repo stringclasses 1
value | github_id int64 1.27B 4.63B | github_node_id stringlengths 18 24 | number int64 8 13.9k | html_url stringlengths 49 53 | api_url stringlengths 59 63 | title stringlengths 1 402 | body stringlengths 1 62.9k ⌀ | state stringclasses 2
values | state_reason stringclasses 4
values | locked bool 2
classes | comments_count int64 0 99 | labels listlengths 0 5 | assignees listlengths 0 5 | created_at stringdate 2022-06-09 16:28:35 2026-06-10 12:14:31 | updated_at stringdate 2022-06-12 22:18:01 2026-06-10 22:20:39 | closed_at stringdate 2022-06-12 22:18:01 2026-06-10 12:07:08 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 70
values | extracted_at stringdate 2026-04-07 13:34:13 2026-06-10 22:35:20 | author_login stringlengths 3 28 | author_id int64 1.54k 290M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 3,065,239,047 | I_kwDOHa8MBc62s9YH | 11,560 | https://github.com/huggingface/diffusers/issues/11560 | https://api.github.com/repos/huggingface/diffusers/issues/11560 | import error | ### Describe the bug
from diffusers import HunyuanVideoFramepackPipeline, HunyuanVideoFramepackTransformer3DModel
ImportError: cannot import name 'HunyuanVideoFramepackPipeline' from 'diffusers' (/opt/conda/lib/python3.11/site-packages/diffusers/__init__.py)
### Reproduction
import torch
from diffusers import Hu... | closed | completed | false | 1 | [
"bug"
] | [] | 2025-05-15T07:25:31Z | 2025-06-03T02:37:30Z | 2025-06-03T02:37:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vickeyhw | 20,540,333 | MDQ6VXNlcjIwNTQwMzMz | User | false |
huggingface/diffusers | 3,065,495,666 | I_kwDOHa8MBc62t8By | 11,561 | https://github.com/huggingface/diffusers/issues/11561 | https://api.github.com/repos/huggingface/diffusers/issues/11561 | FluxFillPipeline Support load IP Adapter. | ### Model/Pipeline/Scheduler description
'FluxFillPipeline' object has no attribute 'load_ip_adapter'
I really need this,Thanks!
### Open source status
- [ ] The model implementation is available.
- [ ] The model weights are available (Only relevant if addition is not a scheduler).
### Provide useful links for the ... | closed | completed | false | 6 | [
"help wanted",
"Good second issue"
] | [] | 2025-05-15T08:58:42Z | 2025-06-17T08:48:28Z | 2025-06-17T08:48:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | PineREN | 35,160,614 | MDQ6VXNlcjM1MTYwNjE0 | User | false |
huggingface/diffusers | 3,067,004,511 | I_kwDOHa8MBc62zsZf | 11,563 | https://github.com/huggingface/diffusers/issues/11563 | https://api.github.com/repos/huggingface/diffusers/issues/11563 | BlipDiffusionPipeline example breaks in v0.33.1, but works in v0.32.2 | ### Describe the bug
The code crashes in the newest version.
The minimum viable code is part of the official Huggingface BlipDiffusion example [found here](https://huggingface.co/docs/diffusers/en/api/pipelines/blip_diffusion#diffusers.BlipDiffusionPipeline.__call__.example), and the full example works on diffusers `... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-05-15T17:52:13Z | 2025-05-22T08:23:01Z | 2025-05-22T08:22:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Gorluxor | 18,031,411 | MDQ6VXNlcjE4MDMxNDEx | User | false |
huggingface/diffusers | 3,070,698,459 | I_kwDOHa8MBc63ByPb | 11,569 | https://github.com/huggingface/diffusers/issues/11569 | https://api.github.com/repos/huggingface/diffusers/issues/11569 | [Feature request] Please provide diffusers repo for ltxv-13b-0.9.7-distilled | **Is your feature request related to a problem? Please describe.**
The repo for 0.9.7-dev exists (https://huggingface.co/a-r-r-o-w/LTX-Video-0.9.7-diffusers) but recently released distilled version is also needed.
**Describe the solution you'd like.**
https://huggingface.co/a-r-r-o-w/LTX-Video-0.9.7-distilled-diffuse... | closed | completed | false | 2 | [] | [] | 2025-05-17T11:34:22Z | 2025-05-18T06:02:34Z | 2025-05-18T06:02:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,070,901,688 | I_kwDOHa8MBc63Cj24 | 11,570 | https://github.com/huggingface/diffusers/issues/11570 | https://api.github.com/repos/huggingface/diffusers/issues/11570 | .to() cast in .load_lora_weights breaks on bitsandbytes quantized text_encoder's for SD/SDXL | ### Describe the bug
This .to() cast on the text encoder: https://github.com/huggingface/diffusers/blob/9836f0e000cfd826a7a5099002253ed2becc13e0/src/diffusers/loaders/lora_base.py#L421
Is invalid when working with an SD1.5 / SDXL pipeline that has a bitsandbytes quantization config used on the text encoder
perhaps s... | closed | completed | false | 6 | [
"bug"
] | [] | 2025-05-17T15:59:59Z | 2025-05-21T07:02:27Z | 2025-05-21T07:02:27Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Teriks | 14,919,098 | MDQ6VXNlcjE0OTE5MDk4 | User | false |
huggingface/diffusers | 3,071,917,266 | I_kwDOHa8MBc63GbzS | 11,574 | https://github.com/huggingface/diffusers/issues/11574 | https://api.github.com/repos/huggingface/diffusers/issues/11574 | KeyError: 'vae.encoder.conv_out.conv.weight' | ### Describe the bug
Getting error with GGUF
```
(sddw-dev) C:\aiOWN\diffuser_webui>python LTXVideo-0.9.7-distilled.py
W0518 21:55:20.177000 10208 site-packages\torch\distributed\elastic\multiprocessing\redirects.py:29] NOTE: Redirects are currently not supported in Windows or MacOs.
Traceback (most recent call last)... | closed | completed | false | 0 | [
"bug"
] | [] | 2025-05-18T16:48:21Z | 2025-05-20T04:10:37Z | 2025-05-20T04:10:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,072,189,710 | I_kwDOHa8MBc63HeUO | 11,575 | https://github.com/huggingface/diffusers/issues/11575 | https://api.github.com/repos/huggingface/diffusers/issues/11575 | Hidream Model loading takes too long — any way to speed it up? | Hi, thanks for this great project.
I'm running Hidream with this library in a serverless environment and facing major delays during model loading. It can be very frustrating, especially for time-sensitive or ephemeral deployments.
I've tried everything I could think of to reduce the loading time, but nothing has work... | open | null | false | 8 | [
"stale"
] | [] | 2025-05-19T00:49:00Z | 2026-02-03T15:21:54Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Me-verner | 170,902,809 | U_kgDOCi_FGQ | User | false |
huggingface/diffusers | 1,437,495,795 | I_kwDOHa8MBc5VrnHz | 1,158 | https://github.com/huggingface/diffusers/issues/1158 | https://api.github.com/repos/huggingface/diffusers/issues/1158 | pytest tests/test_config.py tests interfere with each other | ### Describe the bug
If I run
`pytest ./tests/test_config.py::ConfigTester::test_load_ddim_from_pndm`
It passes but if I run,
`pytest ./tests/test_config.py`
it fails with,
```
========================================================================= FAILURES =========================================... | closed | completed | false | 4 | [
"bug"
] | [] | 2022-11-06T18:26:50Z | 2022-11-09T19:41:01Z | 2022-11-09T19:41:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Thomas-MMJ | 112,830,596 | U_kgDOBrmohA | User | false |
huggingface/diffusers | 3,073,643,475 | I_kwDOHa8MBc63NBPT | 11,580 | https://github.com/huggingface/diffusers/issues/11580 | https://api.github.com/repos/huggingface/diffusers/issues/11580 | Can diffusers support loading and running FLUX with fp8 ? | This is how I use diffusers to load flux model:
```
import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_pretrained(
"/ckptstorage/repo/pretrained_weights/black-forest-labs/FLUX.1-dev",
torch_dtype=torch.float16,
)
device = torch.device(f"cuda:{device_number}" if torch.cuda.is_available() e... | open | null | false | 6 | [
"stale"
] | [] | 2025-05-19T12:18:13Z | 2026-01-09T15:20:48Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | EmmaThompson123 | 138,575,543 | U_kgDOCEJ-tw | User | false |
huggingface/diffusers | 3,074,442,528 | I_kwDOHa8MBc63QEUg | 11,581 | https://github.com/huggingface/diffusers/issues/11581 | https://api.github.com/repos/huggingface/diffusers/issues/11581 | Unable to load IP Adapter into FLUX ( TypeError: CLIPVisionModelWithProjection.__init__() got an unexpected keyword argument 'dtype' ) | ### Describe the bug
IP Adapters fail to load with the Flux pipeline that has had a dtype passed to it.
### Reproduction
import torch
from diffusers import FluxPipeline
from diffusers.hooks import apply_group_offloading
from diffusers.utils import load_image
model_id = "./models/black-forest-labs/FLUX.1-dev"
dty... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2025-05-19T16:43:19Z | 2026-01-09T20:29:33Z | 2026-01-09T20:29:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Meatfucker | 74,834,323 | MDQ6VXNlcjc0ODM0MzIz | User | false |
huggingface/diffusers | 3,075,395,448 | I_kwDOHa8MBc63Ts94 | 11,584 | https://github.com/huggingface/diffusers/issues/11584 | https://api.github.com/repos/huggingface/diffusers/issues/11584 | torch.compile can't be used with groupoffloading on hunyuanvideo_frampack | ### Describe the bug
when i use block_level groupoffloading with use_stream=True on hunyuanvideo_framepack, i want to use torch compile to accelerate video generation. but it seems apply compile and groupoffloading at the same time won't accelerate the model forward.
### Reproduction
i'm using the code below on rtx4... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2025-05-20T01:53:25Z | 2026-01-09T20:34:01Z | 2026-01-09T20:34:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhangvia | 38,352,569 | MDQ6VXNlcjM4MzUyNTY5 | User | false |
huggingface/diffusers | 3,075,749,790 | I_kwDOHa8MBc63VDee | 11,585 | https://github.com/huggingface/diffusers/issues/11585 | https://api.github.com/repos/huggingface/diffusers/issues/11585 | Not getting the desired result as Stable Diffusion Web UI | ### Describe the bug
While using the Stable Diffusion Web UI and Hugging face Diffuser's Stable Diffusion for inpainting task with ControlNet, I am not getting the desired result in HF diffusers. But I want to get the same result as Stable Diffusion Web UI.
### Reproduction
controlnet = ControlNetModel.from_pretrain... | closed | completed | true | 6 | [
"bug",
"stale"
] | [] | 2025-05-20T05:55:55Z | 2026-01-09T20:34:27Z | 2026-01-09T20:34:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aniket-professional2025 | 208,433,529 | U_kgDODGxxeQ | User | false |
huggingface/diffusers | 1,437,496,602 | I_kwDOHa8MBc5VrnUa | 1,159 | https://github.com/huggingface/diffusers/issues/1159 | https://api.github.com/repos/huggingface/diffusers/issues/1159 | Stitching images & Generating larger images | **Is your feature request related to a problem? Please describe.**
For diversity diffusion model applications, include Conditional image generation & Image in-painting & Stitching images & Generating larger images, would you mind providing some examples for training or inference?
Thank you!
| closed | completed | false | 4 | [] | [] | 2022-11-06T18:29:46Z | 2022-11-10T00:41:53Z | 2022-11-10T00:41:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elricwan | 33,402,371 | MDQ6VXNlcjMzNDAyMzcx | User | false |
huggingface/diffusers | 3,077,065,631 | I_kwDOHa8MBc63aEuf | 11,590 | https://github.com/huggingface/diffusers/issues/11590 | https://api.github.com/repos/huggingface/diffusers/issues/11590 | Infinite (not literally) length video creation using LTX-Video? | First of all thanks to Aryan (0.9.7 integration) and DN6 (adding GGUF). Model is quite good and output is also promising.
I need help in creating continuous video using the last frame. 1 trick is to generate the video, extract the last frame and do inference. Is there any easy way where I can do this in loop.
My thou... | closed | completed | false | 1 | [] | [] | 2025-05-20T13:37:36Z | 2025-05-20T19:51:20Z | 2025-05-20T15:07:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,078,120,929 | I_kwDOHa8MBc63eGXh | 11,592 | https://github.com/huggingface/diffusers/issues/11592 | https://api.github.com/repos/huggingface/diffusers/issues/11592 | 0.33.X FluxLoraLoaderMixin `delete_adapters` Regression | ### Describe the bug
Loading a flux lora, using `delete_adapters` then loading another lora is basically a guaranteed key error because `delete_adapters` leaves the state dict inconsistent with what `load_lora_weights` expects.
Possibly related to https://github.com/huggingface/diffusers/issues/11003 but that was ope... | closed | completed | false | 2 | [
"bug"
] | [
"sayakpaul"
] | 2025-05-20T20:10:41Z | 2025-05-22T10:14:46Z | 2025-05-22T10:14:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Beinsezii | 39,478,211 | MDQ6VXNlcjM5NDc4MjEx | User | false |
huggingface/diffusers | 3,081,964,803 | I_kwDOHa8MBc63sw0D | 11,599 | https://github.com/huggingface/diffusers/issues/11599 | https://api.github.com/repos/huggingface/diffusers/issues/11599 | Scheduling LORA | In Comfy, it's possible to schedule loras and their weights. This is a really helpful feature, is it already possible or easy to implement in Diffusers?
https://blog.comfy.org/p/masking-and-scheduling-lora-and-model-weights | open | null | false | 3 | [
"stale",
"consider-for-modular-diffusers"
] | [] | 2025-05-22T04:00:21Z | 2026-02-03T15:21:49Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nom | 57,461 | MDQ6VXNlcjU3NDYx | User | false |
huggingface/diffusers | 1,437,499,778 | I_kwDOHa8MBc5VroGC | 1,160 | https://github.com/huggingface/diffusers/issues/1160 | https://api.github.com/repos/huggingface/diffusers/issues/1160 | the pytest test_local_custom_pipeline hangs/freezes the testing - have to ctrl-c | ### Describe the bug
running
`pytest -v -k "not test_local_custom_pipeline" tests/test_pipelines.py`
the hanging test is skipped, and all of the other tests either pass or are skipped due to slowness
`pytest -v tests/test_pipelines.py`
hangs on test_local_custom_pipeline, and I have to ctrl-c
### Rep... | closed | completed | false | 5 | [
"bug"
] | [
"anton-l"
] | 2022-11-06T18:41:15Z | 2022-11-08T02:44:53Z | 2022-11-08T02:44:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Thomas-MMJ | 112,830,596 | U_kgDOBrmohA | User | false |
huggingface/diffusers | 3,082,482,873 | I_kwDOHa8MBc63uvS5 | 11,601 | https://github.com/huggingface/diffusers/issues/11601 | https://api.github.com/repos/huggingface/diffusers/issues/11601 | `use_safetensors` is not passed to `load_sub_model` | ### Describe the bug
In `DiffusionPipeline.from_pretrained()` (`pipeline_utils.py`) the `use_safetensors` option is only passed to the downloader but not to `load_sub_model()` which means if the safetensors are in the cached folder then they will be loaded.
### Reproduction
```
from diffusers import StableDiffusio... | closed | completed | false | 3 | [
"bug"
] | [
"DN6"
] | 2025-05-22T08:16:26Z | 2025-05-27T08:16:30Z | 2025-05-27T08:16:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AnthonyBarbier | 5,241,909 | MDQ6VXNlcjUyNDE5MDk= | User | false |
huggingface/diffusers | 3,088,167,960 | I_kwDOHa8MBc64EbQY | 11,607 | https://github.com/huggingface/diffusers/issues/11607 | https://api.github.com/repos/huggingface/diffusers/issues/11607 | with a custom attention processor for Flux.dev, inference time changes when manually load and inject the transformer model into a flux pipeline versus let the flux pipeline constructor load the transformer internally. | With a custom attention processor for Flux.dev transformer, the inference time is different between the following two ways:
1. Manually load and inject the transformer into a flux.dev pipeline
2. Let the pipeline constructor load the transformer internally
The inference time of the first way is about 15% slower than... | closed | completed | false | 1 | [] | [] | 2025-05-24T06:42:11Z | 2025-05-26T01:27:00Z | 2025-05-26T01:27:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LinchuanXuTheSEAAI | 139,752,368 | U_kgDOCFRzsA | User | false |
huggingface/diffusers | 1,437,507,089 | I_kwDOHa8MBc5Vrp4R | 1,161 | https://github.com/huggingface/diffusers/issues/1161 | https://api.github.com/repos/huggingface/diffusers/issues/1161 | Convert Flax trained model to PyTorch | **Is your feature request related to a problem? Please describe.**
I'm using the new Flax pipeline for dreambooth training and want to be able to use the trained model on systems that don't run jax.
**Describe the solution you'd like**
Add from_flax parameter to PyTorch model.from_pretrained() functions like Flax ... | closed | completed | false | 11 | [
"stale"
] | [] | 2022-11-06T19:07:20Z | 2023-01-22T19:28:43Z | 2023-01-22T19:28:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jorahn | 13,120,204 | MDQ6VXNlcjEzMTIwMjA0 | User | false |
huggingface/diffusers | 3,093,316,779 | I_kwDOHa8MBc64YESr | 11,615 | https://github.com/huggingface/diffusers/issues/11615 | https://api.github.com/repos/huggingface/diffusers/issues/11615 | AnimateDiffPipeline: Tensor Mismatch with num_frames > motion_max_seq_length (chunking failure) | ### Describe the bug
When using AnimateDiffPipeline with a motion adapter (e.g., guoyww/animatediff-motion-adapter-v1-5-2 with motion_max_seq_length=32) and requesting num_frames in the pipeline call that is greater than the adapter's motion_max_seq_length (e.g., num_frames=64), a RuntimeError: The size of tensor a (6... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-05-27T09:43:47Z | 2025-06-18T17:44:18Z | 2025-06-18T17:44:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bpaviral | 76,731,338 | MDQ6VXNlcjc2NzMxMzM4 | User | false |
huggingface/diffusers | 3,099,292,412 | I_kwDOHa8MBc64u3L8 | 11,627 | https://github.com/huggingface/diffusers/issues/11627 | https://api.github.com/repos/huggingface/diffusers/issues/11627 | StableDiffusion3Pipeline with InstantX/SD3.5-Large-IP-Adapter | ### Describe the bug
Example comes from the Stable Diffusion 3 documentation:
https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_3#image-prompting-with-ip-adapters
Using InstantX/SD3.5-Large-IP-Adapter with StableDiffusion3Pipeline fails because the code expects an ip-adapt... | closed | completed | false | 1 | [
"bug"
] | [] | 2025-05-29T06:08:14Z | 2025-06-02T09:22:35Z | 2025-06-02T09:22:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alpha-magic | 16,140,190 | MDQ6VXNlcjE2MTQwMTkw | User | false |
huggingface/diffusers | 3,099,411,221 | I_kwDOHa8MBc64vUMV | 11,628 | https://github.com/huggingface/diffusers/issues/11628 | https://api.github.com/repos/huggingface/diffusers/issues/11628 | DreamO/Flux group offload: Expected all tensors to be on the same device, but found at least two devices | `removed for privacy` | closed | not_planned | false | 0 | [
"bug"
] | [] | 2025-05-29T07:06:52Z | 2025-06-03T08:22:00Z | 2025-06-02T15:23:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | realisticdreamer114514 | 96,880,303 | U_kgDOBcZGrw | User | false |
huggingface/diffusers | 3,100,931,357 | I_kwDOHa8MBc641HUd | 11,630 | https://github.com/huggingface/diffusers/issues/11630 | https://api.github.com/repos/huggingface/diffusers/issues/11630 | [FR] Please support WanVACETransformer3DModel.from_single_file | **Is your feature request related to a problem? Please describe.**
Need WanVACETransformer3DModel.from_single_file support to use GGUF format
**Describe the solution you'd like.**
Install diffusers
pip install git+https://github.com/huggingface/diffusers.git@refs/pull/11582/head
```python
from typing import List
imp... | closed | completed | false | 3 | [] | [
"a-r-r-o-w"
] | 2025-05-29T17:00:19Z | 2025-07-05T17:22:40Z | 2025-07-05T17:22:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,101,039,316 | I_kwDOHa8MBc641hrU | 11,631 | https://github.com/huggingface/diffusers/issues/11631 | https://api.github.com/repos/huggingface/diffusers/issues/11631 | OMI Format Compatibility | Opening up this thread to raise the conversation on compatibility with the Open Model Initiative standardized key format. We'd engaged w/ Marc Sun a while back on this topic, but wanted to centralize the discussion w/ respect to diffusers here.
@Nerogar has consolidated the OMI formats for models in this repo:
https:/... | open | null | false | 7 | [
"stale"
] | [] | 2025-05-29T17:46:22Z | 2026-02-03T15:21:41Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hipsterusername | 31,807,370 | MDQ6VXNlcjMxODA3Mzcw | User | false |
huggingface/diffusers | 3,101,534,173 | I_kwDOHa8MBc643afd | 11,633 | https://github.com/huggingface/diffusers/issues/11633 | https://api.github.com/repos/huggingface/diffusers/issues/11633 | When I use deepspeed_zero3 to train the controlnet on flux, an error occured "RuntimeError: tensor does not have a device". | [rank0]: File "/home/z/.local/lib/python3.10/site-packages/deepspeed/ops/adam/cpu_adam.py", line 163, in step
[rank0]: self.ds_opt_adam.adam_update(self.opt_id, state['step'], group['lr'], beta1, beta2, group['eps'],
[rank0]: RuntimeError: tensor does not have a device
Steps: 0%| | 0/500000 [00:04<?, ?... | open | null | false | 2 | [
"stale"
] | [] | 2025-05-29T21:30:40Z | 2026-01-09T15:20:14Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Daryu-Fan | 57,171,029 | MDQ6VXNlcjU3MTcxMDI5 | User | false |
huggingface/diffusers | 3,102,311,849 | I_kwDOHa8MBc646YWp | 11,637 | https://github.com/huggingface/diffusers/issues/11637 | https://api.github.com/repos/huggingface/diffusers/issues/11637 | How to load lora weight in distribution applications? | If I want to use xDiT with 2 GPU inference FluxControlPipeline, how should I do
I write a xFuserFluxControlPipeline class, but it can not load lora weight with right way
xFuserFluxTransformer in 1GPU have some parameters and another GPU have others.
How should I do ?? | closed | completed | false | 3 | [
"stale"
] | [] | 2025-05-30T07:14:50Z | 2026-01-09T20:48:00Z | 2026-01-09T20:48:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Johnson-yue | 10,268,274 | MDQ6VXNlcjEwMjY4Mjc0 | User | false |
huggingface/diffusers | 3,105,390,777 | I_kwDOHa8MBc65GIC5 | 11,641 | https://github.com/huggingface/diffusers/issues/11641 | https://api.github.com/repos/huggingface/diffusers/issues/11641 | [BUG] [CleanCode] Tuple[int] = (16, 56, 56) in FluxTransformer2DModel | ### Describe the bug
PyLance fixes error.
```
Выражение типа "tuple[Literal[16], Literal[56], Literal[56]]" не может быть назначено параметру типа "Tuple[int]"
"tuple[Literal[16], Literal[56], Literal[56]]" невозможно назначить "Tuple[int]"
Несоответствие размеров tuple: ожидается "1", но получено "3"Pylance
```... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-05-31T12:03:50Z | 2025-06-08T19:46:39Z | 2025-06-08T19:46:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SlimRG | 39,348,033 | MDQ6VXNlcjM5MzQ4MDMz | User | false |
huggingface/diffusers | 3,106,385,002 | I_kwDOHa8MBc65J6xq | 11,642 | https://github.com/huggingface/diffusers/issues/11642 | https://api.github.com/repos/huggingface/diffusers/issues/11642 | prompt_embeds_scale in FluxPriorReduxPipeline seems to have no effect. | ### Describe the bug
When using the FluxPriorReduxPipeline the prompt_embeds_scale and pooled_prompt_embeds_scale seem to have no effect on the final generation.
### Reproduction
```
async def get_redux_embeds(image, prompt, strength):
redux_repo = "black-forest-labs/FLUX.1-Redux-dev"
text_encoder, tokenizer,... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2025-06-01T01:47:57Z | 2026-01-09T20:57:30Z | 2026-01-09T20:57:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Meatfucker | 74,834,323 | MDQ6VXNlcjc0ODM0MzIz | User | false |
huggingface/diffusers | 3,109,140,778 | I_kwDOHa8MBc65Ubkq | 11,644 | https://github.com/huggingface/diffusers/issues/11644 | https://api.github.com/repos/huggingface/diffusers/issues/11644 | Error in init from pretrained for LTXConditionPipeline | ### Describe the bug
pipe = LTXConditionPipeline.from_pretrained("Lightricks/LTX-Video-0.9.7-dev", torch_dtype=torch.bfloat16)
File "/opt/nas/p/conda/envs/diffuser0526/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn
return fn(*args, **kwargs)
File "/opt/nas/p/cond... | closed | completed | false | 8 | [
"bug"
] | [] | 2025-06-02T08:24:37Z | 2025-07-09T05:36:46Z | 2025-07-09T05:36:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vickeyhw | 20,540,333 | MDQ6VXNlcjIwNTQwMzMz | User | false |
huggingface/diffusers | 3,113,397,731 | I_kwDOHa8MBc65kq3j | 11,648 | https://github.com/huggingface/diffusers/issues/11648 | https://api.github.com/repos/huggingface/diffusers/issues/11648 | how to load lora weight with fp8 transfomer model? | Hi, I want to run fluxcontrolpipeline with transformer_fp8 reference the code :
https://huggingface.co/docs/diffusers/api/pipelines/flux#quantization
```
import torch
from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig, FluxTransformer2DModel, FluxControlPipeline
from transformers import BitsAndB... | open | reopened | false | 23 | [
"stale"
] | [] | 2025-06-03T10:31:23Z | 2026-01-09T15:19:55Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Johnson-yue | 10,268,274 | MDQ6VXNlcjEwMjY4Mjc0 | User | false |
huggingface/diffusers | 1,437,553,050 | I_kwDOHa8MBc5Vr1Ga | 1,165 | https://github.com/huggingface/diffusers/issues/1165 | https://api.github.com/repos/huggingface/diffusers/issues/1165 | prompt endoftext bug - results with prompt > max-length are way better then prompt with length <= max-length | ### Describe the bug
Hi I trained a personal model (key: `smnb`, class: `person`) with dreambooth, and i found, that I can't replicate the results with automatic1111 on the one side and huggingface/transformers + notebook on the other side.
the fascinating thing is, that if I use exactly max_words + 1, then my prom... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2022-11-06T21:53:29Z | 2022-12-15T15:04:42Z | 2022-12-15T15:04:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | petekay | 17,798,967 | MDQ6VXNlcjE3Nzk4OTY3 | User | false |
huggingface/diffusers | 3,114,321,132 | I_kwDOHa8MBc65oMTs | 11,650 | https://github.com/huggingface/diffusers/issues/11650 | https://api.github.com/repos/huggingface/diffusers/issues/11650 | max_shard_size | ### Describe the bug
max_shard_size=1GB is set but the size of each safetensor is still greater than 1GB
### Reproduction
pipe.save_pretrained(save_directory='FLUX.1-dev-Control-all', max_shard_size="1GB")
### Logs
```shell
```
### System Info
diffusers=0.30.2
accelerate=1.7.0
transformers=4.52.4
### Who can h... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2025-06-03T14:51:59Z | 2026-01-09T20:48:49Z | 2026-01-09T20:48:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Ranking666 | 74,530,242 | MDQ6VXNlcjc0NTMwMjQy | User | false |
huggingface/diffusers | 3,114,343,229 | I_kwDOHa8MBc65oRs9 | 11,651 | https://github.com/huggingface/diffusers/issues/11651 | https://api.github.com/repos/huggingface/diffusers/issues/11651 | Can't load flux-fill-lora with FluxControl | ### Describe the bug
I am trying to load lora model trained with flux-fill pipeline using `FluxControlInpaintPipeline`. But it is not able to load the lora model into transformers. Any advice is appreciated. I want to have flux fill pipeline with control.
### Reproduction
Download sample flux fill lora model
```
w... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2025-06-03T14:57:24Z | 2026-01-09T20:59:56Z | 2026-01-09T20:59:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hardikdava | 39,372,750 | MDQ6VXNlcjM5MzcyNzUw | User | false |
huggingface/diffusers | 3,115,549,266 | I_kwDOHa8MBc65s4JS | 11,653 | https://github.com/huggingface/diffusers/issues/11653 | https://api.github.com/repos/huggingface/diffusers/issues/11653 | support .alpha keys in HiDream loras trained using OneTrainer | ### Describe the bug
If a Hidream lora is trained using OneTrainer, some keys of the checkpoint have the ".alpha" in them instead of all of them including 'lora'. So this condition fails: `is_correct_format = all("lora" in key for key in state_dict.keys())` in the `load_lora_weights` method of `HiDreamImageLoraLoaderM... | open | null | false | 1 | [
"bug",
"stale"
] | [
"sayakpaul"
] | 2025-06-03T21:59:24Z | 2026-01-09T15:19:43Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ali-afridi26 | 188,188,833 | U_kgDOCzeIoQ | User | false |
huggingface/diffusers | 3,118,944,329 | I_kwDOHa8MBc6551BJ | 11,657 | https://github.com/huggingface/diffusers/issues/11657 | https://api.github.com/repos/huggingface/diffusers/issues/11657 | Custom Wan diffusion Lora runs without error but doesn't apply effect and gives warning: No LoRA keys associated to WanTransformer3DModel found with the prefix='transformer'. | ### Describe the bug
I run the diffusers pipe using the standard process with a custom diffusers trained lora:
pipe = WanPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
pipe.scheduler = scheduler
pipe.load_lora_weights("lora/customdiffusers_lora.safetensors")
etc...
it runs without error but... | closed | completed | false | 3 | [
"bug"
] | [] | 2025-06-04T19:50:14Z | 2025-09-12T03:32:17Z | 2025-06-07T07:49:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | st-projects-00 | 51,993,981 | MDQ6VXNlcjUxOTkzOTgx | User | false |
huggingface/diffusers | 3,119,259,740 | I_kwDOHa8MBc657CBc | 11,659 | https://github.com/huggingface/diffusers/issues/11659 | https://api.github.com/repos/huggingface/diffusers/issues/11659 | LoRA load issue | ### Describe the bug
FluxPipeline fails to load LoRA weights (https://civitai.com/models/876464):
### Reproduction
```
from diffusers import FluxPipeline
from torch import bfloat16
prompt_from_repo = "Beautiful female woodelf in shiny iron and silver armor, full body view, full body shot, feet and arms seen in th... | open | null | false | 6 | [
"bug",
"stale"
] | [] | 2025-06-04T22:11:32Z | 2026-01-09T15:19:38Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | D1-3105 | 65,292,437 | MDQ6VXNlcjY1MjkyNDM3 | User | false |
huggingface/diffusers | 1,437,556,334 | I_kwDOHa8MBc5Vr15u | 1,166 | https://github.com/huggingface/diffusers/issues/1166 | https://api.github.com/repos/huggingface/diffusers/issues/1166 | [TPU] Textual Inversion Training hangs when saving model | ### Describe the bug
I am running the `textual_inversion.py` on a v3-8 TPU VM, and the script hangs at the model saving (save_progress) step.
Any clue why that may be happening?
### Reproduction
_No response_
### Logs
_No response_
### System Info
TPU VM - tpu-vm-pt-1.10
- `diffusers` version: 0.7.1
- Platfo... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2022-11-06T22:03:42Z | 2022-12-15T15:04:41Z | 2022-12-15T15:04:41Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dhruvrnaik | 22,565,320 | MDQ6VXNlcjIyNTY1MzIw | User | false |
huggingface/diffusers | 3,120,122,551 | I_kwDOHa8MBc65-Uq3 | 11,661 | https://github.com/huggingface/diffusers/issues/11661 | https://api.github.com/repos/huggingface/diffusers/issues/11661 | [BUG]: Using args.max_train_steps even if it is None in diffusers/examples/flux-control | ### Describe the bug
Under [https://github.com/huggingface/diffusers/tree/main/examples/flux-control](examples/flux-control) there are two files showing how to fine tune flux-control:
- [train_control_flux.py](https://github.com/huggingface/diffusers/blob/main/examples/flux-control/train_control_flux.py)
- [train_cont... | closed | completed | false | 0 | [
"bug"
] | [] | 2025-06-05T07:18:06Z | 2025-06-05T09:26:26Z | 2025-06-05T09:26:26Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Markus-Pobitzer | 17,593,719 | MDQ6VXNlcjE3NTkzNzE5 | User | false |
huggingface/diffusers | 3,124,234,125 | I_kwDOHa8MBc66OAeN | 11,674 | https://github.com/huggingface/diffusers/issues/11674 | https://api.github.com/repos/huggingface/diffusers/issues/11674 | [FR] Please support ref image and multiple control videos in Wan VACE | **Is your feature request related to a problem? Please describe.**
Not a problem but a good feature to be supported in Wan Vace pipeline as model supports it.
**Describe the solution you'd like.**
```python
from controlnet_aux import OpenposeDetector, MidasDetector
from diffusers.utils import load_video, export_to_vi... | closed | completed | false | 0 | [
"enhancement",
"contributions-welcome"
] | [] | 2025-06-06T09:40:08Z | 2025-10-09T15:56:35Z | 2025-10-09T15:56:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,125,370,348 | I_kwDOHa8MBc66SV3s | 11,675 | https://github.com/huggingface/diffusers/issues/11675 | https://api.github.com/repos/huggingface/diffusers/issues/11675 | Error in loading the pretrained lora weights | Hi, I am using the script https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora_sdxl.py to train a lora.
An error is raised on https://github.com/huggingface/diffusers/blob/73a9d5856f2d7ae3637c484d83cd697284ad3962/examples/text_to_image/train_text_to_image_lora_sdxl.py#L131... | closed | completed | false | 1 | [] | [] | 2025-06-06T17:09:45Z | 2025-06-07T07:40:14Z | 2025-06-07T07:39:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | garychan22 | 108,175,311 | U_kgDOBnKfzw | User | false |
huggingface/diffusers | 3,126,190,598 | I_kwDOHa8MBc66VeIG | 11,676 | https://github.com/huggingface/diffusers/issues/11676 | https://api.github.com/repos/huggingface/diffusers/issues/11676 | HunyuanVideoImageToVideoPipeline memory leak | ### Describe the bug
`RuntimeError: Invalid buffer size: 635.13 GB` on Apple Silicon with mps, when the readme claims 60GB. Smaller res possible but still with 10x the memory claimed. What's going on here? Is it leaking memory?
Reproducible from boilerplate example code:
```import torch
from diffusers import Hunyuan... | open | null | false | 9 | [
"bug",
"stale"
] | [] | 2025-06-07T00:45:56Z | 2026-02-03T15:21:27Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | openSourcerer9000 | 61,931,826 | MDQ6VXNlcjYxOTMxODI2 | User | false |
huggingface/diffusers | 1,437,717,166 | I_kwDOHa8MBc5VsdKu | 1,168 | https://github.com/huggingface/diffusers/issues/1168 | https://api.github.com/repos/huggingface/diffusers/issues/1168 | What is "class images" mean for dreambooth training? | What is "class images" mean for dreambooth training?
If instance images meaning the subject i want to train on ， what does "class images" mean? | closed | completed | false | 3 | [] | [] | 2022-11-07T03:41:07Z | 2022-11-08T06:07:10Z | 2022-11-08T06:07:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | universewill | 6,790,730 | MDQ6VXNlcjY3OTA3MzA= | User | false |
huggingface/diffusers | 3,133,147,976 | I_kwDOHa8MBc66wAtI | 11,687 | https://github.com/huggingface/diffusers/issues/11687 | https://api.github.com/repos/huggingface/diffusers/issues/11687 | [DOCS] Add `pruna` as optimization framework | **Is your feature request related to a problem? Please describe.**
N.A.
**Describe the solution you'd like.**
Have a section on Pruna AI within the documentation
**Describe alternatives you've considered.**
N.A.
**Additional context.**
N.A. | closed | completed | false | 0 | [] | [] | 2025-06-10T11:02:46Z | 2025-06-16T19:25:06Z | 2025-06-16T19:25:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | davidberenstein1957 | 25,269,220 | MDQ6VXNlcjI1MjY5MjIw | User | false |
huggingface/diffusers | 3,135,313,666 | I_kwDOHa8MBc664RcC | 11,691 | https://github.com/huggingface/diffusers/issues/11691 | https://api.github.com/repos/huggingface/diffusers/issues/11691 | <spam> | Pixel art sprite sheet of a Pokémon trainer character, Game Boy Advance style, 4 directions (front, back, left, right), 3 walking frames per direction, 24x32 pixels per frame, in a grid layPixel art sprite sheet of a Pokémon trainer character, Game Boy Advance style, 4 directions (front, back, left, right), 3 walking f... | closed | completed | false | 0 | [] | [] | 2025-06-11T02:48:31Z | 2025-06-11T02:53:06Z | 2025-06-11T02:53:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Gtvkun-ai | 184,384,261 | U_kgDOCv17BQ | User | false |
huggingface/diffusers | 3,135,315,176 | I_kwDOHa8MBc664Rzo | 11,692 | https://github.com/huggingface/diffusers/issues/11692 | https://api.github.com/repos/huggingface/diffusers/issues/11692 | <spam> | Pixel art sprite sheet of a Pokémon trainer character, Game Boy Advance style, 4 directions (front, back, left, right), 3 walking frames per direction, 24x32 pixels per frame, in a grid layPixel art sprite sheet of a Pokémon trainer character, Game Boy Advance style, 4 directions (front, back, left, right), 3 walking f... | closed | completed | false | 0 | [] | [] | 2025-06-11T02:49:15Z | 2025-06-11T02:53:19Z | 2025-06-11T02:53:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Gtvkun-ai | 184,384,261 | U_kgDOCv17BQ | User | false |
huggingface/diffusers | 3,135,895,093 | I_kwDOHa8MBc666fY1 | 11,694 | https://github.com/huggingface/diffusers/issues/11694 | https://api.github.com/repos/huggingface/diffusers/issues/11694 | Convert VAE from latent-diffusion to diffusers | Hi, I trained a VAE-KL as suggested [here](https://github.com/CompVis/latent-diffusion/tree/main#).
Now, I would like to plug this into diffusers-stable diffusion `train_text_to_image.py.`
To convert the VAE, I'm using `convert_vae_pt_to_diffusers.py`, however I'm getting this error at line 165:
```
RuntimeError: Er... | closed | completed | false | 0 | [] | [] | 2025-06-11T08:07:12Z | 2025-06-11T09:44:12Z | 2025-06-11T09:44:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | smandelli | 28,630,855 | MDQ6VXNlcjI4NjMwODU1 | User | false |
huggingface/diffusers | 3,139,382,012 | I_kwDOHa8MBc67Hyr8 | 11,700 | https://github.com/huggingface/diffusers/issues/11700 | https://api.github.com/repos/huggingface/diffusers/issues/11700 | Add pruna integration for loading model through diffusers.from_pretrained / pipeline | **Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...].
[Pruna](https://github.com/PrunaAI/pruna) is an open-source AI model optimisation framework.
As discussed with @SunMarc [about doing this for transformers](h... | closed | completed | false | 15 | [] | [] | 2025-06-12T08:42:07Z | 2025-08-12T06:48:49Z | 2025-08-12T06:48:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | davidberenstein1957 | 25,269,220 | MDQ6VXNlcjI1MjY5MjIw | User | false |
huggingface/diffusers | 3,141,696,074 | I_kwDOHa8MBc67QnpK | 11,702 | https://github.com/huggingface/diffusers/issues/11702 | https://api.github.com/repos/huggingface/diffusers/issues/11702 | AccVid LoRA key error when loading with diffusers | ### Describe the bug
Loading the AccVid LoRA isn't working with diffusers. For context [AccVid LoRA for Wan 2.1](https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_AccVid_T2V_14B_lora_rank32_fp16.safetensors) - it is another LoRA extracted by Kijai from a distilled model ([AccVid](https://github.com/aejion/Ac... | closed | completed | false | 0 | [
"bug"
] | [] | 2025-06-12T22:20:32Z | 2025-06-13T06:25:09Z | 2025-06-13T06:25:09Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 3,143,587,841 | I_kwDOHa8MBc67X1gB | 11,709 | https://github.com/huggingface/diffusers/issues/11709 | https://api.github.com/repos/huggingface/diffusers/issues/11709 | CausVid LoRA key error when loading with diffusers | ### Describe the bug
when use this [lora](https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_CausVid_14B_T2V_lora_rank32_v2.safetensors) with vace-diffusers. There will be a key loading error problem
### Reproduction
```python
import torch
import PIL.Image
from diffusers import AutoencoderKLWan, WanVACEPipe... | closed | completed | false | 1 | [
"bug"
] | [] | 2025-06-13T13:29:05Z | 2025-06-17T05:09:19Z | 2025-06-17T05:09:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | todochenxi | 108,149,467 | U_kgDOBnI62w | User | false |
huggingface/diffusers | 3,143,776,026 | I_kwDOHa8MBc67Yjca | 11,710 | https://github.com/huggingface/diffusers/issues/11710 | https://api.github.com/repos/huggingface/diffusers/issues/11710 | [FR] Please support MagCache: Fast Video Generation with Magnitude-Aware Cache | **Is your feature request related to a problem? Please describe.**
No problem. The results are good and less-latency.
Following models are supported
[Wan2.1](https://github.com/Wan-Video/Wan2.1)
[HunyuanVideo](https://github.com/Tencent/HunyuanVideo)
[FLUX](https://github.com/black-forest-labs/flux),
**Describe the so... | closed | completed | false | 3 | [] | [] | 2025-06-13T14:30:39Z | 2025-10-09T15:56:31Z | 2025-10-09T15:56:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,147,211,707 | I_kwDOHa8MBc67lqO7 | 11,714 | https://github.com/huggingface/diffusers/issues/11714 | https://api.github.com/repos/huggingface/diffusers/issues/11714 | Reporting a vulnerability | ### Describe the bug
Hello!
I hope you are doing well!
I am a security researcher and a bug bounty hunter . I detected a vulnerability in this repository. I want to disclose it responsibly, I got to know that **huggingface** have an active private bug bounty program running on **Hackerone**.
I found different vulns... | closed | completed | false | 3 | [
"bug"
] | [] | 2025-06-15T07:39:06Z | 2025-06-20T14:34:53Z | 2025-06-20T14:34:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nirajmodi19 | 26,089,044 | MDQ6VXNlcjI2MDg5MDQ0 | User | false |
huggingface/diffusers | 3,149,066,310 | I_kwDOHa8MBc67svBG | 11,720 | https://github.com/huggingface/diffusers/issues/11720 | https://api.github.com/repos/huggingface/diffusers/issues/11720 | add T5 lora train and infer for FluxPipeline | train and infer code of transformer lora + clip text encoder lora + T5 text encoder lora with FluxPipeline and FluxFillPipeline
| closed | completed | false | 2 | [
"stale"
] | [] | 2025-06-16T08:35:59Z | 2026-01-09T21:20:37Z | 2026-01-09T21:20:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | huangjun12 | 22,365,664 | MDQ6VXNlcjIyMzY1NjY0 | User | false |
huggingface/diffusers | 3,149,688,477 | I_kwDOHa8MBc67vG6d | 11,721 | https://github.com/huggingface/diffusers/issues/11721 | https://api.github.com/repos/huggingface/diffusers/issues/11721 | SanaPipeline crash on load | ### Describe the bug
The `SanaPipeline` crashes when loading from pre-trained. The issue seems to be in the Gemma-2 model (and in the `transformers` library).
All libraries (diffusers, transformers, accelerate) are up to date. PyTorch is a recent version (2.4.1).
### Reproduction
```python
from diffusers import San... | closed | completed | false | 12 | [
"bug"
] | [] | 2025-06-16T11:43:57Z | 2025-06-20T09:31:12Z | 2025-06-20T09:31:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | francois-rozet | 37,352,336 | MDQ6VXNlcjM3MzUyMzM2 | User | false |
huggingface/diffusers | 3,150,412,064 | I_kwDOHa8MBc67x3kg | 11,724 | https://github.com/huggingface/diffusers/issues/11724 | https://api.github.com/repos/huggingface/diffusers/issues/11724 | Chroma fails to mask attention in the transformer | ### Describe the bug
Chroma has attention masks on both the transformer and T5. Right now only T5 computes with the proper mask. The attention masks from T5 should carry forward into the transformer and mask padding tokens except for the last padding token. See the reference implementation.
https://github.com/lodesto... | closed | completed | false | 15 | [
"bug"
] | [] | 2025-06-16T15:21:53Z | 2025-06-18T17:03:09Z | 2025-06-18T17:03:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 3,152,778,316 | I_kwDOHa8MBc6765RM | 11,730 | https://github.com/huggingface/diffusers/issues/11730 | https://api.github.com/repos/huggingface/diffusers/issues/11730 | Add `--lora_alpha` and metadata handling in training scripts follow up | With #11707, #11723 we pushed some small changes to the way we save and parse metadata for trained LoRAs, which also allow us to add a `--lora_alpha` arg to the Dreambooth LoRA training scripts, making LoRA alpha also configurable.
This issue is to ask for help from the community to bring these changes to the other t... | closed | completed | false | 8 | [
"good first issue",
"contributions-welcome"
] | [
"linoytsaban"
] | 2025-06-17T09:29:24Z | 2025-06-24T10:58:54Z | 2025-06-23T12:46:45Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | linoytsaban | 57,615,435 | MDQ6VXNlcjU3NjE1NDM1 | User | false |
huggingface/diffusers | 3,153,129,018 | I_kwDOHa8MBc678O46 | 11,732 | https://github.com/huggingface/diffusers/issues/11732 | https://api.github.com/repos/huggingface/diffusers/issues/11732 | Regression? load_lora_weights ValueError: not enough values to unpack (expected 3, got 2) | ### Describe the bug
load_lora_weights seems to be broken recently?
### Reproduction
```python
import torch
from diffusers import AutoPipelineForText2Image
pipeline = AutoPipelineForText2Image.from_pretrained(
"ChuckMcSneed/FLUX.1-dev",
torch_dtype=torch.float16
).to("cuda")
pipeline.load_lora_weights(
... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-06-17T11:21:36Z | 2025-06-17T11:37:12Z | 2025-06-17T11:37:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 3,153,855,111 | I_kwDOHa8MBc67_AKH | 11,733 | https://github.com/huggingface/diffusers/issues/11733 | https://api.github.com/repos/huggingface/diffusers/issues/11733 | ControlNet LoRA support more broadly across models | **Is your feature request related to a problem? Please describe.**
In SimpleTuner, I'm currently extending ControlNet training to all implemented model architectures.
So far, Control LoRA seems to be the preferred way to handle these in Diffusers, but in my limited experimentation across architectures, ControlNet LoR... | open | null | false | 2 | [
"stale"
] | [] | 2025-06-17T15:07:01Z | 2026-01-09T15:19:01Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 3,154,714,904 | I_kwDOHa8MBc68CSEY | 11,734 | https://github.com/huggingface/diffusers/issues/11734 | https://api.github.com/repos/huggingface/diffusers/issues/11734 | FLUX.1-dev Controlnet Training Script for Custom Dataset. | Hi @sayakpaul @qgallouedec @DN6 , I’m currently working on a [FLUX.1-dev text-to-image pipeline](https://github.com/huggingface/diffusers/blob/1bc6f3dc0f21779480db70a4928d14282c0198ed/src/diffusers/pipelines/flux/pipeline_flux_controlnet.py#L67) and have been experimenting with InstantX/FLUX.1-dev-controlnet-canny. Whi... | closed | completed | true | 0 | [] | [] | 2025-06-17T20:40:03Z | 2025-06-18T01:17:06Z | 2025-06-18T01:17:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anandguptacv | 90,339,262 | MDQ6VXNlcjkwMzM5MjYy | User | false |
huggingface/diffusers | 3,156,671,715 | I_kwDOHa8MBc68Jvzj | 11,742 | https://github.com/huggingface/diffusers/issues/11742 | https://api.github.com/repos/huggingface/diffusers/issues/11742 | Optional 'prompt_2' parameter in encode_prompt method is not optional | ### Describe the bug
in pipeline_flux.py
The docstring says prompt_2 is optional but it is not declared as Optional[] and has no default None value
Minor issue, could be in other pipelines too
```
def encode_prompt(
self,
prompt: Union[str, List[str]],
prompt_2: Union[str, List[str]],
... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2025-06-18T12:22:26Z | 2026-01-09T21:23:57Z | 2026-01-09T21:23:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | japppie | 131,072,659 | U_kgDOB9ACkw | User | false |
huggingface/diffusers | 3,157,968,646 | I_kwDOHa8MBc68OscG | 11,745 | https://github.com/huggingface/diffusers/issues/11745 | https://api.github.com/repos/huggingface/diffusers/issues/11745 | Pipeline Quantization silent ignore of components not present in pipeline | ### Describe the bug
if you add a component that doesn't exist in the pipeline to the `PipelineQuantizationConfig`, the quantization works for the components present but it doesn't throw a warning that one of them doesn't exist, this could lead to wrong assumptions that it worked.
### Reproduction
```python
repo_id ... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-06-18T19:37:14Z | 2025-06-20T02:24:13Z | 2025-06-20T02:24:13Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | asomoza | 5,442,875 | MDQ6VXNlcjU0NDI4NzU= | User | false |
huggingface/diffusers | 3,158,756,121 | I_kwDOHa8MBc68RssZ | 11,749 | https://github.com/huggingface/diffusers/issues/11749 | https://api.github.com/repos/huggingface/diffusers/issues/11749 | FlowMatchEulerDiscreteScheduler Timesteps have Floating Point Errors | ### Describe the bug
When using the index_for_timestep method from FlowMatchEulerDiscreteScheduler in scheduling_flow_match_euler_discrete.py, the self.timesteps variable after __init__ maintains floating point values for timesteps (by default, 1000 to 1). Some timestep values have floating point discrepancies that ca... | open | null | false | 1 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2025-06-19T03:12:58Z | 2026-01-09T15:18:47Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JStyborski | 49,502,351 | MDQ6VXNlcjQ5NTAyMzUx | User | false |
huggingface/diffusers | 3,159,137,816 | I_kwDOHa8MBc68TJ4Y | 11,752 | https://github.com/huggingface/diffusers/issues/11752 | https://api.github.com/repos/huggingface/diffusers/issues/11752 | Failed to load lora in int8 mode | Let me introduce the background. The fill model changes the picture background. Torchao quantization is used for inference, but it fails when loading Lora. If Lora is not quantized, it can be loaded.
`transformer = FluxTransformer2DModel.from_pretrained(
model_path
, subfolder = "transformer"
... | open | null | false | 5 | [] | [] | 2025-06-19T06:34:40Z | 2026-02-04T15:18:02Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | babyta | 37,787,795 | MDQ6VXNlcjM3Nzg3Nzk1 | User | false |
huggingface/diffusers | 1,438,625,111 | I_kwDOHa8MBc5Vv61X | 1,176 | https://github.com/huggingface/diffusers/issues/1176 | https://api.github.com/repos/huggingface/diffusers/issues/1176 | AdamW degrades word embeddings in textual inversion training | ### Describe the bug
Text embeddings intended to be frozen/unchanged, but it doesn't happen.
First embedding trains against real embeddings, but as training go on, word embeddings decays to zero.
### Reproduction
Not sure if really a bug.
### Logs
_No response_
### System Info
Example on colab.
Code here too. | closed | completed | false | 8 | [
"bug",
"stale"
] | [
"patil-suraj"
] | 2022-11-07T16:17:38Z | 2022-12-22T15:18:17Z | 2022-12-22T15:18:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hadaev8 | 20,247,085 | MDQ6VXNlcjIwMjQ3MDg1 | User | false |
huggingface/diffusers | 3,162,388,480 | I_kwDOHa8MBc68fjgA | 11,761 | https://github.com/huggingface/diffusers/issues/11761 | https://api.github.com/repos/huggingface/diffusers/issues/11761 | Avoid creating tensor in CosmosAttnProcessor2_0 | ERROR: type should be string, got "https://github.com/huggingface/diffusers/blob/42077e6c734df2fc7bbed373abceab99635500ad/src/diffusers/models/transformers/transformer_cosmos.py#L188-L193\n```\n # 4. Prepare for GQA\n query_idx = query.size(3)\n key_idx = key.size(3)\n value_idx = value.size(3)\n key = key.repeat_interleave(query_idx // key_idx, dim=3)\n value = value.repeat_interleave(query_idx // value_idx, dim=3)\n```\nSpeedup ~10% here in Cosmos2TextToImagePipeline and Cosmos2VideoToWorldPipeline." | closed | completed | false | 1 | [] | [] | 2025-06-20T08:34:31Z | 2025-07-10T21:51:07Z | 2025-07-10T21:51:07Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chenxiao111222 | 154,797,505 | U_kgDOCToFwQ | User | false |
huggingface/diffusers | 3,162,569,136 | I_kwDOHa8MBc68gPmw | 11,762 | https://github.com/huggingface/diffusers/issues/11762 | https://api.github.com/repos/huggingface/diffusers/issues/11762 | Could you help fix the backdoor vulnerability caused by two risky pre-trained models used in this repo? | ### Describe the bug
Hi, @patrickvonplaten, @sayakpaul, I'd like to report that two potentially risky pretrained models are being used in this project, which may pose **backdoor threats**.Please check the following code example:
### Reproduction
• **tests/pipelines/stable_diffusion/test_onnx_stable_diffusion_upsc... | closed | completed | false | 3 | [
"bug"
] | [] | 2025-06-20T09:31:50Z | 2026-01-09T21:24:37Z | 2026-01-09T21:24:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Rockstar292 | 217,070,783 | U_kgDODPA8vw | User | false |
huggingface/diffusers | 3,165,014,973 | I_kwDOHa8MBc68pku9 | 11,764 | https://github.com/huggingface/diffusers/issues/11764 | https://api.github.com/repos/huggingface/diffusers/issues/11764 | RepositoryNotFoundError on inpainting demo | ### Describe the bug
Diffusers page https://huggingface.co/tasks/image-to-image#useful-resources links to demo https://huggingface.co/spaces/diffusers/stable-diffusion-xl-inpainting which opens with this:
```
runtime error
Exit code: 1. Reason: per(
File "/home/user/.pyenv/versions/3.10.13/lib/python3.10/site-packa... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2025-06-21T11:25:22Z | 2026-01-12T01:27:24Z | 2026-01-09T21:27:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | abitrolly | 8,781,107 | MDQ6VXNlcjg3ODExMDc= | User | false |
huggingface/diffusers | 3,165,675,525 | I_kwDOHa8MBc68sGAF | 11,766 | https://github.com/huggingface/diffusers/issues/11766 | https://api.github.com/repos/huggingface/diffusers/issues/11766 | Bug: set adapters when model is compiled | https://github.com/huggingface/diffusers/blob/7fc53b5d66867f9e59398f5a14dd3dd9fbc700dd/src/diffusers/loaders/lora_base.py#L839
I note that `transformer.__class__` is changed from `diffusers.models.transformers.transformer_flux.FluxTransformer2DModel` to `torch._dynamo.eval_frame.OptimizedModule` when I use `torch.compi... | closed | completed | false | 2 | [] | [] | 2025-06-22T05:26:34Z | 2025-07-05T21:14:06Z | 2025-07-05T21:14:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chenxiao111222 | 154,797,505 | U_kgDOCToFwQ | User | false |
huggingface/diffusers | 3,167,499,029 | I_kwDOHa8MBc68zDMV | 11,777 | https://github.com/huggingface/diffusers/issues/11777 | https://api.github.com/repos/huggingface/diffusers/issues/11777 | Some question about training lora for HiDream-I1 | ### Describe the bug
Has anyone successfully trained HiDream-I1 using the diffusers repository? I trained the lora model of HiDream-I1 with the default learning rate. The output images were extremely prone to crash and even turned into noisy images. However, this problem did not occur when using the ai-toolkit reposit... | closed | completed | true | 1 | [
"bug",
"stale"
] | [] | 2025-06-23T09:41:56Z | 2026-01-09T21:27:17Z | 2026-01-09T21:27:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yinguoweiOvO | 56,142,257 | MDQ6VXNlcjU2MTQyMjU3 | User | false |
huggingface/diffusers | 3,167,644,082 | I_kwDOHa8MBc68zmmy | 11,779 | https://github.com/huggingface/diffusers/issues/11779 | https://api.github.com/repos/huggingface/diffusers/issues/11779 | Incorrect mask padding in Wan VACE pipeline | ### Describe the bug
When using reference images in the Wan VACE pipeline, the reference image latents should be at the front of other latents, as well as the mask. However in the main branch the mask padding for the reference latents comes after masks for other latents:
https://github.com/huggingface/diffusers/blob/... | closed | completed | false | 0 | [
"bug"
] | [] | 2025-06-23T10:26:56Z | 2025-06-23T10:58:22Z | 2025-06-23T10:58:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bennyguo | 19,284,678 | MDQ6VXNlcjE5Mjg0Njc4 | User | false |
huggingface/diffusers | 3,168,215,615 | I_kwDOHa8MBc681yI_ | 11,781 | https://github.com/huggingface/diffusers/issues/11781 | https://api.github.com/repos/huggingface/diffusers/issues/11781 | support for MeanFlow | The meanflow paper requires two things
- support for multiple time embeddings.
- calculating jvp
## support for multiple time embeddings
my idea is to have a parameter called `multiple_time_embeddings`
```python
model = UNet2DModel(
sample_size=32,
in_channels=3,
out_channels=3,
layers_per_block=2... | closed | completed | true | 2 | [] | [] | 2025-06-23T13:28:47Z | 2025-07-05T21:08:59Z | 2025-07-05T21:08:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | prakashjayy | 11,383,400 | MDQ6VXNlcjExMzgzNDAw | User | false |
huggingface/diffusers | 3,168,397,309 | I_kwDOHa8MBc682ef9 | 11,784 | https://github.com/huggingface/diffusers/issues/11784 | https://api.github.com/repos/huggingface/diffusers/issues/11784 | Bug diffuser | ### Describe the bug
### Describe the bug
I'm running a Stable Diffusion-based app on Hugging Face Spaces using a custom Dockerfile. After upgrading to `diffusers==0.24.0`, I still encounter an error related to `hf_hub_download` or model loading, which seems to originate from `diffusers.utils.dynamic_modules_utils`.
... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2025-06-23T14:24:01Z | 2026-01-09T21:30:29Z | 2026-01-09T21:30:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Congofire | 217,612,516 | U_kgDODPiA5A | User | false |
huggingface/diffusers | 3,170,619,545 | I_kwDOHa8MBc68-9CZ | 11,790 | https://github.com/huggingface/diffusers/issues/11790 | https://api.github.com/repos/huggingface/diffusers/issues/11790 | Bug: lora trained with train_dreambooth_lora_sdxl_advanced.py, the weight cannot be loaded | ### Describe the bug
I have trained a sdxl lora with the script https://github.com/huggingface/diffusers/blob/7bc0a07b1947bcbe0e84bbe9ecf8ae2d234382c4/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py , and a loading error was encountered during the validation in the end of the training whenc... | open | null | false | 7 | [
"bug",
"stale"
] | [] | 2025-06-24T06:56:42Z | 2026-02-03T15:21:07Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | garychan22 | 108,175,311 | U_kgDOBnKfzw | User | false |
huggingface/diffusers | 3,170,958,091 | I_kwDOHa8MBc69APsL | 11,791 | https://github.com/huggingface/diffusers/issues/11791 | https://api.github.com/repos/huggingface/diffusers/issues/11791 | can't load lora weights after apply group offloading | ### Describe the bug
i want to dynamically switch the lora with applying group offloading ,but failed
### Reproduction
```python
from diffusers import HunyuanVideoPipeline, HunyuanVideoTransformer3DModel
from diffusers.utils import export_to_video
import torch
from diffusers.hooks import apply_group_offloading
mod... | closed | completed | false | 1 | [
"bug"
] | [] | 2025-06-24T08:31:44Z | 2025-08-06T08:27:04Z | 2025-06-27T17:50:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhangvia | 38,352,569 | MDQ6VXNlcjM4MzUyNTY5 | User | false |
huggingface/diffusers | 3,171,863,132 | I_kwDOHa8MBc69Dspc | 11,798 | https://github.com/huggingface/diffusers/issues/11798 | https://api.github.com/repos/huggingface/diffusers/issues/11798 | Single File and GGUF support of Cosmos-Predict2 | As a Mac user I'm usually restricted in the memory reduction options I can use, so would like to ask for GGUF support for the new Comos models. I have enough RAM for the 2B version but would like to be able to use the GGUF versions for the 14B model.
Unfortunately the current code for the transformer does not allow l... | closed | completed | false | 2 | [] | [] | 2025-06-24T12:43:20Z | 2025-07-01T12:33:00Z | 2025-07-01T12:33:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vargol | 62,868 | MDQ6VXNlcjYyODY4 | User | false |
huggingface/diffusers | 3,175,418,827 | I_kwDOHa8MBc69RQvL | 11,805 | https://github.com/huggingface/diffusers/issues/11805 | https://api.github.com/repos/huggingface/diffusers/issues/11805 | Flash Attention | ### Describe the bug
RuntimeError: Failed to import diffusers.pipelines.pipeline_utils because of the following error (look up to see its traceback):
Failed to import diffusers.models.autoencoders.autoencoder_kl because of the following error (look up to see its traceback):
Requires Flash-Attention version >=2.7.1,<=2... | closed | completed | false | 1 | [
"bug"
] | [] | 2025-06-25T12:11:57Z | 2025-06-25T12:26:31Z | 2025-06-25T12:26:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | johnnynunez | 22,727,137 | MDQ6VXNlcjIyNzI3MTM3 | User | false |
huggingface/diffusers | 1,438,839,528 | I_kwDOHa8MBc5VwvLo | 1,181 | https://github.com/huggingface/diffusers/issues/1181 | https://api.github.com/repos/huggingface/diffusers/issues/1181 | Training a model from scratch, not fine-tuning. | **Is your feature request related to a problem? Please describe.**
For philosophical/ethical reasons, I would like to try my hand at create a version of stable diffusion that uses only public domain images.
**Describe the solution you'd like**
I would like an example in the training scripts that show how to get a ... | closed | completed | false | 15 | [
"stale"
] | [] | 2022-11-07T19:07:13Z | 2024-08-08T13:14:59Z | 2023-01-14T15:03:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MatthewWaller | 5,520,521 | MDQ6VXNlcjU1MjA1MjE= | User | false |
huggingface/diffusers | 3,180,639,188 | I_kwDOHa8MBc69lLPU | 11,816 | https://github.com/huggingface/diffusers/issues/11816 | https://api.github.com/repos/huggingface/diffusers/issues/11816 | set_adapters performance degrades with the number of inactive adapters | ### Describe the bug
### Goal
Build an image-generation service with `StableDiffusionXLPipeline` that:
1. Keeps ~50 LoRA adapters resident in GPU VRAM.
2. For each request:
• activate **≤ 5** specific LoRAs via `pipeline.set_adapters(...)`
• run inference
• deactivate them (ready for the next request).
... | closed | completed | false | 27 | [
"bug"
] | [] | 2025-06-26T22:27:54Z | 2025-09-29T14:33:13Z | 2025-06-30T14:55:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hrazjan | 45,817,795 | MDQ6VXNlcjQ1ODE3Nzk1 | User | false |
huggingface/diffusers | 3,181,103,036 | I_kwDOHa8MBc69m8e8 | 11,817 | https://github.com/huggingface/diffusers/issues/11817 | https://api.github.com/repos/huggingface/diffusers/issues/11817 | flux_control support flux kontext dev | flux_control support flux kontext dev | closed | completed | false | 3 | [] | [] | 2025-06-27T02:24:10Z | 2025-07-01T03:45:48Z | 2025-07-01T03:45:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | universewill | 6,790,730 | MDQ6VXNlcjY3OTA3MzA= | User | false |
huggingface/diffusers | 3,182,184,274 | I_kwDOHa8MBc69rEdS | 11,819 | https://github.com/huggingface/diffusers/issues/11819 | https://api.github.com/repos/huggingface/diffusers/issues/11819 | LoRA Training Script for Image Editing with FLUX Kontext | **Is your feature request related to a problem? Please describe.**
Currently, the provided fine-tuning script for FLUX Kontext (train_dreambooth_lora_flux_kontext.py) focuses exclusively on T2I. This is no example to follow to fine-tune the model for instruction-based image editing tasks.
**Describe the solution you'd... | open | null | false | 2 | [
"stale"
] | [] | 2025-06-27T09:53:04Z | 2026-01-09T15:18:08Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Eastforward | 85,276,865 | MDQ6VXNlcjg1Mjc2ODY1 | User | false |
huggingface/diffusers | 3,182,849,680 | I_kwDOHa8MBc69tm6Q | 11,821 | https://github.com/huggingface/diffusers/issues/11821 | https://api.github.com/repos/huggingface/diffusers/issues/11821 | WanVACEPipeline: Passing a list of prompts is not yet supported. This may be supported in the future. | ### Describe the bug
Is it that the WanVACEPipeline doesn't support prompt_embed?
```
Traceback (most recent call last):
File "C:\aiOWN\diffuser_webui\WanVace_GGUF_BROKEN.py", line 122, in <module>
result = pipe(
^^^^^
File "C:\Users\nitin\miniconda3\envs\sddw-dev\Lib\site-packages\torch\utils\_c... | closed | completed | false | 0 | [
"bug"
] | [
"a-r-r-o-w"
] | 2025-06-27T13:28:30Z | 2025-09-24T09:45:05Z | 2025-09-24T09:45:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nitinmukesh | 2,102,186 | MDQ6VXNlcjIxMDIxODY= | User | false |
huggingface/diffusers | 3,184,488,435 | I_kwDOHa8MBc69z2_z | 11,824 | https://github.com/huggingface/diffusers/issues/11824 | https://api.github.com/repos/huggingface/diffusers/issues/11824 | Run FluxKontextPipeline with multi-images input, But An error occurred | ### Describe the bug
I Use this examples:
conditioning_images = [
Image.open("image1.jpg").convert("RGB").resize((512, 512)),
Image.open("image2.jpg").convert("RGB").resize((512, 512)),
]
output = pipe(
prompt="A surreal fusion of the input images in a fantasy environment",
image=conditioning_images,
... | closed | completed | false | 20 | [
"bug"
] | [] | 2025-06-28T03:04:45Z | 2025-09-16T08:42:56Z | 2025-07-09T15:10:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | veridone | 6,941,942 | MDQ6VXNlcjY5NDE5NDI= | User | false |
huggingface/diffusers | 3,184,820,046 | I_kwDOHa8MBc691H9O | 11,827 | https://github.com/huggingface/diffusers/issues/11827 | https://api.github.com/repos/huggingface/diffusers/issues/11827 | FLUX Kontext. Transformers error: safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer | ### Describe the bug
Kontext worked yesterday, but today there is an error running the included example:
`safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer`
### Reproduction
```python
import torch
from diffusers import FluxKontextPipeline
from diffusers.utils import load_im... | closed | completed | false | 2 | [
"bug"
] | [] | 2025-06-28T10:28:34Z | 2025-06-29T09:53:37Z | 2025-06-29T09:53:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 3,185,672,251 | I_kwDOHa8MBc694YA7 | 11,828 | https://github.com/huggingface/diffusers/issues/11828 | https://api.github.com/repos/huggingface/diffusers/issues/11828 | StableDiffusionXLControlNetUnionInpaintPipeline requires a float for controlnet_conditioning_scale & control_guidance_end | ### Describe the bug
### Summary
StableDiffusionXLControlNetUnionInpaintPipeline requires a float for `controlnet_conditioning_scale` & `control_guidance_end` when they should also accept a `List[float]`.
After PR #10723 the txt-to-img Union pipeline `pipeline_controlnet_union_sd_xl.py` accepts a list/tuple so each... | closed | completed | false | 3 | [
"bug",
"stale"
] | [
"asomoza"
] | 2025-06-29T02:30:12Z | 2026-01-09T18:48:18Z | 2026-01-09T18:48:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ewwwgiddings | 26,286,559 | MDQ6VXNlcjI2Mjg2NTU5 | User | false |
huggingface/diffusers | 1,439,077,015 | I_kwDOHa8MBc5VxpKX | 1,183 | https://github.com/huggingface/diffusers/issues/1183 | https://api.github.com/repos/huggingface/diffusers/issues/1183 | MPS device not properly tested in EulerAncestralDiscreteScheduler | ### Describe the bug
When the device set to mps, str(device) can return "mps:0".
The proper way to test the device would be device.type == "mps".
The wrong test is used in EulerAncestralDiscreteScheduler preventing to run the right code:
```
device = model_output.device if torch.is_tensor(model_output) else "cpu"
... | closed | completed | false | 2 | [
"bug"
] | [
"pcuenca"
] | 2022-11-07T22:13:44Z | 2022-11-08T12:13:16Z | 2022-11-08T12:13:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | florian-ivadolabs | 44,813,459 | MDQ6VXNlcjQ0ODEzNDU5 | User | false |
huggingface/diffusers | 3,187,234,004 | I_kwDOHa8MBc69-VTU | 11,830 | https://github.com/huggingface/diffusers/issues/11830 | https://api.github.com/repos/huggingface/diffusers/issues/11830 | framepack pipeline return pure black output when using torch.float16 | ### Describe the bug
when using torch.float16 dtype, frame pack pipeline return pure black output. is this normal? i want to use torch.backends.cuda.matmul.allow_fp16_accumulation = True to accelerate the inference of framepack. it is a new feature of torch 2.7. but ifound that even without this feature, framepack pip... | closed | completed | false | 5 | [
"bug"
] | [] | 2025-06-30T05:48:42Z | 2025-07-03T01:51:23Z | 2025-07-03T01:51:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhangvia | 38,352,569 | MDQ6VXNlcjM4MzUyNTY5 | User | false |
huggingface/diffusers | 3,188,574,934 | I_kwDOHa8MBc6-DcrW | 11,833 | https://github.com/huggingface/diffusers/issues/11833 | https://api.github.com/repos/huggingface/diffusers/issues/11833 | set_lora_device reports KeyError even though lora is loaded | ### Describe the bug
## Goal
Build an image-generation service with `StableDiffusionXLPipeline` that:
1. Keeps large number of LoRA adapters in CPU RAM
2. For each request:
• load ≤ 5 specific LoRAs into cuda using `pipeline.set_lora_device`
• run inference
• offload loras back to cpu.
## Issue
The `pipelin... | closed | completed | false | 5 | [
"bug"
] | [] | 2025-06-30T13:50:24Z | 2025-07-04T13:56:18Z | 2025-07-04T13:56:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hrazjan | 45,817,795 | MDQ6VXNlcjQ1ODE3Nzk1 | User | false |
huggingface/diffusers | 3,190,522,437 | I_kwDOHa8MBc6-K4JF | 11,839 | https://github.com/huggingface/diffusers/issues/11839 | https://api.github.com/repos/huggingface/diffusers/issues/11839 | Try load gguf for FluxKontextPipeline but invalid input size | ### Describe the bug
I try the gguf quantize method like FluxPipeline or FluxFillPipeline but fail, I use https://huggingface.co/bullerwins/FLUX.1-Kontext-dev-GGUF/blob/main/flux1-kontext-dev-Q4_K_M.gguf. I also tried resize image to width = (width // 64) * 64, height = (height // 64) * 64
### Reproduction
```
trans... | open | null | false | 13 | [
"bug"
] | [] | 2025-07-01T03:09:32Z | 2026-02-06T03:19:35Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dangph-alala | 198,682,075 | U_kgDOC9el2w | User | false |
huggingface/diffusers | 3,190,579,264 | I_kwDOHa8MBc6-LGBA | 11,840 | https://github.com/huggingface/diffusers/issues/11840 | https://api.github.com/repos/huggingface/diffusers/issues/11840 | finetune flux kontext on custom dataset | i want to finetune flux kontext on custom dataset, can diffusers do it? | open | null | false | 2 | [
"stale"
] | [] | 2025-07-01T03:47:21Z | 2026-01-09T15:17:53Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | universewill | 6,790,730 | MDQ6VXNlcjY3OTA3MzA= | User | false |
huggingface/diffusers | 3,192,293,707 | I_kwDOHa8MBc6-RolL | 11,846 | https://github.com/huggingface/diffusers/issues/11846 | https://api.github.com/repos/huggingface/diffusers/issues/11846 | Finetuning DreamBoot Inpainting | While finetuning the dream boot inpainting there are many issues coming related to the libraries and also in that script the masks on images are created randomply but i want to be the one to provide my own image mask to the model | closed | completed | false | 2 | [
"stale"
] | [] | 2025-07-01T12:28:35Z | 2026-01-09T21:44:33Z | 2026-01-09T21:44:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kkewat | 60,972,748 | MDQ6VXNlcjYwOTcyNzQ4 | User | false |
huggingface/diffusers | 3,195,991,069 | I_kwDOHa8MBc6-fvQd | 11,849 | https://github.com/huggingface/diffusers/issues/11849 | https://api.github.com/repos/huggingface/diffusers/issues/11849 | Can not load fusionx_lora into original wan2.1-14b | hello, i am adding the fusionx_lora into original wan2.1-14b-i2v, my code is as follow:
> pipe = WanImageToVideoPipeline.from_pretrained(my_local_path + "Wan2.1-I2V-14B-480P-Diffusers", vae=vae, image_encoder=image_encoder, torch_dtype=torch.bfloat16)
> pipe.load_lora_weights(
> my_local_path + "Wan14BT2VFusio... | open | null | false | 1 | [
"stale"
] | [] | 2025-07-02T13:48:17Z | 2026-01-09T15:17:45Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fzuo1230 | 84,018,094 | MDQ6VXNlcjg0MDE4MDk0 | User | false |
huggingface/diffusers | 1,439,576,509 | I_kwDOHa8MBc5VzjG9 | 1,185 | https://github.com/huggingface/diffusers/issues/1185 | https://api.github.com/repos/huggingface/diffusers/issues/1185 | Dreambooth results are not reproducible | ### Describe the bug
Disappointed that the results are not reproducible (others say the same as well).
In my case, the output image is very similar to one of the input images.
My steps:
```
pip install --upgrade diffusers[torch]
git clone https://github.com/huggingface/diffusers.git
```
```
cd diffusers/e... | closed | completed | false | 9 | [
"bug"
] | [
"patil-suraj"
] | 2022-11-08T06:29:27Z | 2022-11-13T00:45:10Z | 2022-11-09T17:33:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aurotripathy | 1,825,684 | MDQ6VXNlcjE4MjU2ODQ= | User | false |
huggingface/diffusers | 3,198,013,743 | I_kwDOHa8MBc6-ndEv | 11,853 | https://github.com/huggingface/diffusers/issues/11853 | https://api.github.com/repos/huggingface/diffusers/issues/11853 | Conflix size when load controlnet+IP-Adapter when load stable-diffusion-xl-base-1.0 | ### Describe the bug
It work when loading stable-diffusion-v1-5
But not work with stable-diffusion-xl-base-1.0
```python
depth_estimator = DPTForDepthEstimation.from_pretrained("Intel/dpt-hybrid-midas").to("cuda")
feature_extr... | closed | completed | false | 0 | [
"bug"
] | [] | 2025-07-03T04:52:40Z | 2025-07-03T04:54:34Z | 2025-07-03T04:54:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dangph-alala | 198,682,075 | U_kgDOC9el2w | User | false |
huggingface/diffusers | 3,203,348,041 | I_kwDOHa8MBc6-7zZJ | 11,859 | https://github.com/huggingface/diffusers/issues/11859 | https://api.github.com/repos/huggingface/diffusers/issues/11859 | STORK Scheduler | ### Model/Pipeline/Scheduler description
I would like to upload my new scheduler called STORK onto diffuser. The full description can be seen in the paper https://arxiv.org/abs/2505.24210.
### Open source status
- [x] The model implementation is available.
- [ ] The model weights are available (Only relevant if addi... | open | null | false | 1 | [
"stale"
] | [] | 2025-07-04T17:23:23Z | 2026-01-09T15:17:42Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ZT220501 | 61,394,893 | MDQ6VXNlcjYxMzk0ODkz | User | false |
huggingface/diffusers | 3,203,658,727 | I_kwDOHa8MBc6-8_Pn | 11,861 | https://github.com/huggingface/diffusers/issues/11861 | https://api.github.com/repos/huggingface/diffusers/issues/11861 | ControlnetUnion not working correctly | ### Describe the bug
After https://github.com/huggingface/diffusers/pull/10723 the way controlnet union works with the control images is not correct.
The more clear and concise way of showing this is using a depth map with the tile condition, the tile condition should return almost the same image but in this case it ... | closed | completed | false | 2 | [
"bug"
] | [
"asomoza"
] | 2025-07-04T20:35:21Z | 2025-07-10T18:35:30Z | 2025-07-10T18:35:30Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | asomoza | 5,442,875 | MDQ6VXNlcjU0NDI4NzU= | User | false |
huggingface/diffusers | 3,204,416,253 | I_kwDOHa8MBc6-_4L9 | 11,864 | https://github.com/huggingface/diffusers/issues/11864 | https://api.github.com/repos/huggingface/diffusers/issues/11864 | AutoencoderDC.encode fails with torch.compile(fullgraph=True) - "name 'torch' is not defined" | ### Describe the bug
I'm trying to optimize my data preprocessing pipeline for the Sana model by using `torch.compile` on the DC-AE encoder. Following PyTorch's best practices, I attempted to compile only the `encode` method with `fullgraph=True` for better performance, but I'm encountering an error.
When I try:
```p... | closed | completed | false | 6 | [
"bug"
] | [] | 2025-07-05T06:15:11Z | 2025-07-09T01:32:39Z | 2025-07-09T01:32:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SingleBicycle | 109,320,283 | U_kgDOBoQYWw | User | false |
huggingface/diffusers | 3,204,866,396 | I_kwDOHa8MBc6_BmFc | 11,865 | https://github.com/huggingface/diffusers/issues/11865 | https://api.github.com/repos/huggingface/diffusers/issues/11865 | AttributeError: type object 'CosmosTransformer3DModel' has no attribute 'from_single_file' | ### Describe the bug
I would like to run the Cosmos-Predict2-14B-Text2Image model, but it is too large to fit in 24GB of VRAM normally, so I tried to load a Q8_0 GGUF quantization. I copied some code from the [HiDreamImageTransformer2DModel](https://huggingface.co/docs/diffusers/en/api/models/hidream_image_transformer... | closed | completed | false | 9 | [
"bug"
] | [] | 2025-07-05T12:14:50Z | 2025-07-11T07:15:23Z | 2025-07-11T07:15:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mingyi456 | 65,432,252 | MDQ6VXNlcjY1NDMyMjUy | User | false |
huggingface/diffusers | 3,205,075,148 | I_kwDOHa8MBc6_CZDM | 11,866 | https://github.com/huggingface/diffusers/issues/11866 | https://api.github.com/repos/huggingface/diffusers/issues/11866 | [Docs] Visualizing the codebase for diffusers | I am creating this issue as a follow up from https://github.com/huggingface/diffusers/discussions/11668. Big thanks to @TensorCruncher for the support.
Based on the positive feedback, I’d like to propose a small documentation improvement for `diffusers`: integrating a visual onboarding diagram to help new contributors... | closed | completed | true | 0 | [] | [] | 2025-07-05T15:11:29Z | 2025-07-15T19:41:40Z | 2025-07-15T19:41:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ivanmilevtues | 10,105,224 | MDQ6VXNlcjEwMTA1MjI0 | User | false |
huggingface/diffusers | 3,205,278,336 | I_kwDOHa8MBc6_DKqA | 11,867 | https://github.com/huggingface/diffusers/issues/11867 | https://api.github.com/repos/huggingface/diffusers/issues/11867 | Question about control sd3 | I've noticed a potential inconsistency in how the VAE-encoded control_image is processed between the training script for ControlNet with Stable Diffusion 3 and the corresponding inference pipeline.
In the inference pipeline (pipeline_stable_diffusion_3_controlnet.py):
The control_image latent is processed by both sub... | closed | completed | false | 5 | [] | [] | 2025-07-05T17:07:13Z | 2025-07-14T18:55:56Z | 2025-07-14T18:55:56Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Henry-Bi | 25,079,406 | MDQ6VXNlcjI1MDc5NDA2 | User | false |
huggingface/diffusers | 3,206,278,254 | I_kwDOHa8MBc6_G-xu | 11,872 | https://github.com/huggingface/diffusers/issues/11872 | https://api.github.com/repos/huggingface/diffusers/issues/11872 | CPU Offloading Inefficiency: UNet/Transformer Repeatedly Moved Between CPU/GPU During Loops | ### Describe the bug
# CPU Offloading Inefficiency: UNet/Transformer Repeatedly Moved Between CPU/GPU During Denoising Loop
## Issue
The current CPU offloading mechanism (`enable_model_cpu_offload()`) in Diffusers is inefficient for models with denoising loops. It repeatedly moves the transformer/UNet component betw... | closed | completed | false | 14 | [
"bug"
] | [] | 2025-07-06T08:09:12Z | 2025-08-07T12:15:50Z | 2025-08-07T12:15:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JoeGaffney | 11,944,251 | MDQ6VXNlcjExOTQ0MjUx | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.