | [] | 2024-03-31T15:21:05Z | 2024-05-10T19:53:32Z | 2024-05-10T19:53:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tianze0926 | 66,857,124 | MDQ6VXNlcjY2ODU3MTI0 | User | false |
huggingface/diffusers | 2,217,077,308 | I_kwDOHa8MBc6EJeo8 | 7,536 | https://github.com/huggingface/diffusers/issues/7536 | https://api.github.com/repos/huggingface/diffusers/issues/7536 | Train text encoder in train_text_to_image_sdxl.py | I have used this script successfully on a A100 40GB to train a full SDXL model, I would like to try training the text_encoder as well and that is currently not an option.
I'm wondering if that option can be added with possibly a separate learning rate argument for it. I know things are done to make things simple a... | closed | completed | false | 8 | [
"stale"
] | [] | 2024-03-31T16:42:31Z | 2024-06-29T13:30:22Z | 2024-06-29T13:30:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stpg06 | 96,386,710 | U_kgDOBb6-lg | User | false |
huggingface/diffusers | 2,217,464,598 | I_kwDOHa8MBc6EK9MW | 7,538 | https://github.com/huggingface/diffusers/issues/7538 | https://api.github.com/repos/huggingface/diffusers/issues/7538 | Issue of training on Custom Fill50k style dataset. | ### Describe the bug
I was trying to train T2I-Adapter segmentation on my Custom Fill50k style dataset but i'm getting **"AttributeError: 'str' object has no attribute 'convert'"** Error.
E... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-04-01T03:59:36Z | 2024-04-01T14:06:08Z | 2024-04-01T05:50:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Shuvo001 | 61,898,803 | MDQ6VXNlcjYxODk4ODAz | User | false |
huggingface/diffusers | 2,217,671,115 | I_kwDOHa8MBc6ELvnL | 7,539 | https://github.com/huggingface/diffusers/issues/7539 | https://api.github.com/repos/huggingface/diffusers/issues/7539 | bug in load lora weights when add align_device_hook to model | ### Describe the bug
i noticed that when i add ```align_device_hook``` to module in pipeline manually, then ```load_lora_weights``` function will enable the sequential cpu offload. so i dig deeper and find that ```load_lora_weights``` function use ```_optionally_disable_offloading``` function to decide whether to seq... | open | null | false | 19 | [
"bug",
"wip"
] | [] | 2024-04-01T07:00:56Z | 2025-02-27T06:39:22Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhangvia | 38,352,569 | MDQ6VXNlcjM4MzUyNTY5 | User | false |
huggingface/diffusers | 2,218,223,794 | I_kwDOHa8MBc6EN2iy | 7,545 | https://github.com/huggingface/diffusers/issues/7545 | https://api.github.com/repos/huggingface/diffusers/issues/7545 | RepaintPipeline: Preserve Original Tensor Model Output | I am trying to use the output of the RepaintPipeline in a torch.tensor, but the inherent scaling and conversion to PIL Image or numpy makes assumptions about the original normalization used for the input model.
Add extra case for non-rescaled tensor passed to ImagePipelineOutput in repaint_pipeline.py.
I have cas... | closed | completed | false | 5 | [
"stale"
] | [] | 2024-04-01T13:07:09Z | 2024-05-01T16:04:38Z | 2024-05-01T16:04:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | c-merchant | 13,489,410 | MDQ6VXNlcjEzNDg5NDEw | User | false |
huggingface/diffusers | 2,218,789,782 | I_kwDOHa8MBc6EQAuW | 7,548 | https://github.com/huggingface/diffusers/issues/7548 | https://api.github.com/repos/huggingface/diffusers/issues/7548 | Flicker in Animate Diff Pipe Outputs | ### Describe the bug
For some reason, using AnimateDiffPipeline leads to some flicker in the output that is not present with the same settings in comfyui, for instance. I notice a very apparent lighting flicker between the scenes. I've attached an example here that was produced from the code snippet.
![animation](htt... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-04-01T18:10:58Z | 2024-04-02T06:28:12Z | 2024-04-01T23:11:27Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AbhinavGopal | 43,016,805 | MDQ6VXNlcjQzMDE2ODA1 | User | false |
huggingface/diffusers | 2,218,832,403 | I_kwDOHa8MBc6EQLIT | 7,549 | https://github.com/huggingface/diffusers/issues/7549 | https://api.github.com/repos/huggingface/diffusers/issues/7549 | Potential UNet EMA bug in `train_text_to_image_sdxl` | ### Describe the bug
Hi,
Am I seeing a bug here? https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_sdxl.py#L1198-L1201
At the end of each epoch we perform a validation step which involves having `ema_unet` store `unet`'s params temporarily, and then loading `ema_unet`... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-04-01T18:35:41Z | 2024-04-09T12:37:56Z | 2024-04-09T12:37:56Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher-beckham | 2,417,792 | MDQ6VXNlcjI0MTc3OTI= | User | false |
huggingface/diffusers | 2,219,155,579 | I_kwDOHa8MBc6ERaB7 | 7,551 | https://github.com/huggingface/diffusers/issues/7551 | https://api.github.com/repos/huggingface/diffusers/issues/7551 | [Commuity Scripts]🤗 | ## Community Scripts
We added a "Community Scripts" README file to our community folder! 🤗
Please check it out [here](https://github.com/huggingface/diffusers/blob/main/examples/community/README_community_scripts.md)
The purpose of this section is to gather tips, tricks, and code examples from the community... | open | null | false | 2 | [
"wip"
] | [] | 2024-04-01T22:05:00Z | 2024-11-14T15:49:49Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 2,219,517,024 | I_kwDOHa8MBc6ESyRg | 7,554 | https://github.com/huggingface/diffusers/issues/7554 | https://api.github.com/repos/huggingface/diffusers/issues/7554 | When will you release your preprocessed dataset? | in `bidiff_if_objaverse.yaml`,
- root: './dataset/objaverse_dep_512'
- extra_view_root: './dataset/objaverse_dep_512_extra'
- prompt_file: './data/objaverse_32k/Cap3D_automated_Objaverse.csv'
- pre_prompt_embed_cache: './dataset/embed/embed_obj_cap3d'
- latent_root: './dataset/latents'
- sdf_root: './dataset/o... | closed | completed | false | 2 | [] | [] | 2024-04-02T04:06:45Z | 2024-04-02T07:21:05Z | 2024-04-02T07:21:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thucz | 135,981,918 | U_kgDOCBrrXg | User | false |
huggingface/diffusers | 2,219,660,338 | I_kwDOHa8MBc6ETVQy | 7,556 | https://github.com/huggingface/diffusers/issues/7556 | https://api.github.com/repos/huggingface/diffusers/issues/7556 | Update Animatediff docs with Scheduler improvements | Residual noise from DDIM and issues with GIF export are affecting the quality of AnimateDiff results. Update docs based on info from this issue:
https://github.com/huggingface/diffusers/issues/7548 | closed | completed | false | 3 | [] | [
"DN6"
] | 2024-04-02T06:25:44Z | 2024-11-19T09:13:34Z | 2024-11-19T09:13:33Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DN6 | 7,529,846 | MDQ6VXNlcjc1Mjk4NDY= | User | false |
huggingface/diffusers | 2,219,696,664 | I_kwDOHa8MBc6ETeIY | 7,557 | https://github.com/huggingface/diffusers/issues/7557 | https://api.github.com/repos/huggingface/diffusers/issues/7557 | none | null | closed | completed | false | 0 | [] | [] | 2024-04-02T06:48:24Z | 2024-04-02T06:57:59Z | 2024-04-02T06:51:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yaojinzhen | 63,828,175 | MDQ6VXNlcjYzODI4MTc1 | User | false |
huggingface/diffusers | 2,220,028,120 | I_kwDOHa8MBc6EUvDY | 7,559 | https://github.com/huggingface/diffusers/issues/7559 | https://api.github.com/repos/huggingface/diffusers/issues/7559 | TextToVideoSDPipeline outputs blank video | ### Describe the bug
I am encountering an issue when using the TextToVideoSDPipeline. It is generating blank videos when I run the model in Replicate. The model is running on A100 (80GB) hardware.
### Reproduction
```
pipe = TextToVideoSDPipeline.from_pretrained(
"cerspense/zeroscope_v2_576w",
... | closed | completed | false | 12 | [
"bug"
] | [] | 2024-04-02T09:36:34Z | 2024-10-15T21:25:31Z | 2024-10-15T21:25:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rafationgson | 70,418,702 | MDQ6VXNlcjcwNDE4NzAy | User | false |
huggingface/diffusers | 1,400,098,489 | I_kwDOHa8MBc5Tc865 | 756 | https://github.com/huggingface/diffusers/issues/756 | https://api.github.com/repos/huggingface/diffusers/issues/756 | ValueError: The deprecation tuple ('tensor_format' ... | ### Describe the bug
I am getting the following error
```
ValueError: The deprecation tuple ('tensor_format', '0.5.0', "If you're running your code in PyTorch, you can safely remove this argument.") should be removed since diffusers' version 0.5.0.dev0 is >= 0.5.0
```
When running diffusers
What does this... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2022-10-06T18:05:19Z | 2022-11-15T15:03:01Z | 2022-11-15T15:03:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WASasquatch | 1,151,589 | MDQ6VXNlcjExNTE1ODk= | User | false |
huggingface/diffusers | 2,220,227,951 | I_kwDOHa8MBc6EVf1v | 7,562 | https://github.com/huggingface/diffusers/issues/7562 | https://api.github.com/repos/huggingface/diffusers/issues/7562 | `enable_model_cpu_offload` fails with multicontrolnet on `StableDiffusionXLControlNetImg2ImgPipeline` | ### Describe the bug
If I provide a `pipe.enable_model_cpu_offload()`,
The `StableDiffusionXLControlNetImg2ImgPipeline` pipe fails with
```
RuntimeError: Input type (float) and bias type (c10::Half) should be the same
```
### Reproduction
Repro Colab: https://colab.research.google.com/drive/1kT0BTwVRIIQm... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-04-02T11:09:31Z | 2024-04-02T12:50:51Z | 2024-04-02T12:50:50Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,220,740,597 | I_kwDOHa8MBc6EXc_1 | 7,563 | https://github.com/huggingface/diffusers/issues/7563 | https://api.github.com/repos/huggingface/diffusers/issues/7563 | [mps] training / inference dtype issues | when training on Diffusers without attention slicing, we see:
```
/AppleInternal/Library/BuildRoots/ce725a5f-c761-11ee-a4ec-b6ef2fd8d87b/Library/Caches/com.apple.xbs/Sources/MetalPerformanceShaders/MPSCore/Types/MPSNDArray.mm:788: failed assertion `[MPSNDArray initWithDevice:descriptor:] Error: total bytes of NDArr... | open | null | false | 33 | [
"stale"
] | [] | 2024-04-02T14:54:41Z | 2024-09-14T15:16:26Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,223,040,962 | I_kwDOHa8MBc6EgOnC | 7,568 | https://github.com/huggingface/diffusers/issues/7568 | https://api.github.com/repos/huggingface/diffusers/issues/7568 | Use prompt tokens instead of prompt for sdxl? for the purpose of interpolation | Working from the following code, but would like to use the tokenized prompt so that I can interpolate between 2 prompts and produce a video output. The creation of the video from frames is easy enough, but I’m stuck here at the beginning. Any advice is appreciated. Perhaps this already exists but I can't find the answe... | closed | completed | false | 5 | [
"stale"
] | [] | 2024-04-03T14:05:24Z | 2024-05-03T21:43:29Z | 2024-05-03T21:43:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | prismspecs | 3,711,404 | MDQ6VXNlcjM3MTE0MDQ= | User | false |
huggingface/diffusers | 2,223,546,296 | I_kwDOHa8MBc6EiJ-4 | 7,569 | https://github.com/huggingface/diffusers/issues/7569 | https://api.github.com/repos/huggingface/diffusers/issues/7569 | `test_freeu_enabled()` fails on Mac | ### Describe the bug
The test `test_freeu_enabled` fails on Mac. The underlying issue is this method chain `utils.torch_utils.apply_freeu` -> `utils.torch_utils.fourier_filter` -> `torch.fft.fftn` ultimately tries to convert a `ComplexFloat` to MPS, which MPS doesn't seem to support.
I'll submit a PR to skip this t... | closed | completed | false | 0 | [
"bug"
] | [] | 2024-04-03T17:49:13Z | 2024-04-04T10:16:05Z | 2024-04-04T10:16:05Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | UmerHA | 40,663,591 | MDQ6VXNlcjQwNjYzNTkx | User | false |
huggingface/diffusers | 1,400,160,901 | I_kwDOHa8MBc5TdMKF | 757 | https://github.com/huggingface/diffusers/issues/757 | https://api.github.com/repos/huggingface/diffusers/issues/757 | Implement attention splicing for textual inversion | **Is your feature request related to a problem? Please describe.**
Textual inversion requires a lot of (V)RAM and could possibly benefit from attention splicing.
**Describe the solution you'd like**
Implement attention splicing for textual inversion
**Describe alternatives you've considered**
I tried to use `u... | closed | completed | false | 5 | [] | [] | 2022-10-06T18:56:41Z | 2022-10-10T11:02:41Z | 2022-10-07T09:21:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | allo- | 1,482,341 | MDQ6VXNlcjE0ODIzNDE= | User | false |
huggingface/diffusers | 2,223,975,041 | I_kwDOHa8MBc6EjyqB | 7,571 | https://github.com/huggingface/diffusers/issues/7571 | https://api.github.com/repos/huggingface/diffusers/issues/7571 | Stable Cascade Image to Image Pipeline | Are there any plans for Image2Image pipeline for the StableCascade model ?
| closed | completed | false | 6 | [
"stale"
] | [] | 2024-04-03T21:36:48Z | 2024-06-29T13:28:26Z | 2024-06-29T13:28:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ekurtulus | 66,876,436 | MDQ6VXNlcjY2ODc2NDM2 | User | false |
huggingface/diffusers | 2,224,146,151 | I_kwDOHa8MBc6Ekcbn | 7,572 | https://github.com/huggingface/diffusers/issues/7572 | https://api.github.com/repos/huggingface/diffusers/issues/7572 | RuntimeError: Input type (c10::Half) and bias type (float) should be the same | ### Describe the bug
I'm running my usual command in Google Colab, but there's something wrong today, i can't get it fix
I check all the pipeline it's already set float16, i don't know what's wrong when i'm running inference
### Reproduction
vae = AsymmetricAutoencoderKL.from_pretrained("cross-attention/asymmetric-... | closed | completed | false | 0 | [
"bug"
] | [] | 2024-04-04T00:08:42Z | 2024-04-04T06:46:28Z | 2024-04-04T06:46:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TheRide05 | 165,733,612 | U_kgDOCeDk7A | User | false |
huggingface/diffusers | 2,225,381,136 | I_kwDOHa8MBc6EpJ8Q | 7,574 | https://github.com/huggingface/diffusers/issues/7574 | https://api.github.com/repos/huggingface/diffusers/issues/7574 | KeyError: 'middle_block_out.0.weight' | ### Describe the bug
I'm encountering a `KeyError: 'middle_block_out.0.weight'` while attempting to utilize the **ControlNet** for an image colorization task. I have followed most of the guidelines outlined in the [tutorial](https://huggingface.co/docs/diffusers/training/controlnet?installation=PyTorch&gpu-select=1... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-04-04T12:36:27Z | 2024-04-04T13:13:40Z | 2024-04-04T13:13:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nick8592 | 56,616,275 | MDQ6VXNlcjU2NjE2Mjc1 | User | false |
huggingface/diffusers | 2,225,922,988 | I_kwDOHa8MBc6ErOOs | 7,575 | https://github.com/huggingface/diffusers/issues/7575 | https://api.github.com/repos/huggingface/diffusers/issues/7575 | Shape Error with Transformer2DModel and "adanorm" | ### Describe the bug
I have this shape error where `scale` and `shift` are different sizes from `x`. I'm not positive this is a bug so it could be me doing something wrong. Thanks in advance for taking a look at this.
### Reproduction
```
import torch
from diffusers import Transformer2DModel
model = Trans... | closed | not_planned | false | 3 | [
"bug",
"stale"
] | [] | 2024-04-04T16:03:01Z | 2024-12-22T00:08:01Z | 2024-12-22T00:08:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | will-rice | 25,072,137 | MDQ6VXNlcjI1MDcyMTM3 | User | false |
huggingface/diffusers | 1,400,172,106 | I_kwDOHa8MBc5TdO5K | 758 | https://github.com/huggingface/diffusers/issues/758 | https://api.github.com/repos/huggingface/diffusers/issues/758 | ImportError: cannot import name 'PROTOCOL_TLS' from 'urllib3.util.ssl_' | ### Describe the bug
When I import accelerate after installing it, and urllib3 (which was verified via `urllib3.__version__`) I get the following error *only one time*. When I run the cell again, the error is gone. Is there in order to these installations preferred?
```
[<ipython-input-2-14c74dad3ffb>](https://lo... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2022-10-06T19:07:16Z | 2023-02-11T04:44:12Z | 2022-11-15T15:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WASasquatch | 1,151,589 | MDQ6VXNlcjExNTE1ODk= | User | false |
huggingface/diffusers | 2,229,128,587 | I_kwDOHa8MBc6E3c2L | 7,589 | https://github.com/huggingface/diffusers/issues/7589 | https://api.github.com/repos/huggingface/diffusers/issues/7589 | GLIGEN for layout-to-image synthesis does not seem to work. | ### Describe the bug
I attempted to execute the python script for GLIGEN-based layout-to-image synthesis,
and it results in an error as follows:
```
AttributeError: 'UNet2DConditionModel' object has no attribute 'position_net'
```
### Reproduction
The script to reproduce the bug is as follows:
```
fr... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-04-06T07:48:24Z | 2024-04-08T05:34:42Z | 2024-04-08T05:34:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | joonjeon | 29,110,031 | MDQ6VXNlcjI5MTEwMDMx | User | false |
huggingface/diffusers | 1,400,195,762 | I_kwDOHa8MBc5TdUqy | 759 | https://github.com/huggingface/diffusers/issues/759 | https://api.github.com/repos/huggingface/diffusers/issues/759 | Make saving the full model optional for textual inversion | **Is your feature request related to a problem? Please describe.**
The resulting `.bin` file from the textual inversion script can be loaded with a few lines of code into the standard model. Saving a full model only to have an additional token is a waste of space (and possibly VRAM and CPU/GPU time).
**Describe the... | closed | completed | false | 7 | [
"stale"
] | [
"patil-suraj"
] | 2022-10-06T19:28:20Z | 2022-11-06T15:18:17Z | 2022-11-06T15:18:17Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | allo- | 1,482,341 | MDQ6VXNlcjE0ODIzNDE= | User | false |
huggingface/diffusers | 2,229,316,709 | I_kwDOHa8MBc6E4Kxl | 7,591 | https://github.com/huggingface/diffusers/issues/7591 | https://api.github.com/repos/huggingface/diffusers/issues/7591 | from_single_file use local_files_only=True got Error: "HeaderTooLarge" | @asomoza
Hi there, I follow you example to made mine:
```
from diffusers import StableDiffusionXLPipeline
from transformers import CLIPTextModel, CLIPTextModelWithProjection
import torch
text_encoder = CLIPTextModel.from_pretrained(
"model/stabilityai/stable-diffusion-xl-base-1.0/text_encoder",
var... | closed | completed | false | 2 | [] | [] | 2024-04-06T16:27:39Z | 2024-04-07T05:23:04Z | 2024-04-07T05:23:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SSLemon | 49,478,858 | MDQ6VXNlcjQ5NDc4ODU4 | User | false |
huggingface/diffusers | 2,229,464,460 | I_kwDOHa8MBc6E4u2M | 7,592 | https://github.com/huggingface/diffusers/issues/7592 | https://api.github.com/repos/huggingface/diffusers/issues/7592 | DoRA loading does not load all keys from the state_dict | ### Describe the bug
When loading a DoRA model from a kohya state_dict some keys are silently skipped in the state_dict.
DoRA loading was added in https://github.com/huggingface/diffusers/pull/7371.
This feature has not been released yet, so I am encountering this issue on `main` (commit: 6133d98ff70eafad7b9f65d... | closed | completed | false | 11 | [
"bug",
"stale"
] | [
"sayakpaul"
] | 2024-04-06T23:51:26Z | 2024-11-30T02:58:22Z | 2024-11-30T02:58:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | RyanJDick | 14,897,797 | MDQ6VXNlcjE0ODk3Nzk3 | User | false |
huggingface/diffusers | 2,229,535,019 | I_kwDOHa8MBc6E5AEr | 7,593 | https://github.com/huggingface/diffusers/issues/7593 | https://api.github.com/repos/huggingface/diffusers/issues/7593 | StableDiffusionPipeline prepare_ip_adapter_image_embeds, 'NoneType' object has no attribute 'image_projection_layers' | ### Describe the bug
I want to convert images into Image embeddings. When using StableDiffusionPipeline.prepare_ip_adapter_image_embeds, an error occurs:
AttributeError: 'NoneType' object has no attribute 'image_projection_layers'
### Reproduction
```python
pipeline = StableDiffusionPipeline.from_single_... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-04-07T03:06:05Z | 2024-04-07T17:31:56Z | 2024-04-07T15:10:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ShylockGao | 16,751,902 | MDQ6VXNlcjE2NzUxOTAy | User | false |
huggingface/diffusers | 2,229,817,553 | I_kwDOHa8MBc6E6FDR | 7,598 | https://github.com/huggingface/diffusers/issues/7598 | https://api.github.com/repos/huggingface/diffusers/issues/7598 | CombinedPipeline fails to accept bfloat16 image tensor as input | `StableCascadeCombinedPipeLine` accepts an `images` argument that can be a PIL image, a torch tensor, or a list of either. Unfortunately, I can't get it to accept a bfloat16 type for the image. It raises a runtime error in CLIP. I tried float32, but HF's A10G Large runs out of memory.
Here's the error I'm seeing wh... | closed | completed | false | 14 | [] | [] | 2024-04-07T15:23:31Z | 2024-05-09T16:32:56Z | 2024-05-09T16:32:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Michael-F-Ellis | 6,699,301 | MDQ6VXNlcjY2OTkzMDE= | User | false |
huggingface/diffusers | 1,400,524,373 | I_kwDOHa8MBc5Tek5V | 760 | https://github.com/huggingface/diffusers/issues/760 | https://api.github.com/repos/huggingface/diffusers/issues/760 | StableDiffusionPipeline producing unexpected output with MPS device using diffusers==0.4.0 | ### Describe the bug
I tried testing the potential speed updates of diffusers 0.4.0 on my M1 mac using an existing StableDiffusionPipeline-based script, and I found that a large image that would take ~3 min to generate in diffusers 0.3.0 was estimated to take more than 10x as long.
Since my existing script had a ... | closed | completed | false | 16 | [
"bug"
] | [] | 2022-10-07T02:05:54Z | 2022-10-18T00:22:30Z | 2022-10-11T18:30:11Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jncasey | 31,020,859 | MDQ6VXNlcjMxMDIwODU5 | User | false |
huggingface/diffusers | 2,230,010,515 | I_kwDOHa8MBc6E60KT | 7,600 | https://github.com/huggingface/diffusers/issues/7600 | https://api.github.com/repos/huggingface/diffusers/issues/7600 | DreamBooth: Not applying accelerator.accumulate on text_encoder | ### Describe the bug
At [examples/dreambooth/train_dreambooth_lora_sdxl.py#L1618](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_sdxl.py#L1618), only `unet` is applied with `accelerator.accumulate()`.
```python
for step, batch in enumerate(train_dataloader):
... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-04-07T23:38:27Z | 2024-06-29T13:27:06Z | 2024-06-29T13:27:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | immortalCO | 22,522,323 | MDQ6VXNlcjIyNTIyMzIz | User | false |
huggingface/diffusers | 2,230,174,952 | I_kwDOHa8MBc6E7cTo | 7,601 | https://github.com/huggingface/diffusers/issues/7601 | https://api.github.com/repos/huggingface/diffusers/issues/7601 | StableDiffusionXLControlNetPipeline bug in >0.26.3 with control_guidance_start type | ### Describe the bug
When I try to run InstantID pipeline, which is StableDiffusionXLInstantIDPipeline pipeline, on version newer than 0.26.3 I get very strange bug:
```
716 raise ValueError(
717 f"`control_guidance_start`: {control_guidance_start} has {len(control_guidance_start)} elemen... | closed | completed | false | 3 | [
"bug",
"needs-code-example"
] | [] | 2024-04-08T03:25:01Z | 2024-04-11T02:07:27Z | 2024-04-11T02:07:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | GrigoryEvko | 24,699,844 | MDQ6VXNlcjI0Njk5ODQ0 | User | false |
huggingface/diffusers | 2,230,330,355 | I_kwDOHa8MBc6E8CPz | 7,602 | https://github.com/huggingface/diffusers/issues/7602 | https://api.github.com/repos/huggingface/diffusers/issues/7602 | There may be a mistake in apply_forward_hook | ### Describe the bug
After I put different models of my own diffusion pipeline based on _DiffusionPipeline_ to different devices, in case the model's inputs and weights are not on the same device, I attach _AlignDevicesHooks_ on the models, such as VAE.
However, when I call _vae.encode_ or _vae.decode_, the error ... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-04-08T06:01:48Z | 2024-11-30T02:58:46Z | 2024-11-30T02:58:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | InuyashaLee | 114,284,586 | U_kgDOBs_YKg | User | false |
huggingface/diffusers | 2,230,612,470 | I_kwDOHa8MBc6E9HH2 | 7,606 | https://github.com/huggingface/diffusers/issues/7606 | https://api.github.com/repos/huggingface/diffusers/issues/7606 | Kandinsky 2.2: LoRA method - LoRAAttnAddedKVProcessor Decaprated | ### Describe the bug
I have install diffusers from source and I am trying to train kandisnky 2.2 decoder with LoRa method but I get this error.
Traceback (most recent call last):
File "/home/erfan/TORUS/sd_finetune/kan2/train_text_to_image_lora_decoder.py", line 812, in <module>
main()
File "/home/... | closed | not_planned | false | 4 | [
"bug",
"stale"
] | [] | 2024-04-08T08:43:15Z | 2024-11-30T02:59:14Z | 2024-11-30T02:59:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | erfansoheil | 145,917,256 | U_kgDOCLKFSA | User | false |
huggingface/diffusers | 2,230,825,689 | I_kwDOHa8MBc6E97LZ | 7,608 | https://github.com/huggingface/diffusers/issues/7608 | https://api.github.com/repos/huggingface/diffusers/issues/7608 | Want to use "strict=False" in from_pretrained | **What API design would you like to have changed or added to the library? Why?**
I want to add param "strict" in from_pretrained.
I want to add zerosft controlnet, but it can only implement by adding the set layers to unet class,
when I use from_pretrained, it can't use "strict=False"
**What use case would this e... | open | null | false | 9 | [
"stale"
] | [] | 2024-04-08T10:25:05Z | 2025-01-08T15:07:14Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CuddleSabe | 61,224,076 | MDQ6VXNlcjYxMjI0MDc2 | User | false |
huggingface/diffusers | 1,400,682,164 | I_kwDOHa8MBc5TfLa0 | 761 | https://github.com/huggingface/diffusers/issues/761 | https://api.github.com/repos/huggingface/diffusers/issues/761 | DreamBooth fails when training with 8bit adam optimizer. | ### Describe the bug
Seems that 8bit adam optimizer fails when training dreambooth.
The command I used:
python train_dreambooth.py \
--pretrained_model_name_or_path="CompVis/stable-diffusion-v1-4" \
--instance_data_dir=$instance_folder \
--class_data_dir=$class_folder \
--output_dir=$model_checkpoints_fol... | closed | completed | false | 7 | [
"bug"
] | [
"patil-suraj"
] | 2022-10-07T06:25:50Z | 2022-10-27T10:05:00Z | 2022-10-07T17:20:26Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nanlliu | 45,443,761 | MDQ6VXNlcjQ1NDQzNzYx | User | false |
huggingface/diffusers | 2,232,102,677 | I_kwDOHa8MBc6FCy8V | 7,611 | https://github.com/huggingface/diffusers/issues/7611 | https://api.github.com/repos/huggingface/diffusers/issues/7611 | Tensor size mismatch for unet2dconditional | ### Describe the bug
So I just started with the diffusers. I was following the tutorial: https://huggingface.co/docs/diffusers/using-diffusers/write_own_pipeline
An error occurred when I followed the code to this line:
with torch.no_grad():
noise_pred = unet(latent_model_input, t, encoder_hidden_states=text... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-04-08T21:21:56Z | 2024-04-09T14:53:51Z | 2024-04-09T14:53:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | metatl | 15,329,386 | MDQ6VXNlcjE1MzI5Mzg2 | User | false |
huggingface/diffusers | 2,232,436,063 | I_kwDOHa8MBc6FEEVf | 7,612 | https://github.com/huggingface/diffusers/issues/7612 | https://api.github.com/repos/huggingface/diffusers/issues/7612 | AttributeError: 'list' object has no attribute '__module__' when calling from_pretrained using file system | ### Discussed in https://github.com/huggingface/diffusers/discussions/7610
<div type='discussions-op-text'>
<sup>Originally posted by **dferguson992** April 8, 2024</sup>
I downloaded the stabilityai/stable-cascade model from HuggingFace and saved it to my file system using the following:
```python
model = St... | open | null | false | 7 | [
"stale"
] | [
"DN6"
] | 2024-04-09T01:26:44Z | 2024-09-20T20:06:48Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dferguson992 | 5,060,436 | MDQ6VXNlcjUwNjA0MzY= | User | false |
huggingface/diffusers | 2,233,281,257 | I_kwDOHa8MBc6FHSrp | 7,616 | https://github.com/huggingface/diffusers/issues/7616 | https://api.github.com/repos/huggingface/diffusers/issues/7616 | Img2Img with Playground 2.5 has artifacts when denoise is low | ### Describe the bug
I tried to generate an img2img with Playground2.5 while setting denoise is low, such as 0.5, the result image will have artifacts. If the denoise was high, such as 0.8, the result was fine, but of course it varied a lot.
I tried both EDMDP2M and EDMDP2MK, and I got the same results.
Here ar... | closed | completed | false | 9 | [
"bug"
] | [] | 2024-04-09T12:00:48Z | 2024-04-20T15:45:22Z | 2024-04-11T03:37:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MikeHanKK | 123,093,985 | U_kgDOB1ZD4Q | User | false |
huggingface/diffusers | 2,234,031,466 | I_kwDOHa8MBc6FKJ1q | 7,619 | https://github.com/huggingface/diffusers/issues/7619 | https://api.github.com/repos/huggingface/diffusers/issues/7619 | [Dreambooth] number of channels error in train_dreambooth.py | ### Describe the bug
I am trying to use train_dreambooth.py to train a personalized model by following https://github.com/huggingface/diffusers/tree/main/examples/dreambooth. I got the following error:
```
File "C:\WPy64-31090\python-3.10.9.amd64\lib\site-packages\torch\nn\modules\conv.py", line 458, in _conv_f... | closed | completed | false | 17 | [
"bug"
] | [] | 2024-04-09T17:54:06Z | 2024-05-14T02:49:55Z | 2024-05-14T02:49:55Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bssrdf | 689,043 | MDQ6VXNlcjY4OTA0Mw== | User | false |
huggingface/diffusers | 1,400,685,296 | I_kwDOHa8MBc5TfMLw | 762 | https://github.com/huggingface/diffusers/issues/762 | https://api.github.com/repos/huggingface/diffusers/issues/762 | CI testing on `mps` | We should add a couple of simple tests that use the `mps` device to detect issues like #760.
I'm not familiar with how the CI system is set up or how to trigger a test in a M1 mac, perhaps @anton-l can provide some ideas? | closed | completed | false | 2 | [] | [
"pcuenca",
"anton-l"
] | 2022-10-07T06:29:40Z | 2022-10-18T18:09:05Z | 2022-10-18T18:09:05Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 2,234,066,351 | I_kwDOHa8MBc6FKSWv | 7,621 | https://github.com/huggingface/diffusers/issues/7621 | https://api.github.com/repos/huggingface/diffusers/issues/7621 | `StableDiffusionXLInstructPix2PixPipeline` doesn't work with cosxl_edit | ### Describe the bug
CosXL Edit is an InstructPix2Pix model (https://huggingface.co/stabilityai/cosxl) released together with CosXL, however trying to load it gives a size mismatch error
### Reproduction
```py
import torch
from diffusers import StableDiffusionXLInstructPix2PixPipeline
pipe = StableDiffusionXL... | closed | completed | false | 14 | [
"bug",
"stale"
] | [] | 2024-04-09T18:17:20Z | 2024-05-13T03:27:07Z | 2024-05-13T03:27:07Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,234,332,869 | I_kwDOHa8MBc6FLTbF | 7,622 | https://github.com/huggingface/diffusers/issues/7622 | https://api.github.com/repos/huggingface/diffusers/issues/7622 | Inference speed | What is the fastest way to do inference with sdxl and sd1.5 these days without suffering on quality?
Seen a lot of 3rd party libs like openvivo , stable fast , stream diffusion , deepcache etc
But there is some quality loss or problems in inference.
Looking for inpainting specifically | closed | completed | false | 6 | [] | [] | 2024-04-09T21:32:17Z | 2024-10-15T21:27:08Z | 2024-10-15T21:27:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rishabh063 | 55,680,979 | MDQ6VXNlcjU1NjgwOTc5 | User | false |
huggingface/diffusers | 2,234,774,990 | I_kwDOHa8MBc6FM_XO | 7,626 | https://github.com/huggingface/diffusers/issues/7626 | https://api.github.com/repos/huggingface/diffusers/issues/7626 | Confusion about the use of from_config and its relation to _use_default_values | ### Describe the bug
I'm not sure if this qualifies as a bug because it depends on what the intent is behind the code. At the very least, I would classify it as peculiar behaviour. Suppose I instantiate some scheduler (doesn't matter which one) and extract (a copy of) its config to modify after:
```
sched = Eule... | open | null | false | 7 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2024-04-10T05:19:02Z | 2024-09-14T15:15:48Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher-beckham | 2,417,792 | MDQ6VXNlcjI0MTc3OTI= | User | false |
huggingface/diffusers | 2,235,272,030 | I_kwDOHa8MBc6FO4te | 7,629 | https://github.com/huggingface/diffusers/issues/7629 | https://api.github.com/repos/huggingface/diffusers/issues/7629 | svd can be run in pytorch 2.x version, but running in Pytorch 1.13.1 version will appear OOM | ### Describe the bug
I use diffusers 0.27.2 version to run svd, input the image size (1024, 576), When I use pytorch2.0.1 svd can run normally, and the VRAM usage is only about 12GB, which is far less than the maximum VRAM of my machine(80G), but when I use pytorch1.13.1 svd appears OOM. So I don't know if it's the py... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-04-10T10:35:16Z | 2024-04-10T12:07:09Z | 2024-04-10T12:07:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | buxianggaimingzi | 135,929,703 | U_kgDOCBofZw | User | false |
huggingface/diffusers | 1,400,729,590 | I_kwDOHa8MBc5TfW_2 | 763 | https://github.com/huggingface/diffusers/issues/763 | https://api.github.com/repos/huggingface/diffusers/issues/763 | Change from error to warning when moving an fp16 pipeline to CPU | https://github.com/huggingface/diffusers/commit/6c64741933c64df276b5ede21f62777dbe079cfd introduced an error moving an fp16 pipeline to CPU, which makes sense as it cannot run on CPU; however I would suggest changing it to a warning as it broke my use-case that I think is a legitimate one
This is the use-case that b... | closed | completed | false | 0 | [
"bug"
] | [] | 2022-10-07T07:17:58Z | 2022-10-07T08:34:47Z | 2022-10-07T08:34:47Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,236,833,063 | I_kwDOHa8MBc6FU10n | 7,636 | https://github.com/huggingface/diffusers/issues/7636 | https://api.github.com/repos/huggingface/diffusers/issues/7636 | how to use the controlnet sdxl tile model in diffusers | ### Describe the bug
I want to use [this model](https://huggingface.co/TTPlanet/TTPLanet_SDXL_Controlnet_Tile_Realistic_V1) to make my slightly blurry photos clear, so i found this model.
I follow the code [here](https://huggingface.co/lllyasviel/control_v11f1e_sd15_tile) , but as the model mentioned above is XL not... | closed | completed | false | 17 | [
"bug",
"stale"
] | [] | 2024-04-11T03:20:42Z | 2024-06-29T13:26:58Z | 2024-06-29T13:26:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xinli2008 | 71,311,087 | MDQ6VXNlcjcxMzExMDg3 | User | false |
huggingface/diffusers | 2,237,543,572 | I_kwDOHa8MBc6FXjSU | 7,640 | https://github.com/huggingface/diffusers/issues/7640 | https://api.github.com/repos/huggingface/diffusers/issues/7640 | StableDiffusionXL from_single_file get RuntimeError with dimensions smaller | ### Describe the bug
I'm using StableDiffusionXLPipeline From safetensor predict, got RuntimeError
very sure safetensor is OK, use [sd_scripts](https://github.com/kohya-ss/sd-scripts) with ckpt infer get Expected result
### Reproduction
```python
tokenizer = CLIPTokenizer.from_pretrained(
"/aigc-na... | closed | completed | false | 6 | [
"bug"
] | [] | 2024-04-11T11:49:04Z | 2024-05-20T14:11:18Z | 2024-04-12T03:55:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DeJoker | 18,501,966 | MDQ6VXNlcjE4NTAxOTY2 | User | false |
huggingface/diffusers | 2,237,588,049 | I_kwDOHa8MBc6FXuJR | 7,641 | https://github.com/huggingface/diffusers/issues/7641 | https://api.github.com/repos/huggingface/diffusers/issues/7641 | Intel MKL FATAL ERROR: Cannot load /home/sh0006yi/miniconda3/envs/pytorch/lib/python3.9/site-packages/torch/lib/libtorch_cpu.so. | ### Describe the bug
### Reproduction
export MODEL_NAM... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-04-11T12:14:40Z | 2024-06-29T13:26:43Z | 2024-06-29T13:26:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | crystal-kur | 56,252,135 | MDQ6VXNlcjU2MjUyMTM1 | User | false |
huggingface/diffusers | 2,238,211,159 | I_kwDOHa8MBc6FaGRX | 7,642 | https://github.com/huggingface/diffusers/issues/7642 | https://api.github.com/repos/huggingface/diffusers/issues/7642 | Add native support to InstantStyle by allowing users to choose target_blocks for IPAdapters | [InstantStyle](https://github.com/InstantStyle/InstantStyle) is proving to be really good quality and super simple approach - the authors realized certain blocks of the IPAdapter are responsible for composition and style
```py
# target_blocks=["block"] for original IP-Adapter
# target_blocks=["up_blocks.0.attentions... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-04-11T17:28:23Z | 2024-05-13T01:17:19Z | 2024-05-13T01:17:19Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,238,898,642 | I_kwDOHa8MBc6FcuHS | 7,644 | https://github.com/huggingface/diffusers/issues/7644 | https://api.github.com/repos/huggingface/diffusers/issues/7644 | non power of 2 sized images with stable cascade fail | ### Describe the bug
I am not sure if this is a bug, but the documentation does not mention a limitation to power of 2 that I can find.
Generating a non power of 2 sized image fails.
### Reproduction
```python
import torch
from diffusers import StableCascadeDecoderPipeline, StableCascadePriorPipeline
d... | open | null | false | 5 | [
"bug",
"stale"
] | [] | 2024-04-12T02:04:41Z | 2024-09-14T15:15:45Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Teriks | 14,919,098 | MDQ6VXNlcjE0OTE5MDk4 | User | false |
huggingface/diffusers | 2,238,939,261 | I_kwDOHa8MBc6Fc4B9 | 7,646 | https://github.com/huggingface/diffusers/issues/7646 | https://api.github.com/repos/huggingface/diffusers/issues/7646 | About dataloader_num_workers in train_text_to_image_lora.py | ### Describe the bug
I can run train_text_to_image_lora.py with dataloader_num_workers=0. But it does not work with dataloader_num_workers>0.
### Reproduction
I set dataloader_num_workers=4, here is the ouput.
The following values were not passed to `accelerate launch` and had defaults used instead:
`--n... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2024-04-12T02:47:07Z | 2024-11-30T02:56:57Z | 2024-11-30T02:56:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Hellcat1005 | 7,982,023 | MDQ6VXNlcjc5ODIwMjM= | User | false |
huggingface/diffusers | 1,400,884,072 | I_kwDOHa8MBc5Tf8to | 765 | https://github.com/huggingface/diffusers/issues/765 | https://api.github.com/repos/huggingface/diffusers/issues/765 | How to add Undesired Content with the txt2img pipe | With Undesired Content, you can list anything you want the AI to avoid in future generations.
But，how to add the uc like prompt with the piep? can you help me | closed | completed | false | 2 | [] | [] | 2022-10-07T09:26:18Z | 2022-10-07T12:07:24Z | 2022-10-07T12:07:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CrazyBoyM | 35,400,185 | MDQ6VXNlcjM1NDAwMTg1 | User | false |
huggingface/diffusers | 2,239,308,810 | I_kwDOHa8MBc6FeSQK | 7,651 | https://github.com/huggingface/diffusers/issues/7651 | https://api.github.com/repos/huggingface/diffusers/issues/7651 | Timesteps are not calculated correctly for img2img pipelines. | ### Describe the bug
I have arrived at this conclusion after experimenting with the LCM models/schedulers, and I believe it applies to all schedulers and img2img pipelines. For longer diffusion processes, the effects are more subtle but clearly surface in the case of the LCMScheduler.
For img2img, when we select ... | closed | completed | false | 29 | [
"bug",
"stale"
] | [] | 2024-04-12T07:39:01Z | 2024-10-15T15:12:34Z | 2024-06-29T13:26:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | okaris | 1,448,702 | MDQ6VXNlcjE0NDg3MDI= | User | false |
huggingface/diffusers | 2,240,224,008 | I_kwDOHa8MBc6FhxsI | 7,657 | https://github.com/huggingface/diffusers/issues/7657 | https://api.github.com/repos/huggingface/diffusers/issues/7657 | Add the ability to turn off classifer free guidance when sampling for a timestep range | Please refer to this paper.
## Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models
> Guidance is a crucial technique for extracting the best performance out of image-generating diffusion models. Traditionally, a constant guidance weight has been applied throughout ... | closed | completed | false | 15 | [
"stale"
] | [] | 2024-04-12T14:01:41Z | 2024-05-16T17:07:05Z | 2024-05-16T17:07:05Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 2,240,436,691 | I_kwDOHa8MBc6FilnT | 7,658 | https://github.com/huggingface/diffusers/issues/7658 | https://api.github.com/repos/huggingface/diffusers/issues/7658 | Is Multiple Conditioning in Stable Diffusion XL with ControlNet Possible? | **Is your feature request related to a problem? Please describe.**
I am encountering issues when trying to use multiple conditionings with the Stable Diffusion XL model using ControlNet. I followed a [guide](https://huggingface.co/blog/controlnet#canny-conditioning) and successfully ran ControlNet with depth and segme... | closed | completed | false | 4 | [
"stale"
] | [] | 2024-04-12T15:57:24Z | 2024-05-16T03:14:50Z | 2024-05-16T03:14:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SinaDavari | 126,830,506 | U_kgDOB49Hqg | User | false |
huggingface/diffusers | 2,241,069,644 | I_kwDOHa8MBc6FlAJM | 7,660 | https://github.com/huggingface/diffusers/issues/7660 | https://api.github.com/repos/huggingface/diffusers/issues/7660 | Trace example does not work | ### Describe the bug
The example of the trace for memory efficiency in the docs errors
### Reproduction
```
import time
import torch
from diffusers import StableDiffusionPipeline
import functools
torch.set_grad_enabled(False)
n_experiments = 2
unet_runs_per_experiment = 50
def generate_inputs():
... | closed | completed | false | 10 | [
"bug"
] | [] | 2024-04-12T22:33:24Z | 2024-04-16T04:30:52Z | 2024-04-16T04:28:34Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AbhinavGopal | 43,016,805 | MDQ6VXNlcjQzMDE2ODA1 | User | false |
huggingface/diffusers | 2,241,125,236 | I_kwDOHa8MBc6FlNt0 | 7,661 | https://github.com/huggingface/diffusers/issues/7661 | https://api.github.com/repos/huggingface/diffusers/issues/7661 | make it possible to create an animate diff pipeline with unet loaded with ip-adapter | [to-do] make it possible to create an animate diff pipeline with unet loaded with ip-adapter
currently this will fail
```python
from diffusers import DiffusionPipeline, AnimateDiffPipeline, MotionAdapter, DDIMScheduler
from diffusers.utils import export_to_gif
import torch
from diffusers.utils import load_im... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-04-12T23:48:56Z | 2024-05-13T18:15:02Z | 2024-05-13T18:15:02Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 2,241,991,604 | I_kwDOHa8MBc6FohO0 | 7,665 | https://github.com/huggingface/diffusers/issues/7665 | https://api.github.com/repos/huggingface/diffusers/issues/7665 | instruct_pix2pix problem | ### Describe the bug
I'm using the instruct_pix2pix training method to regenerate backgrounds for cut-out food images. However, I've noticed that the generated backgrounds often contain numerous fragmented and distorted cups, plates, and bowls. What could be the reason for this? I've examined my training data, and alt... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2024-04-14T05:57:22Z | 2024-05-16T03:11:36Z | 2024-05-16T03:11:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mechigonft | 90,537,707 | MDQ6VXNlcjkwNTM3NzA3 | User | false |
huggingface/diffusers | 2,242,018,069 | I_kwDOHa8MBc6FonsV | 7,666 | https://github.com/huggingface/diffusers/issues/7666 | https://api.github.com/repos/huggingface/diffusers/issues/7666 | `lpw_stable_diffusion_xl` custom pipeline doesn't work with `from_single_file` | ### Describe the bug
I tested `lpw_stable_diffusion_xl`, it works with `StableDiffusionXLPipeline.from_pretrained` but doesn't work with `StableDiffusionXLPipeline.from_single_file`. I tried to delete the truncated prompt to see if it's really truncating the prompt or just an ignorable log. Here are the results:
... | closed | completed | false | 12 | [
"bug"
] | [] | 2024-04-14T07:22:24Z | 2025-05-13T01:31:57Z | 2024-04-16T14:16:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nawka12 | 54,880,732 | MDQ6VXNlcjU0ODgwNzMy | User | false |
huggingface/diffusers | 2,242,128,053 | I_kwDOHa8MBc6FpCi1 | 7,669 | https://github.com/huggingface/diffusers/issues/7669 | https://api.github.com/repos/huggingface/diffusers/issues/7669 | Updated from 23.1 to 27.2 and generations are off. | ### Describe the bug
So I just been using the Abyss Orange Mix model from warrior in the diffusers file format. I get messages now about how the cliptextmodel isn't loading values anymore. Please keep in mind that it was loading them just fine before.
@sayakpaul Is this something new where things are intentionally ... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-04-14T12:22:50Z | 2024-04-19T11:00:16Z | 2024-04-19T11:00:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JemiloII | 5,192,788 | MDQ6VXNlcjUxOTI3ODg= | User | false |
huggingface/diffusers | 2,242,130,669 | I_kwDOHa8MBc6FpDLt | 7,670 | https://github.com/huggingface/diffusers/issues/7670 | https://api.github.com/repos/huggingface/diffusers/issues/7670 | How to use IDDPM in diffusers ? | The code base is here:
https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py | closed | completed | false | 3 | [
"should-move-to-discussion"
] | [] | 2024-04-14T12:30:34Z | 2024-11-20T00:17:18Z | 2024-11-20T00:17:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jiarenyf | 12,405,578 | MDQ6VXNlcjEyNDA1NTc4 | User | false |
huggingface/diffusers | 2,242,182,342 | I_kwDOHa8MBc6FpPzG | 7,672 | https://github.com/huggingface/diffusers/issues/7672 | https://api.github.com/repos/huggingface/diffusers/issues/7672 | error in using stable cascade with long prompt | Hi,
When I use stable cascade model with long prompt, I get below error.
`Token indices sequence length is longer than the specified maximum sequence length for this model (165 > 77). Running this sequence through the model will result in indexing errors`
I try to use `compel` library to fix this problem, but... | closed | completed | false | 6 | [
"stale"
] | [] | 2024-04-14T13:59:52Z | 2024-08-08T04:20:18Z | 2024-05-21T11:23:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | saeedkhanehgir | 65,589,645 | MDQ6VXNlcjY1NTg5NjQ1 | User | false |
huggingface/diffusers | 2,242,217,719 | I_kwDOHa8MBc6FpYb3 | 7,674 | https://github.com/huggingface/diffusers/issues/7674 | https://api.github.com/repos/huggingface/diffusers/issues/7674 | Deprecate VQDiffusionScheduler | **What API design would you like to have changed or added to the library? Why?**
Since the vq diffusion pipeline is deprecated in #6169 (sadly no docs can be found about why it is deprecated and what to use instead), there are no actively-supported modules or models in the code that depends on VQDiffusionScheduler any... | closed | completed | false | 2 | [] | [] | 2024-04-14T15:19:09Z | 2024-06-08T02:10:21Z | 2024-06-08T02:10:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | HydrogenC | 33,123,710 | MDQ6VXNlcjMzMTIzNzEw | User | false |
huggingface/diffusers | 2,242,236,901 | I_kwDOHa8MBc6FpdHl | 7,675 | https://github.com/huggingface/diffusers/issues/7675 | https://api.github.com/repos/huggingface/diffusers/issues/7675 | apple: support for MLX quantized linear in diffusers | **Is your feature request related to a problem? Please describe.**
As an Apple MPS user, it always feels somewhat like we're second-class citizens with respect to the latest and greatest optimisations that only happen for other platforms. The biggest deal is likely xformers/bitsandbytes which remain CUDA-only, but t... | open | reopened | false | 17 | [
"wip"
] | [] | 2024-04-14T16:04:39Z | 2024-12-15T04:00:39Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,242,438,611 | I_kwDOHa8MBc6FqOXT | 7,676 | https://github.com/huggingface/diffusers/issues/7676 | https://api.github.com/repos/huggingface/diffusers/issues/7676 | How to determine the type of file, such as checkpoint, etc. | Hello.
Is there some kind of script that determines the type of file "checkpoint", "LORA", "textual_inversion", etc.? | closed | completed | false | 2 | [] | [] | 2024-04-14T23:58:08Z | 2024-04-15T02:50:43Z | 2024-04-15T02:50:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | suzukimain | 131,413,573 | U_kgDOB9U2RQ | User | false |
huggingface/diffusers | 2,242,604,196 | I_kwDOHa8MBc6Fq2yk | 7,677 | https://github.com/huggingface/diffusers/issues/7677 | https://api.github.com/repos/huggingface/diffusers/issues/7677 | [Tests] help us speed up the fast pipeline tests | Currently, the pipeline tests (located under [`tests/pipelines`](https://github.com/huggingface/diffusers/tree/main/tests/pipelines/)) take a bit of time to run. We recently performed some optimizations to speed them up but, we can do better.
This is especially desirable because it greatly improves the contribution... | closed | completed | false | 7 | [
"Good second issue",
"contributions-welcome"
] | [] | 2024-04-15T03:07:17Z | 2024-11-20T03:23:18Z | 2024-11-20T03:23:18Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.