repo stringclasses 1
value | github_id int64 1.27B 4.24B | github_node_id stringlengths 18 18 | number int64 8 13.4k | 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 2026-04-10 09:58:27 | updated_at stringdate 2022-06-12 22:18:01 2026-04-10 19:54:38 | closed_at stringdate 2022-06-12 22:18:01 2026-04-10 19:54:38 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 2
values | extracted_at stringdate 2026-04-07 13:34:13 2026-04-10 21:59:46 | author_login stringlengths 3 28 | author_id int64 1.54k 258M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 1,272,010,353 | I_kwDOHa8MBc5L0VZx | 10 | https://github.com/huggingface/diffusers/issues/10 | https://api.github.com/repos/huggingface/diffusers/issues/10 | Missing `transformers` requirement | Dependencies don't mention `transformers`
https://github.com/huggingface/diffusers/blob/850d43450f2154d586c1cb226ca17ec53765e2a2/setup.py#L78-L91
However, importing from the quickstart fails without `transformers` installed:
```py
In [1]: import torch
In [2]: from diffusers import UNetModel, DDPMScheduler
... | closed | completed | false | 2 | [] | [
"anton-l"
] | 2022-06-15T10:21:14Z | 2022-06-22T11:48:28Z | 2022-06-22T11:48:28Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LysandreJik | 30,755,778 | MDQ6VXNlcjMwNzU1Nzc4 | User | false |
huggingface/diffusers | 1,424,414,610 | I_kwDOHa8MBc5U5teS | 1,000 | https://github.com/huggingface/diffusers/issues/1000 | https://api.github.com/repos/huggingface/diffusers/issues/1000 | `StableDiffusionInpaintPipeline` should handle batches | I'd like to batch inferences with `StableDiffusionInpaintPipeline`. Most of the code in `pipeline_stable_diffusion_inpaint.py` can do it **except** for `prepare_mask_and_masked_image(image, mask)` which **requires** a `PIL.Image`.
**Describe the solution you'd like**
A simple fix would be to just let `tensors` alon... | closed | completed | false | 2 | [
"stale"
] | [] | 2022-10-26T17:17:48Z | 2022-11-30T12:04:46Z | 2022-11-30T12:04:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vict0rsch | 9,283,470 | MDQ6VXNlcjkyODM0NzA= | User | false |
huggingface/diffusers | 2,686,122,280 | I_kwDOHa8MBc6gGvko | 10,003 | https://github.com/huggingface/diffusers/issues/10003 | https://api.github.com/repos/huggingface/diffusers/issues/10003 | DDIMScheduler (or DDPM) of StableDiffusion with 1000 num_inference_steps bugs out | ### Describe the bug
There's something going on with the set_timesteps offset parameters on Stable Diffusion (1v4). The timesteps are set from 1->1000 instead from 0, and so it tries to index out of bounds
### Reproduction
```
model_id = "CompVis/stable-diffusion-v1-4"
scheduler = DDIMScheduler.from_pretrained... | open | null | false | 10 | [
"bug",
"stale"
] | [] | 2024-11-23T15:18:23Z | 2025-02-03T15:04:00Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | HilaManor | 53,814,901 | MDQ6VXNlcjUzODE0OTAx | User | false |
huggingface/diffusers | 2,686,596,889 | I_kwDOHa8MBc6gIjcZ | 10,004 | https://github.com/huggingface/diffusers/issues/10004 | https://api.github.com/repos/huggingface/diffusers/issues/10004 | how to use kohya sd-scripts flux loras with text encoder keys in diffusers? | resulting lora weights from setting train text encoder to true is incompatible with diffusers load_lora_weights. the script networks/convert_flux_lora.py does not convert the text encoder keys either. | open | null | false | 12 | [
"contributions-welcome"
] | [] | 2024-11-23T20:54:30Z | 2025-03-16T15:39:25Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 2,689,130,458 | I_kwDOHa8MBc6gSN_a | 10,010 | https://github.com/huggingface/diffusers/issues/10010 | https://api.github.com/repos/huggingface/diffusers/issues/10010 | FLUX Fill example not working due to Not Found for url: https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev/resolve/main/model_index.json | ### Describe the bug
FLUX Fill is example not working due to Not Found for url: https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev/resolve/main/model_index.json
### Reproduction
Execute the FLUX Fill example according to doc.
### Logs
_No response_
### System Info
diffusers main
### Who can help?
_No res... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-11-25T04:50:06Z | 2024-11-25T04:53:33Z | 2024-11-25T04:53:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | petergerten | 17,756,065 | MDQ6VXNlcjE3NzU2MDY1 | User | false |
huggingface/diffusers | 2,690,235,944 | I_kwDOHa8MBc6gWb4o | 10,013 | https://github.com/huggingface/diffusers/issues/10013 | https://api.github.com/repos/huggingface/diffusers/issues/10013 | [Core] refactor model loading | Currently, we have got two codepaths:
1. For non-sharded checkpoints we do: https://github.com/huggingface/diffusers/blob/047bf492914ddc9393070b8f73bba5ad5823eb29/src/diffusers/models/modeling_utils.py#L855
2. For sharded checkpoints we do: https://github.com/huggingface/diffusers/blob/047bf492914ddc9393070b8f73bba... | closed | completed | false | 2 | [
"wip",
"roadmap"
] | [
"sayakpaul",
"SunMarc"
] | 2024-11-25T10:44:58Z | 2025-02-19T12:04:55Z | 2025-02-19T12:04:55Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,690,920,364 | I_kwDOHa8MBc6gZC-s | 10,016 | https://github.com/huggingface/diffusers/issues/10016 | https://api.github.com/repos/huggingface/diffusers/issues/10016 | Cannot load SD3.5M with from_single_file, mismatched shape for pos_embed.pos_embed in empty state dict | ### Describe the bug
When trying to load the Stable Diffusion 3.5 Medium checkpoint from Stability AI with the following command:
pipe = diffusers.StableDiffusion3Pipeline.from_single_file( '...path/sd3.5_medium.safetensors', text_encoder=None, text_encoder_2=None, text_encoder_3=None )
The following error is ... | closed | completed | false | 3 | [
"bug"
] | [
"DN6"
] | 2024-11-25T14:25:18Z | 2024-12-11T07:41:41Z | 2024-12-03T01:21:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CodeExplode | 79,982,924 | MDQ6VXNlcjc5OTgyOTI0 | User | false |
huggingface/diffusers | 2,691,321,209 | I_kwDOHa8MBc6gak15 | 10,017 | https://github.com/huggingface/diffusers/issues/10017 | https://api.github.com/repos/huggingface/diffusers/issues/10017 | loading dataset issue with load_dataset() | ### Describe the bug
i'm unable to load my dataset for [controlnet training](https://github.com/huggingface/diffusers/blob/074e12358bc17e7dbe111ea4f62f05dbae8a49d5/examples/controlnet/train_controlnet.py#L606) using load_dataset(). however, load_from_disk() seems to work?
would appreciate if someone can explain wh... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-11-25T16:15:45Z | 2024-11-26T03:32:51Z | 2024-11-26T03:32:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sarahahtee | 81,594,044 | MDQ6VXNlcjgxNTk0MDQ0 | User | false |
huggingface/diffusers | 2,691,771,169 | I_kwDOHa8MBc6gcSsh | 10,018 | https://github.com/huggingface/diffusers/issues/10018 | https://api.github.com/repos/huggingface/diffusers/issues/10018 | no model_index.json for flux tools | cannot load any flux tools pipelines (fluxfill, fluxcontrol, etc) all throw a model_index.json missing error. | closed | completed | false | 2 | [] | [] | 2024-11-25T18:22:53Z | 2024-11-25T23:55:23Z | 2024-11-25T23:55:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 2,692,979,032 | I_kwDOHa8MBc6gg5lY | 10,022 | https://github.com/huggingface/diffusers/issues/10022 | https://api.github.com/repos/huggingface/diffusers/issues/10022 | [core] refactor `attention_processor.py` the easy way | With @DN6 we have been discussing an idea about breaking up `src/diffusers/models/attention_processor.py` as it's getting excruciatingly longer and longer. The idea is simple and won't very likely require multiple rounds of PRs.
* Create a module named `attention_processor`.
* Split the attention processor classes... | open | null | false | 6 | [
"wip"
] | [
"DN6"
] | 2024-11-26T03:37:33Z | 2026-04-01T07:55:23Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,692,982,333 | I_kwDOHa8MBc6gg6Y9 | 10,023 | https://github.com/huggingface/diffusers/issues/10023 | https://api.github.com/repos/huggingface/diffusers/issues/10023 | Parallel Inference support ffor Dit model | does diffusers have any plan to support Parallel Inference like TP/PP/SP？ | open | null | false | 3 | [
"stale"
] | [] | 2024-11-26T03:40:48Z | 2024-12-26T15:02:57Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | enozhu | 37,290,567 | MDQ6VXNlcjM3MjkwNTY3 | User | false |
huggingface/diffusers | 2,693,777,714 | I_kwDOHa8MBc6gj8ky | 10,025 | https://github.com/huggingface/diffusers/issues/10025 | https://api.github.com/repos/huggingface/diffusers/issues/10025 | attention mask for transformer Flux | ### Describe the bug
Is it possible to get back the `attention_mask` argument in the flux attention processor
```
hidden_states = F.scaled_dot_product_attention(query, key, value, dropout_p=0.0, is_causal=False,attn_mask=attention_mask)
```
https://github.com/huggingface/diffusers/blob/main/src/diffusers/mo... | closed | completed | false | 19 | [
"bug"
] | [] | 2024-11-26T08:51:20Z | 2024-12-05T00:22:37Z | 2024-12-05T00:22:37Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,693,996,392 | I_kwDOHa8MBc6gkx9o | 10,026 | https://github.com/huggingface/diffusers/issues/10026 | https://api.github.com/repos/huggingface/diffusers/issues/10026 | Flux ControlNet Training Multi-GPU DeepSpeed Stage-3 doesn't reduce memory compared to Single GPU | ### Describe the bug
I am running a slightly modified version of [Flux ControlNet training script in diffusers](https://github.com/huggingface/diffusers/blob/main/examples/controlnet/train_controlnet_flux.py). The script is attached below. I am using DeepSpeed Stage-3 with the accelerate config below.
When I use on... | closed | completed | true | 1 | [
"bug"
] | [] | 2024-11-26T09:59:56Z | 2024-11-26T11:13:26Z | 2024-11-26T11:13:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | enesmsahin | 36,894,859 | MDQ6VXNlcjM2ODk0ODU5 | User | false |
huggingface/diffusers | 2,695,890,786 | I_kwDOHa8MBc6gsAdi | 10,029 | https://github.com/huggingface/diffusers/issues/10029 | https://api.github.com/repos/huggingface/diffusers/issues/10029 | Linear' object has no attribute 'base_layer | ### Describe the bug
I was trying to inference flux lora , lora was finetuned on ai toolkit
I have installed diffusers from source
while inferencing with lora , I am getting this error ===> Linear' object has no attribute 'base_layer
Please let me know if you need any info
Thanks
... | open | null | false | 2 | [
"bug",
"stale"
] | [] | 2024-11-26T19:37:55Z | 2024-12-27T15:02:59Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sandeep-Narahari | 46,187,761 | MDQ6VXNlcjQ2MTg3NzYx | User | false |
huggingface/diffusers | 2,696,254,448 | I_kwDOHa8MBc6gtZPw | 10,030 | https://github.com/huggingface/diffusers/issues/10030 | https://api.github.com/repos/huggingface/diffusers/issues/10030 | ControlNet expects added_cond_kwargs to not be null, which pipeline_controlnet_img2img does not respect | ### Describe the bug
Please see this code in pipeline_control_img2img.py
https://github.com/huggingface/diffusers/blob/ad5ecd1251472dbc69da1268671d41bc2d8c1caa/src/diffusers/pipelines/controlnet/pipeline_controlnet_img2img.py#L1199-L1203
### Reproduction
I think the pipeline is just straight up bugged sin... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-11-26T22:19:13Z | 2024-11-26T22:55:35Z | 2024-11-26T22:55:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | oxysoft | 3,945,277 | MDQ6VXNlcjM5NDUyNzc= | User | false |
huggingface/diffusers | 2,697,762,971 | I_kwDOHa8MBc6gzJib | 10,034 | https://github.com/huggingface/diffusers/issues/10034 | https://api.github.com/repos/huggingface/diffusers/issues/10034 | Add Support for Lightricks/LTX-Video Model and Video Generation Features (text2video, image2video, video2video) in Diffusers with Sequential CPU Offload Compatibility | **Is your feature request related to a problem? Please describe.**
Currently, the Diffusers library does not provide native support for advanced video generation tasks such as text-to-video (text2video), image-to-video (image2video), or video-to-video (video2video) generation for lighttricks/LTX-Video model. With the ... | closed | completed | false | 2 | [] | [] | 2024-11-27T09:10:41Z | 2024-11-28T05:33:37Z | 2024-11-28T05:33:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | BasimBashir | 45,428,008 | MDQ6VXNlcjQ1NDI4MDA4 | User | false |
huggingface/diffusers | 2,698,659,585 | I_kwDOHa8MBc6g2kcB | 10,035 | https://github.com/huggingface/diffusers/issues/10035 | https://api.github.com/repos/huggingface/diffusers/issues/10035 | CogvideoX VAE decoder consumes significantly more memory in the latest version | ### Describe the bug
The memory consumption for CogVideoX decoder in diffusers 0.31.0 version consumes significantly more memory. To the point where the model goes OOM even on 80G H100 GPUs with a relatively modest frame count.
I include two profiles for very small input tensors of only 5 frames where its visible ... | closed | completed | true | 1 | [
"bug"
] | [] | 2024-11-27T14:02:53Z | 2024-11-27T15:49:18Z | 2024-11-27T15:49:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ic-synth | 170,024,644 | U_kgDOCiJexA | User | false |
huggingface/diffusers | 2,701,086,646 | I_kwDOHa8MBc6g_0-2 | 10,039 | https://github.com/huggingface/diffusers/issues/10039 | https://api.github.com/repos/huggingface/diffusers/issues/10039 | Forgot to pass the output sample size in AttnUpBlock2D() in unet_2d_blocks.py | ### Describe the bug
Hi, I found that the parameter upsample_size was not passed into the upsampler in AttnUpBlock2D() in unet_2d_blocks.py [Line: 2386]. This can result in mismatched dimensions when processing images using AttnUpBlock2D(). This does not seem to have been corrected in the latest GitHub version [v0.31.... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-11-28T07:14:19Z | 2024-12-28T19:58:17Z | 2024-12-28T19:58:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xyliu6666 | 92,036,814 | U_kgDOBXxezg | User | false |
huggingface/diffusers | 2,701,283,255 | I_kwDOHa8MBc6hAk-3 | 10,040 | https://github.com/huggingface/diffusers/issues/10040 | https://api.github.com/repos/huggingface/diffusers/issues/10040 | [BUG - STABLE DIFFUSION 3] Grey images generated | ### Describe the bug
I'm running the SD3 model [stabilityai/stable-diffusion-3-medium](https://huggingface.co/stabilityai/stable-diffusion-3-medium) with the following settings:
- Height: 1024
- Width: 1024
- Inference steps: 50
- Guidance scale: 7
- Prompts length: 32 (using the same input prompt: "A men jumps... | closed | completed | false | 13 | [
"bug"
] | [] | 2024-11-28T08:31:27Z | 2024-12-02T23:33:58Z | 2024-12-02T23:33:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hoangvictor | 41,613,226 | MDQ6VXNlcjQxNjEzMjI2 | User | false |
huggingface/diffusers | 2,701,318,031 | I_kwDOHa8MBc6hAteP | 10,041 | https://github.com/huggingface/diffusers/issues/10041 | https://api.github.com/repos/huggingface/diffusers/issues/10041 | 🙋Flexible Directory Structure Support for Custom Pipeline Scenarios | ## 🚨 Flexible Directory Structure for Custom Pipeline Scenarios Raised Error!
When I have defined some custom_pipelines in local and the directory structure looks like:
```bash
custom_pipelines
├── custom_modules
│ ├── autoencoders
│ │ └── vae_a.py
│ └── transformers
│ ├── transformer_a.py
│ ... | open | null | false | 4 | [
"stale"
] | [
"a-r-r-o-w"
] | 2024-11-28T08:45:20Z | 2025-01-22T15:03:43Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | townwish4git | 143,256,262 | U_kgDOCInqxg | User | false |
huggingface/diffusers | 2,701,802,452 | I_kwDOHa8MBc6hCjvU | 10,043 | https://github.com/huggingface/diffusers/issues/10043 | https://api.github.com/repos/huggingface/diffusers/issues/10043 | F5-TTS Integration | ### Model/Pipeline/Scheduler description
F5-TTS is a fully non-autoregressive text-to-speech system based on flow matching with Diffusion Transformer (DiT).
It has excellent voice cloning capabilities, and audio generation is of quite high quality.
### Open source status
- [X] The model implementation is available.... | open | reopened | false | 11 | [
"help wanted",
"contributions-welcome"
] | [] | 2024-11-28T11:14:18Z | 2025-11-02T18:46:02Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nityanandmathur | 77,379,835 | MDQ6VXNlcjc3Mzc5ODM1 | User | false |
huggingface/diffusers | 2,702,055,025 | I_kwDOHa8MBc6hDhZx | 10,045 | https://github.com/huggingface/diffusers/issues/10045 | https://api.github.com/repos/huggingface/diffusers/issues/10045 | Problem with importing pipelines | ### Describe the bug
When trying to import pipelines in colab i have the following error:
RuntimeError Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/transformers/utils/import_utils.py](https://localhost:8080/#) in _get_module(self, module_name)
1777 ... | closed | completed | false | 0 | [
"bug"
] | [] | 2024-11-28T12:50:50Z | 2024-11-28T18:03:57Z | 2024-11-28T18:03:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stickman712 | 159,720,673 | U_kgDOCYUk4Q | User | false |
huggingface/diffusers | 2,702,916,663 | I_kwDOHa8MBc6hGzw3 | 10,049 | https://github.com/huggingface/diffusers/issues/10049 | https://api.github.com/repos/huggingface/diffusers/issues/10049 | Cannot import MochiPipeline | ### Describe the bug
Can't import mochi in colab. When trying have a following error:
ImportError Traceback (most recent call last)
[<ipython-input-5-6aeec58dc926>](https://localhost:8080/#) in <cell line: 2>()
1 import torch
----> 2 from diffusers import MochiPipeline
... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-11-28T18:11:49Z | 2024-11-28T18:21:12Z | 2024-11-28T18:21:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stickman712 | 159,720,673 | U_kgDOCYUk4Q | User | false |
huggingface/diffusers | 1,424,800,551 | I_kwDOHa8MBc5U7Lsn | 1,005 | https://github.com/huggingface/diffusers/issues/1005 | https://api.github.com/repos/huggingface/diffusers/issues/1005 | num_inference_timesteps can't exceed num_train_timesteps? | ### Describe the bug
Hey all,
I recently trained an unconditional image generation DDPM pipeline on a custom dataset using the `train_unconditional.py` script. The scheduler specified during training had `num_train_timesteps=1000` as follows:
```
noise_scheduler = DDPMScheduler(num_train_timesteps=1000)
```
... | closed | completed | false | 2 | [
"bug"
] | [] | 2022-10-26T23:08:36Z | 2022-10-30T02:43:50Z | 2022-10-30T02:43:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mihail911 | 2,789,441 | MDQ6VXNlcjI3ODk0NDE= | User | false |
huggingface/diffusers | 2,704,257,129 | I_kwDOHa8MBc6hL7Bp | 10,050 | https://github.com/huggingface/diffusers/issues/10050 | https://api.github.com/repos/huggingface/diffusers/issues/10050 | Is there any img2img KDiffusion equivalent of StableDiffusionKDiffusionPipeline? | ### Model/Pipeline/Scheduler description
I'm working on result alignment between diffusers and A1111 webui.
In txt2img scene, I can achieve via `StableDiffusionKDiffusionPipeline`, refer to https://github.com/huggingface/diffusers/issues/3253.
But in img2img scene, is there any KDiffusion pipeline equivalent?
I... | open | null | false | 2 | [
"stale"
] | [] | 2024-11-29T07:47:11Z | 2024-12-29T15:03:05Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | juju812 | 19,584,910 | MDQ6VXNlcjE5NTg0OTEw | User | false |
huggingface/diffusers | 2,704,631,827 | I_kwDOHa8MBc6hNWgT | 10,052 | https://github.com/huggingface/diffusers/issues/10052 | https://api.github.com/repos/huggingface/diffusers/issues/10052 | syntax error on ctx | ### Describe the bug
Across [several files](https://github.com/search?q=repo%3Ahuggingface%2Fdiffusers%20CTX&type=code), a bool statement is wrongly constructed IMO
One example: https://github.com/huggingface/diffusers/blob/c96bfa5c80eca798d555a79a491043c311d0f608/scripts/convert_sd3_to_diffusers.py#L14 => instea... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-11-29T10:13:59Z | 2024-12-02T05:58:02Z | 2024-12-02T05:58:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Clement-Lelievre | 70,368,164 | MDQ6VXNlcjcwMzY4MTY0 | User | false |
huggingface/diffusers | 2,704,908,993 | I_kwDOHa8MBc6hOaLB | 10,054 | https://github.com/huggingface/diffusers/issues/10054 | https://api.github.com/repos/huggingface/diffusers/issues/10054 | Testing beta sigmas in EulerDiscreteScheduler passes wrong parameters to torch.linspace | ### Describe the bug
Creating a EulerDiscreteScheduler using an existing schedulers config seems to pass a `FrozenDict` for `num_train_timesteps` instead of the expected `int`
e.g.
```py
pipe.scheduler = EulerDiscreteScheduler(pipe.scheduler.config, timestep_spacing="linspace", use_beta_sigmas=True)
```
fails w... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-11-29T11:47:28Z | 2024-11-29T11:55:34Z | 2024-11-29T11:55:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vargol | 62,868 | MDQ6VXNlcjYyODY4 | User | false |
huggingface/diffusers | 2,705,183,952 | I_kwDOHa8MBc6hPdTQ | 10,055 | https://github.com/huggingface/diffusers/issues/10055 | https://api.github.com/repos/huggingface/diffusers/issues/10055 | Training script for a Controlnet based on SD3 does not work | ### Describe the bug
Hi @sayakpaul and all others :)
The training script for a Control-net based on Stable Diffusion 3 seems to not work.
**RuntimeError: Given groups=1, weight of size [1536, 17, 2, 2], expected input[4, 16, 64, 64] to have 17 channels, but got 16 channels instead**
I tried to follow th... | open | null | false | 17 | [
"bug",
"stale"
] | [] | 2024-11-29T13:46:29Z | 2025-02-03T15:03:46Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Putzzmunta | 29,607,612 | MDQ6VXNlcjI5NjA3NjEy | User | false |
huggingface/diffusers | 2,705,420,279 | I_kwDOHa8MBc6hQW_3 | 10,057 | https://github.com/huggingface/diffusers/issues/10057 | https://api.github.com/repos/huggingface/diffusers/issues/10057 | Token Issue with FluxTransformer2DModel.from_single_file Method | ### Describe the bug
When I use the FluxTransformer2DModel.from_single_file method, the provided token seems to be invalid, and I am unable to download files that require token verification, such as the official Flux model files.
However, I can confirm that the token is valid because when using FluxPipeline.from_... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-11-29T15:18:06Z | 2024-12-28T17:06:37Z | 2024-12-03T01:15:34Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CyberVy | 72,680,847 | MDQ6VXNlcjcyNjgwODQ3 | User | false |
huggingface/diffusers | 2,706,472,451 | I_kwDOHa8MBc6hUX4D | 10,059 | https://github.com/huggingface/diffusers/issues/10059 | https://api.github.com/repos/huggingface/diffusers/issues/10059 | [core] introduce `AutoModel` | Now that we have quite a few model classes (and transformer model classes), I wonder if it makes sense to have an AutoModel class. Oftentimes, users initialize the model separately and pass it to a `DiffusionPipeline`:
```py
transformer = "..."
pipeline = DiffusionPipeline.from_pretrained("black-forest-labs/FLUX.1... | closed | completed | false | 0 | [
"enhancement"
] | [] | 2024-11-30T01:24:39Z | 2025-04-09T09:50:10Z | 2025-04-09T09:50:10Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,707,499,364 | I_kwDOHa8MBc6hYSlk | 10,062 | https://github.com/huggingface/diffusers/issues/10062 | https://api.github.com/repos/huggingface/diffusers/issues/10062 | 'UNet2DConditionModel' object has no attribute 'peft_config' | ### Describe the bug
After pulling the latest code from the _examples/research_projects/instructpix2pix_lora_ repository, I encountered the following error when running the default code:
`AttributeError: 'Linear' object has no attribute 'set_lora_layer'`
I resolved this issue by modifying the code.
The training now... | open | null | false | 6 | [
"bug",
"stale"
] | [
"Aiden-Frost"
] | 2024-11-30T15:25:58Z | 2025-01-06T15:03:26Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | duduke121 | 144,907,684 | U_kgDOCKMdpA | User | false |
huggingface/diffusers | 2,708,432,047 | I_kwDOHa8MBc6hb2Sv | 10,066 | https://github.com/huggingface/diffusers/issues/10066 | https://api.github.com/repos/huggingface/diffusers/issues/10066 | self.attn ValueError: not enough values to unpack (expected 2, got 1) | ### Describe the bug
I have encountered an error when trying to sample from SD3 transformer component, where line 208 in `attention.py` is asking for two outputs but only one is given:
```
File "/home/p2p/pytorch/mc3/envs/marigold/lib/python3.10/site-packages/diffusers/src/diffusers/models/attention.py", line ... | open | null | false | 15 | [
"bug",
"stale",
"needs-code-example"
] | [] | 2024-12-01T03:59:35Z | 2025-06-19T07:23:49Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tracyyy-zhu | 61,514,951 | MDQ6VXNlcjYxNTE0OTUx | User | false |
huggingface/diffusers | 2,709,432,084 | I_kwDOHa8MBc6hfqcU | 10,070 | https://github.com/huggingface/diffusers/issues/10070 | https://api.github.com/repos/huggingface/diffusers/issues/10070 | Xformers info , memory efficient atttention unavailable | ### Describe the bug
I just started learning Stable Diffuision on Win11. After I installed xformers, I found several memory_efficient_attention string is unavailable. Is it possible to make them available? Thanks for any help.
### Reproduction
xFormers 0.0.28.post3
memory_efficient_attention.ckF: ... | open | null | false | 1 | [
"bug",
"stale"
] | [] | 2024-12-01T16:14:21Z | 2025-01-01T15:03:09Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Stareshine | 160,834,677 | U_kgDOCZYkdQ | User | false |
huggingface/diffusers | 2,709,616,761 | I_kwDOHa8MBc6hgXh5 | 10,071 | https://github.com/huggingface/diffusers/issues/10071 | https://api.github.com/repos/huggingface/diffusers/issues/10071 | [New feature] A Noise Injection Method for Flux | ### New Feature for FluxPipeline
Paper: [Link](https://openreview.net/pdf?id=KzokzKV4JK)
Code : [Link](https://anonymous.4open.science/r/SSNI-F746/README.md)
workflow : [Link](https://www.dropbox.com/scl/fi/hhitjx6lqpqpv8xjx9ikq/Flux-Noise-Injection.json?rlkey=45xnu45j1i5owiwhc7z1hppn1&e=1&dl=0)
This paper introduc... | open | null | false | 10 | [
"wip",
"consider-for-modular-diffusers"
] | [] | 2024-12-01T18:35:39Z | 2025-02-26T16:36:07Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SahilCarterr | 110,806,554 | U_kgDOBprGGg | User | false |
huggingface/diffusers | 2,710,794,006 | I_kwDOHa8MBc6hk28W | 10,074 | https://github.com/huggingface/diffusers/issues/10074 | https://api.github.com/repos/huggingface/diffusers/issues/10074 | how to install diffusers 0.32.0 | FluxFillPipeline Function need =0.32.0 But I don't know how to install it, can anyone help me? Thanks in advance | closed | completed | false | 2 | [] | [] | 2024-12-02T07:05:24Z | 2024-12-02T19:11:34Z | 2024-12-02T09:35:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | babyta | 37,787,795 | MDQ6VXNlcjM3Nzg3Nzk1 | User | false |
huggingface/diffusers | 2,710,824,259 | I_kwDOHa8MBc6hk-VD | 10,075 | https://github.com/huggingface/diffusers/issues/10075 | https://api.github.com/repos/huggingface/diffusers/issues/10075 | flux:different results on different machines. | Hi！
During the debugging phase, I've noticed that training the [FLUX-Controlnet-Inpainting](https://github.com/alimama-creative/FLUX-Controlnet-Inpainting) model yields different results on different machines.
Specifically, on one machine, everything trains fine,
but on another, I'm getting an error that says "... | open | null | false | 3 | [
"stale"
] | [] | 2024-12-02T07:22:00Z | 2025-01-01T15:03:03Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | D222097 | 50,061,868 | MDQ6VXNlcjUwMDYxODY4 | User | false |
huggingface/diffusers | 2,710,923,544 | I_kwDOHa8MBc6hlWkY | 10,076 | https://github.com/huggingface/diffusers/issues/10076 | https://api.github.com/repos/huggingface/diffusers/issues/10076 | Do we have any script covert from hf format to orginal format? | **Is your feature request related to a problem? Please describe.**
scripts/convert_cogvideox_to_diffusers.py
in this script, we can convert cogvideox -> diffusers. Do we have the opposite script?
cc @yiyixuxu
| open | null | false | 1 | [
"good first issue",
"contributions-welcome",
"conversion script"
] | [] | 2024-12-02T07:49:34Z | 2024-12-02T18:22:50Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | foreverpiano | 29,483,524 | MDQ6VXNlcjI5NDgzNTI0 | User | false |
huggingface/diffusers | 2,711,410,316 | I_kwDOHa8MBc6hnNaM | 10,078 | https://github.com/huggingface/diffusers/issues/10078 | https://api.github.com/repos/huggingface/diffusers/issues/10078 | Cache text encoder embeds in pipelines | **Is your feature request related to a problem? Please describe.**
When reusing a prompt text encoder embeds are recomputed, this can be time consuming for something like T5-XXL with offloading or on CPU.
Text encoder embeds are relatively small, so keeping them in memory is feasible.
```python
import torch
... | open | null | false | 3 | [
"community-examples",
"stale"
] | [] | 2024-12-02T10:40:01Z | 2025-01-01T15:02:59Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hlky | 106,811,348 | U_kgDOBl3P1A | User | false |
huggingface/diffusers | 2,711,545,173 | I_kwDOHa8MBc6hnuVV | 10,080 | https://github.com/huggingface/diffusers/issues/10080 | https://api.github.com/repos/huggingface/diffusers/issues/10080 | Proposal to add sigmas option to FluxPipeline | **Is your feature request related to a problem? Please describe.**
Flux's image generation is great, but it seems to have a tendency to remove too much detail.
**Describe the solution you'd like.**
Add the `sigmas` option to `FluxPipeline` to enable adjustment of the degree of noise removal.
**Additional contex... | closed | completed | false | 3 | [] | [] | 2024-12-02T11:14:40Z | 2024-12-02T18:16:49Z | 2024-12-02T18:16:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | John6666cat | 186,692,226 | U_kgDOCyCygg | User | false |
huggingface/diffusers | 2,711,831,082 | I_kwDOHa8MBc6ho0Iq | 10,085 | https://github.com/huggingface/diffusers/issues/10085 | https://api.github.com/repos/huggingface/diffusers/issues/10085 | RuntimeError with LyCORIS, Batch Inference and skip_guidance_layers | ### Describe the bug
A `RuntimeError` occurs when using the following combination:
* SD3
* Batch inference (`num_images_per_prompt > 1`)
* LyCORIS
* `skip_guidance_layers` is set
The error message is: `"RuntimeError: The size of tensor a (2) must match the size of tensor b (4) at non-singleton dimension 0... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-12-02T12:56:30Z | 2024-12-02T21:43:05Z | 2024-12-02T21:43:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | japppie | 131,072,659 | U_kgDOB9ACkw | User | false |
huggingface/diffusers | 1,424,872,662 | I_kwDOHa8MBc5U7dTW | 1,009 | https://github.com/huggingface/diffusers/issues/1009 | https://api.github.com/repos/huggingface/diffusers/issues/1009 | Weight decay bug in textual_inversion.py | ### Describe the bug
I think there's a bug in the [textual_inversion.py ](https://github.com/huggingface/diffusers/blob/main/examples/textual_inversion/textual_inversion.py)script. Specifically, the word embeddings of all the tokens except the newly added token are supposed to be frozen. The way the script does this i... | closed | completed | false | 5 | [
"bug",
"stale"
] | [
"patil-suraj"
] | 2022-10-27T00:57:36Z | 2023-01-13T15:05:01Z | 2023-01-13T15:05:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | scoutsaachi | 9,390,538 | MDQ6VXNlcjkzOTA1Mzg= | User | false |
huggingface/diffusers | 2,713,832,446 | I_kwDOHa8MBc6hwcv- | 10,090 | https://github.com/huggingface/diffusers/issues/10090 | https://api.github.com/repos/huggingface/diffusers/issues/10090 | Face editor onnxruntime error Fail:Loadlibrary faild with onnxruntime_providers_cuda.dll | ### Describe the bug
After I installed the Face Editor in Webui/Extension. the starter show me such fails. Even though, Face editor will still work fine. That confuses me. These will now show when I disable this extensions.
I am a beginner. I have checked everything I could find on the internet. Onnxruntime-GPU 1.20.... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-12-03T02:32:31Z | 2025-01-02T18:50:22Z | 2025-01-02T18:50:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Stareshine | 160,834,677 | U_kgDOCZYkdQ | User | false |
huggingface/diffusers | 2,713,868,101 | I_kwDOHa8MBc6hwldF | 10,091 | https://github.com/huggingface/diffusers/issues/10091 | https://api.github.com/repos/huggingface/diffusers/issues/10091 | CogvideoX VAE decoder consumes significantly more memory in the latest version | ### Discussed in https://github.com/huggingface/diffusers/discussions/10036
<div type='discussions-op-text'>
<sup>Originally posted by **ic-synth** November 27, 2024</sup>
### Describe the bug
The memory consumption for CogVideoX decoder in diffusers 0.31.0 version consumes significantly more memory. To the p... | open | null | false | 6 | [
"bug",
"wip",
"roadmap"
] | [
"yiyixuxu",
"a-r-r-o-w"
] | 2024-12-03T03:02:42Z | 2025-08-10T18:29:54Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,713,961,860 | I_kwDOHa8MBc6hw8WE | 10,093 | https://github.com/huggingface/diffusers/issues/10093 | https://api.github.com/repos/huggingface/diffusers/issues/10093 | [pipelines] add better checking when a wrong model is passed when initializing a pipeline | With this code:
```py
from diffusers import DiffusionPipeline
from transformers import T5EncoderModel
import torch
repo_id = "black-forest-labs/FLUX.1-dev"
text_encoder_2 = T5EncoderModel.from_pretrained(repo_id, subfolder="text_encoder_2")
pipe = DiffusionPipeline.from_pretrained(
repo_id, text_enco... | closed | completed | false | 1 | [
"roadmap"
] | [] | 2024-12-03T04:33:35Z | 2024-12-19T09:24:53Z | 2024-12-19T09:24:53Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,714,263,146 | I_kwDOHa8MBc6hyF5q | 10,097 | https://github.com/huggingface/diffusers/issues/10097 | https://api.github.com/repos/huggingface/diffusers/issues/10097 | Missing Section in Optimization Docs | ### Describe the bug
There is a missing section in the Optimization section of the diffuser docs. Its location being here in the ReadMe:
Web link for the main version: https://huggingface.co/docs/diffusers/main/en/optimi... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-12-03T07:56:20Z | 2024-12-03T18:23:36Z | 2024-12-03T18:23:36Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ParagEkbote | 69,567,729 | MDQ6VXNlcjY5NTY3NzI5 | User | false |
huggingface/diffusers | 2,714,883,731 | I_kwDOHa8MBc6h0daT | 10,100 | https://github.com/huggingface/diffusers/issues/10100 | https://api.github.com/repos/huggingface/diffusers/issues/10100 | Add support for ConsisID | ### Model/Pipeline/Scheduler description
I have found a awesome work ConsisID, a tuning-free DiT-based controllable IPT2V model to keep human-identity consistent in the generated video. Can diffuser support it, so that I can simply use this model, thanks
### Open source status
- [X] The model implementation is... | closed | completed | false | 1 | [
"New pipeline/model"
] | [] | 2024-12-03T12:21:01Z | 2025-06-10T20:55:00Z | 2025-06-10T20:54:59Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 963658029 | 25,529,908 | MDQ6VXNlcjI1NTI5OTA4 | User | false |
huggingface/diffusers | 2,715,881,329 | I_kwDOHa8MBc6h4Q9x | 10,106 | https://github.com/huggingface/diffusers/issues/10106 | https://api.github.com/repos/huggingface/diffusers/issues/10106 | Add HunyuanVideo | ## HunyuanVideo
_We present HunyuanVideo, a novel open-source video foundation model that exhibits performance in video generation that is comparable to, if not superior to, leading closed-source models. HunyuanVideo features a comprehensive framework that integrates several key contributions, including data curatio... | closed | completed | false | 12 | [
"New pipeline/model"
] | [] | 2024-12-03T20:09:15Z | 2024-12-31T06:30:17Z | 2024-12-17T21:39:19Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hlky | 106,811,348 | U_kgDOBl3P1A | User | false |
huggingface/diffusers | 2,716,073,827 | I_kwDOHa8MBc6h4_9j | 10,107 | https://github.com/huggingface/diffusers/issues/10107 | https://api.github.com/repos/huggingface/diffusers/issues/10107 | gradient checkpointing runs during validations in the training examples | ### Describe the bug
The gradient checkpointing is enabled via self.training, but then the log_validations also unnecessarily encounter this codepath.
I found that I have to disable this when running even under `torch.no_grad()`, looked and saw that the official examples do not do this either.
This gives a sub... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2024-12-03T22:10:25Z | 2025-01-03T16:01:36Z | 2025-01-03T16:01:36Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,716,357,956 | I_kwDOHa8MBc6h6FVE | 10,108 | https://github.com/huggingface/diffusers/issues/10108 | https://api.github.com/repos/huggingface/diffusers/issues/10108 | [pipelines] allow models to run with a user-provided dtype map instead of a single dtype | The newer models like Mochi-1 run the text encoder and VAE decoding in FP32 while keeping the denoising process in `torch.bfloat16` autocast.
Currently, it's not possible for our pipelines to run the different models involved as we set a global `torch_dtype` while initializing the pipeline.
We have some pipelin... | closed | completed | false | 1 | [
"enhancement"
] | [] | 2024-12-04T01:43:38Z | 2025-04-02T08:06:50Z | 2025-04-02T08:05:48Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,717,050,507 | I_kwDOHa8MBc6h8uaL | 10,111 | https://github.com/huggingface/diffusers/issues/10111 | https://api.github.com/repos/huggingface/diffusers/issues/10111 | Some bugs in FLUX pipeline | ### Describe the bug
1. missing self.theta in get_1d_rotary_pos_embed:
https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/embeddings.py#L961-L963
2. if prompt_embeds is None, pooled_prompt_embeds will never be computed:
https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-12-04T09:09:54Z | 2024-12-05T13:12:49Z | 2024-12-05T13:12:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chenxiao111222 | 154,797,505 | U_kgDOCToFwQ | User | false |
huggingface/diffusers | 2,717,061,836 | I_kwDOHa8MBc6h8xLM | 10,112 | https://github.com/huggingface/diffusers/issues/10112 | https://api.github.com/repos/huggingface/diffusers/issues/10112 | Detail-Daemon diffusers | **Describe the solution you'd like.**
Detail-Daemon: https://github.com/Jonseed/ComfyUI-Detail-Daemon
How to implement Detail-Daemon in diffusers, as seen in https://github.com/Jonseed/ComfyUI-Detail-Daemon. Will there be a better official component in the future? | open | null | false | 10 | [
"wip",
"consider-for-modular-diffusers"
] | [] | 2024-12-04T09:14:39Z | 2025-01-03T18:01:24Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NicholasCao | 32,389,766 | MDQ6VXNlcjMyMzg5NzY2 | User | false |
huggingface/diffusers | 2,718,451,920 | I_kwDOHa8MBc6iCEjQ | 10,118 | https://github.com/huggingface/diffusers/issues/10118 | https://api.github.com/repos/huggingface/diffusers/issues/10118 | Shall we have a SD3.5 fine tune script under examples text to image folder | SD3.5 is a full model release. The models are perfect for fine-tune effort. Diffusers already has the fine tune script for SDXL under examples/text_to_image folder. Shall we do the same for SD3.5 | closed | completed | false | 2 | [
"stale",
"training",
"contributions-welcome"
] | [] | 2024-12-04T18:02:46Z | 2025-01-06T00:12:04Z | 2025-01-06T00:12:03Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | linjiapro | 4,970,879 | MDQ6VXNlcjQ5NzA4Nzk= | User | false |
huggingface/diffusers | 2,718,637,895 | I_kwDOHa8MBc6iCx9H | 10,120 | https://github.com/huggingface/diffusers/issues/10120 | https://api.github.com/repos/huggingface/diffusers/issues/10120 | memory consumption of dreambooth+SD3 | Hi, I am running dreambooth SD3 with a single A100 GPU, I reduced resolution to 256; but it still need more memory than a single A100 has? I am wondering is this huge memory consumption normal?
```
!python train_dreambooth_sd3.py \
--pretrained_model_name_or_path="stabilityai/stable-diffusion-3-medium-diffusers"... | closed | completed | false | 5 | [
"bug",
"stale",
"training"
] | [] | 2024-12-04T19:39:04Z | 2025-01-27T01:30:18Z | 2025-01-27T01:30:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KolvacS-W | 55,591,358 | MDQ6VXNlcjU1NTkxMzU4 | User | false |
huggingface/diffusers | 2,718,760,825 | I_kwDOHa8MBc6iDP95 | 10,121 | https://github.com/huggingface/diffusers/issues/10121 | https://api.github.com/repos/huggingface/diffusers/issues/10121 | Flux1.-Dev cannot generate high resolution images in diffusers | ### Describe the bug
When I try to generate images with Flux.1-Dev at high resolutions e.g. above 5000, diffusers crashes.
### Reproduction
```python
import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)
pipe.to('cuda... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-12-04T20:50:18Z | 2025-01-07T23:20:01Z | 2025-01-07T23:20:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clintg6 | 7,388,379 | MDQ6VXNlcjczODgzNzk= | User | false |
huggingface/diffusers | 2,719,203,712 | I_kwDOHa8MBc6iE8GA | 10,124 | https://github.com/huggingface/diffusers/issues/10124 | https://api.github.com/repos/huggingface/diffusers/issues/10124 | [models] allow specifying block numbers in `enable_gradient_checkpointing()` to speed up training | As brought up in https://github.com/huggingface/diffusers/pull/9982#issuecomment-2517826764 by @bghira, I think we could support this directly in the `enable_gradient_checkpointing()` method we expose for the models.
Users could specify the block interval they want gradient checkpointing to be applied in and we tak... | closed | completed | false | 3 | [
"stale",
"training",
"performance"
] | [
"a-r-r-o-w"
] | 2024-12-05T03:09:00Z | 2025-01-28T01:21:47Z | 2025-01-28T01:21:47Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,719,814,867 | I_kwDOHa8MBc6iHRTT | 10,128 | https://github.com/huggingface/diffusers/issues/10128 | https://api.github.com/repos/huggingface/diffusers/issues/10128 | Is there any plan to support fastercache? | Expect to support fastercache, https://github.com/Vchitect/FasterCache | closed | completed | false | 4 | [
"wip",
"performance"
] | [] | 2024-12-05T09:11:19Z | 2025-03-21T04:05:06Z | 2025-03-21T04:05:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | songh11 | 75,419,275 | MDQ6VXNlcjc1NDE5Mjc1 | User | false |
huggingface/diffusers | 2,719,881,196 | I_kwDOHa8MBc6iHhfs | 10,129 | https://github.com/huggingface/diffusers/issues/10129 | https://api.github.com/repos/huggingface/diffusers/issues/10129 | Does StableDiffusion3 have an image2image pipeline with ControlNet? | I want to use `ControlNet` with `StableDiffusion3`, providing a prompt, an original image, and a control image as inputs. However, I found that the `StableDiffusion3ControlNetPipeline` only supports prompts and control images as inputs. The `StableDiffusionControlNetImg2ImgPipeline` allows for providing a prompt, an or... | closed | completed | false | 1 | [
"New pipeline/model",
"contributions-welcome"
] | [] | 2024-12-05T09:40:03Z | 2025-01-02T20:02:33Z | 2025-01-02T20:02:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ZHJ19970917 | 73,893,296 | MDQ6VXNlcjczODkzMjk2 | User | false |
huggingface/diffusers | 1,425,003,624 | I_kwDOHa8MBc5U79Ro | 1,013 | https://github.com/huggingface/diffusers/issues/1013 | https://api.github.com/repos/huggingface/diffusers/issues/1013 | 'StableDiffusionMegaPipeline' not working Diffusers 6.0 | ### Describe the bug
```
from diffusers import DiffusionPipeline
self.pipe = DiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
custom_pipeline="stable_diffusion_mega", revision="fp16", torch_dtype=torch.float16, use_auth_token=hf_token)
self.pipe.to("cuda")
self.pipe.enable_attention... | closed | completed | false | 0 | [
"bug"
] | [] | 2022-10-27T04:16:53Z | 2022-10-27T05:44:48Z | 2022-10-27T05:44:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | adhikjoshi | 11,740,719 | MDQ6VXNlcjExNzQwNzE5 | User | false |
huggingface/diffusers | 2,721,931,060 | I_kwDOHa8MBc6iPV80 | 10,137 | https://github.com/huggingface/diffusers/issues/10137 | https://api.github.com/repos/huggingface/diffusers/issues/10137 | Script to merge Flux diffuser into 1 safetensor file | Hi everyone
I found Flux model which is split into 3 parts. So far I can't find any script to merge splitted parts of Flux model into one
Can anyone help me to do it
Thank in advance | closed | completed | false | 7 | [
"good first issue",
"contributions-welcome",
"conversion script"
] | [] | 2024-12-06T02:58:37Z | 2025-03-10T08:01:13Z | 2025-02-24T11:44:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pivtienduc | 180,398,453 | U_kgDOCsCpdQ | User | false |
huggingface/diffusers | 2,722,156,945 | I_kwDOHa8MBc6iQNGR | 10,139 | https://github.com/huggingface/diffusers/issues/10139 | https://api.github.com/repos/huggingface/diffusers/issues/10139 | Unet sample_size not fully considered in pipeline init | ### Describe the bug
#### Background
- I'm trying to train a small unet from scratch
- For this, I used a modified version of `train_text_to_image.py`
- unet model's `sample_size` is not symmetric. I passed `[60, 80]` when instantiating `UNet2DConditionModel`, which is normally an `int` value with `512`
#### Wha... | closed | completed | false | 6 | [
"bug"
] | [] | 2024-12-06T06:12:30Z | 2024-12-26T07:27:51Z | 2024-12-19T22:24:20Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Foundsheep | 92,705,171 | U_kgDOBYaRkw | User | false |
huggingface/diffusers | 1,425,128,896 | I_kwDOHa8MBc5U8b3A | 1,014 | https://github.com/huggingface/diffusers/issues/1014 | https://api.github.com/repos/huggingface/diffusers/issues/1014 | have a small problem after upgrade diffusers and others | hello!
I upgrade my diffuers and use dreamtbooth but It doesn't work like this code
```
> The following values were not passed to `accelerate launch` and had defaults used instead:
> `--num_cpu_threads_per_process` was set to `4` to improve out-of-box performance
> To avoid this warning pass in values fo... | closed | completed | false | 14 | [
"stale"
] | [
"patil-suraj"
] | 2022-10-27T06:55:23Z | 2022-12-28T15:03:35Z | 2022-12-28T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | A-Polyana | 58,426,735 | MDQ6VXNlcjU4NDI2NzM1 | User | false |
huggingface/diffusers | 2,722,531,692 | I_kwDOHa8MBc6iRols | 10,141 | https://github.com/huggingface/diffusers/issues/10141 | https://api.github.com/repos/huggingface/diffusers/issues/10141 | [New Feature] Flux PuLID | ### PuLID: Pure and Lightning ID Customization via Contrastive Alignment
Paper:[Link](https://arxiv.org/abs/2404.16022)
Code:[Link](https://github.com/ToTheBeginning/PuLID)
This has astonishing result with FLUX Pipeline
![pulid_github](https://github.com/user-attachments/assets/7b772cc4-c274-49e3-af2e-b585e93... | open | null | false | 7 | [
"consider-for-modular-diffusers"
] | [] | 2024-12-06T09:33:18Z | 2025-08-26T06:52:13Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SahilCarterr | 110,806,554 | U_kgDOBprGGg | User | false |
huggingface/diffusers | 2,722,629,149 | I_kwDOHa8MBc6iSAYd | 10,142 | https://github.com/huggingface/diffusers/issues/10142 | https://api.github.com/repos/huggingface/diffusers/issues/10142 | Requirements for advanced_diffusion_training incomplete | ### Describe the bug
When following the https://github.com/huggingface/diffusers/blob/main/examples/advanced_diffusion_training/README.md
I also had to install / do those things to make it work:
```
!pip install wandb prodigyopt datasets
!pip install --upgrade peft
```
- login to huggingface
- setup and login... | open | null | false | 6 | [
"bug",
"stale",
"training",
"contributions-welcome"
] | [
"linoytsaban"
] | 2024-12-06T10:21:00Z | 2025-01-30T15:03:25Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | teshi24 | 29,971,981 | MDQ6VXNlcjI5OTcxOTgx | User | false |
huggingface/diffusers | 2,724,396,369 | I_kwDOHa8MBc6iYv1R | 10,144 | https://github.com/huggingface/diffusers/issues/10144 | https://api.github.com/repos/huggingface/diffusers/issues/10144 | Why mochi diffusers video output is worse than mochi official code? | ### Describe the bug
The quality of video is worse.
### Reproduction
Run the code with official prompt
### Logs
_No response_
### System Info
diffusers@main
### Who can help?
@a-r-r-o-w @yiyixuxu | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2024-12-07T05:53:57Z | 2025-01-07T15:38:38Z | 2025-01-07T15:38:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | foreverpiano | 29,483,524 | MDQ6VXNlcjI5NDgzNTI0 | User | false |
huggingface/diffusers | 2,724,542,274 | I_kwDOHa8MBc6iZTdC | 10,145 | https://github.com/huggingface/diffusers/issues/10145 | https://api.github.com/repos/huggingface/diffusers/issues/10145 | Dreambooth Flux with Lora: OOM | ### Describe the bug
It is pretty wired that even after I set lora rank to only 2, training the dreambooth_flux with lora still returns OOM. My GPU is three L40s (44G-48G). And data is only one image.
If this is still OOM, I'm wondering why should lora version exist? The next GPU memory capbility is 80G which can t... | open | null | false | 13 | [
"bug",
"stale",
"training"
] | [] | 2024-12-07T11:11:32Z | 2025-01-06T15:03:09Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | massyzs | 45,664,796 | MDQ6VXNlcjQ1NjY0Nzk2 | User | false |
huggingface/diffusers | 2,724,919,623 | I_kwDOHa8MBc6iavlH | 10,147 | https://github.com/huggingface/diffusers/issues/10147 | https://api.github.com/repos/huggingface/diffusers/issues/10147 | Support MV-Adapter for multi-view generation | ### Model/Pipeline/Scheduler description
[MV-Adapter](https://huanngzh.github.io/MV-Adapter-Page/) is a creative productivity tool that seamlessly transfer text-to-image models to multi-view generators.
Highlights:
- generate 768x768 multi-view images
- work well with personalized models, LCM, ControlNet
- suppo... | open | null | false | 1 | [
"New pipeline/model",
"contributions-welcome"
] | [] | 2024-12-08T02:59:06Z | 2024-12-26T09:56:30Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | huanngzh | 78,398,294 | MDQ6VXNlcjc4Mzk4Mjk0 | User | false |
huggingface/diffusers | 1,425,302,817 | I_kwDOHa8MBc5U9GUh | 1,015 | https://github.com/huggingface/diffusers/issues/1015 | https://api.github.com/repos/huggingface/diffusers/issues/1015 | JAX / Flax: avoid recompilation on params change: guidance scale, size, number of steps | This is required to improve the demo(s), we currently have the sliders disabled.
Some users are also requesting it in discord and the forums. Our demo notebook is very fast on TPU hardware but users can't quickly explore different parameters. | closed | completed | false | 10 | [
"stale"
] | [
"pcuenca"
] | 2022-10-27T09:15:53Z | 2022-12-06T15:03:21Z | 2022-12-06T15:03:21Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 2,726,306,327 | I_kwDOHa8MBc6igCIX | 10,150 | https://github.com/huggingface/diffusers/issues/10150 | https://api.github.com/repos/huggingface/diffusers/issues/10150 | SDXL Pipelines slow loading and inference | ### Describe the bug
Cached model + lora takes >40 seconds to load and >8 minutes to run a simple image to image task at 1024x1024
### Reproduction
```
import os
import torch
from PIL import Image
from diffusers import StableDiffusionXLImg2ImgPipeline
from diffusers.schedulers import DPMSolverMultistepSchedule... | closed | completed | false | 6 | [
"bug"
] | [] | 2024-12-09T08:08:29Z | 2024-12-09T09:36:22Z | 2024-12-09T09:36:21Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | okaris | 1,448,702 | MDQ6VXNlcjE0NDg3MDI= | User | false |
huggingface/diffusers | 2,726,339,739 | I_kwDOHa8MBc6igKSb | 10,152 | https://github.com/huggingface/diffusers/issues/10152 | https://api.github.com/repos/huggingface/diffusers/issues/10152 | Diffusers Roadmap | The roadmap board shows work that is being prioritized for the upcoming Diffusers release as well as issues that are being considered for future releases.
Diffusers Public Roadmap
https://github.com/orgs/huggingface/projects/61/views/1 | open | null | false | 3 | [
"stale"
] | [] | 2024-12-09T08:24:37Z | 2025-02-02T15:03:10Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DN6 | 7,529,846 | MDQ6VXNlcjc1Mjk4NDY= | User | false |
huggingface/diffusers | 2,726,536,504 | I_kwDOHa8MBc6ig6U4 | 10,153 | https://github.com/huggingface/diffusers/issues/10153 | https://api.github.com/repos/huggingface/diffusers/issues/10153 | In the StableDiffusion3Img2ImgPipeline, the height dimension of latents is odd, which may cause a mismatch with the height dimension of noise_pred. | ### Describe the bug
In the `StableDiffusion3Img2ImgPipeline`, the height dimension of latents is odd, which may cause a mismatch with the height dimension of noise_pred.
During the process of using the `StableDiffusion3Img2ImgPipeline`, I input an image of shape (1, 3, 856, 1088). After passing through the `self... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-12-09T09:52:22Z | 2024-12-12T22:32:20Z | 2024-12-12T22:32:20Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ZHJ19970917 | 73,893,296 | MDQ6VXNlcjczODkzMjk2 | User | false |
huggingface/diffusers | 2,726,802,447 | I_kwDOHa8MBc6ih7QP | 10,154 | https://github.com/huggingface/diffusers/issues/10154 | https://api.github.com/repos/huggingface/diffusers/issues/10154 | When using load_lora_weights, it prompts that object has no attribute | ### Describe the bug
When following the tutorial in version V0.31.0, specifically the section "Load LoRAs for inference," using the original code :
```
from diffusers import DiffusionPipeline
import torch
pipe_id = "stabilityai/stable-diffusion-xl-base-1.0"
pipe = DiffusionPipeline.from_pretrained(pipe_id, torch_... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-12-09T11:43:42Z | 2024-12-10T08:09:39Z | 2024-12-10T08:09:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Purhythm | 78,393,131 | MDQ6VXNlcjc4MzkzMTMx | User | false |
huggingface/diffusers | 2,727,542,669 | I_kwDOHa8MBc6ikv-N | 10,158 | https://github.com/huggingface/diffusers/issues/10158 | https://api.github.com/repos/huggingface/diffusers/issues/10158 | Post processing performance can be improved | ## Problem
Images generated in batches pay a performance penalty in the post-processing step of the diffusion pipeline.
A lot of calls to image_processor.denormalize are made instead of batching the computation.
### Suggested Improvements
#### Using multiplication instead of division
This is a freebie, n... | closed | completed | false | 1 | [] | [] | 2024-12-09T15:59:54Z | 2024-12-10T18:07:27Z | 2024-12-10T18:07:27Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | soof-golan | 83,900,570 | MDQ6VXNlcjgzOTAwNTcw | User | false |
huggingface/diffusers | 2,727,559,307 | I_kwDOHa8MBc6ik0CL | 10,160 | https://github.com/huggingface/diffusers/issues/10160 | https://api.github.com/repos/huggingface/diffusers/issues/10160 | `StableAudioDiTModel` throws `[rank0]: RuntimeError: DDP...` on `accelerator.prepare` when training on 4 GPUs | ### Describe the bug
When trying to train on 4 GPUs, accelerator.prepare fails with:
```
/usr/local/lib/python3.10/dist-packages/transformers/utils/hub.py:128: FutureWarning: Using `TRANSFORMERS_CACHE` is deprecated and will be removed in v5 of Transformers. Use `HF_HOME` instead.
warnings.warn(
/usr/local/l... | open | null | false | 14 | [
"bug",
"training"
] | [] | 2024-12-09T16:05:13Z | 2025-07-05T21:33:16Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mmxgn | 349,290 | MDQ6VXNlcjM0OTI5MA== | User | false |
huggingface/diffusers | 2,727,735,363 | I_kwDOHa8MBc6ilfBD | 10,162 | https://github.com/huggingface/diffusers/issues/10162 | https://api.github.com/repos/huggingface/diffusers/issues/10162 | Move UFOGen Pipeline and Scheduler to Research Projects | **Is your feature request related to a problem? Please describe.**
After reviewing #6133 and a bit of searching has led me to the implementation of the UFOGen paper by the co-authors. By reading the repo ReadME and it seems there is no model checkpoints provided by the authors, only the training code. Does it make sen... | closed | not_planned | false | 2 | [
"stale"
] | [] | 2024-12-09T17:08:39Z | 2025-01-14T19:44:41Z | 2025-01-14T19:44:41Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ParagEkbote | 69,567,729 | MDQ6VXNlcjY5NTY3NzI5 | User | false |
huggingface/diffusers | 2,728,798,065 | I_kwDOHa8MBc6ipidx | 10,165 | https://github.com/huggingface/diffusers/issues/10165 | https://api.github.com/repos/huggingface/diffusers/issues/10165 | LoRA - Seed are not preserved on changing LoRA scale | ### Describe the bug
Hello everyone, I have a question, I noticed that when I use LoRA in diffusers the seeds are not preserved in all generations while I change the LoRA scale. When I do the same thing in auto 1111 this does not happen, to be honest it happens but very little variation
## Diffusers
|**Without Lor... | closed | completed | false | 9 | [
"bug"
] | [] | 2024-12-10T02:39:09Z | 2024-12-12T23:19:05Z | 2024-12-10T18:19:23Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elismasilva | 40,075,615 | MDQ6VXNlcjQwMDc1NjE1 | User | false |
huggingface/diffusers | 2,729,669,917 | I_kwDOHa8MBc6is3Ud | 10,169 | https://github.com/huggingface/diffusers/issues/10169 | https://api.github.com/repos/huggingface/diffusers/issues/10169 | an error that config.json is not found. | ### Describe the bug
Whether it is local download and upload or download model from Hugging face official website, after running on the jupter notebook, it shows an error that config.json is not found.
After checking, the locally downloaded file is exactly the same as the official website, but it still reports an e... | closed | completed | false | 6 | [
"bug",
"needs-code-example"
] | [] | 2024-12-10T10:17:27Z | 2024-12-11T13:37:56Z | 2024-12-11T13:37:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zarato2122 | 167,994,972 | U_kgDOCgNmXA | User | false |
huggingface/diffusers | 2,730,275,161 | I_kwDOHa8MBc6ivLFZ | 10,172 | https://github.com/huggingface/diffusers/issues/10172 | https://api.github.com/repos/huggingface/diffusers/issues/10172 | Raise an error when `len(gligen_images )` is not equal to `len(gligen_phrases)` in `StableDiffusionGLIGENTextImagePipeline` | To whom it may concern,
I found that when using `StableDiffusionGLIGENTextImagePipeline`, there is no error raised when `len(gligen_images )` is not equal to `len(gligen_phrases)`. And when I dig into the source code, it seems that these two features are zipped together in a for loop during the preprocessing. I gues... | closed | completed | false | 1 | [] | [] | 2024-12-10T14:25:48Z | 2024-12-11T08:59:44Z | 2024-12-11T08:59:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | abcdefg133hi | 33,452,670 | MDQ6VXNlcjMzNDUyNjcw | User | false |
huggingface/diffusers | 2,730,910,255 | I_kwDOHa8MBc6ixmIv | 10,178 | https://github.com/huggingface/diffusers/issues/10178 | https://api.github.com/repos/huggingface/diffusers/issues/10178 | unet/unet_3d_blocks/.py - dimension size mismatche | ### Describe the bug
I encountered a RuntimeError while running the pipeline in the [ControlNext-SVD-v2](https://github.com/dvlab-research/ControlNeXt/tree/main/ControlNeXt-SVD-v2) project. When I set the resolution of the video except for certain numbers, I got the following error. The error is related to tensor size... | closed | completed | false | 10 | [
"bug"
] | [] | 2024-12-10T18:41:38Z | 2024-12-19T11:32:11Z | 2024-12-19T11:32:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | HoseinRanjbar | 102,860,600 | U_kgDOBiGHOA | User | false |
huggingface/diffusers | 2,731,243,766 | I_kwDOHa8MBc6iy3j2 | 10,180 | https://github.com/huggingface/diffusers/issues/10180 | https://api.github.com/repos/huggingface/diffusers/issues/10180 | Can't load multiple loras when using Flux Control LoRA | ### Describe the bug
I was trying out the FluxControlPipeline with the Control LoRA introduced in #9999 , but had issues loading in multiple loras.
For example, if I load the depth lora first and then the 8-step lora, it errors on the 8-step lora, and if I load the 8-step lora first and then the depth lora, it err... | closed | completed | false | 11 | [
"bug",
"help wanted",
"lora"
] | [] | 2024-12-10T21:40:24Z | 2024-12-20T09:00:33Z | 2024-12-20T09:00:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jonathanyin12 | 26,175,818 | MDQ6VXNlcjI2MTc1ODE4 | User | false |
huggingface/diffusers | 2,732,213,723 | I_kwDOHa8MBc6i2kXb | 10,186 | https://github.com/huggingface/diffusers/issues/10186 | https://api.github.com/repos/huggingface/diffusers/issues/10186 | ImportError: cannot import name 'FluxFillPipeline' from 'diffusers' (/usr/local/lib/python3.10/dist-packages/diffusers/__init__.py) | ### Describe the bug
ImportError: cannot import name 'FluxFillPipeline' from 'diffusers' (/usr/local/lib/python3.10/dist-packages/diffusers/__init__.py)
### Reproduction
```
import torch
from diffusers import FluxFillPipeline
from diffusers.utils import load_image
image = load_image("https://huggingface.co/d... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-12-11T08:21:24Z | 2025-01-02T20:17:34Z | 2025-01-02T20:17:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Arslan-Mehmood1 | 51,626,734 | MDQ6VXNlcjUxNjI2NzM0 | User | false |
huggingface/diffusers | 2,732,559,043 | I_kwDOHa8MBc6i34rD | 10,190 | https://github.com/huggingface/diffusers/issues/10190 | https://api.github.com/repos/huggingface/diffusers/issues/10190 | How to use fluxfill to repalce background？ | I want to use fluxfill to change the background, but I find that the prompt words are almost useless, and the output image is more like the original image.
I have tested multiple guidance_scale parameters, but found that the resulting image is more related to the original image, and less related to the prompt word. | closed | completed | false | 8 | [] | [] | 2024-12-11T10:48:27Z | 2025-05-23T12:12:28Z | 2025-01-10T20:14:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | babyta | 37,787,795 | MDQ6VXNlcjM3Nzg3Nzk1 | User | false |
huggingface/diffusers | 2,733,091,101 | I_kwDOHa8MBc6i56kd | 10,191 | https://github.com/huggingface/diffusers/issues/10191 | https://api.github.com/repos/huggingface/diffusers/issues/10191 | Unable to specify the GPU device for training; it is set to GPU2 but GPU0 is being used. | ### Describe the bug
I have already set the GPU for training to GPU2 in the program using the statement os.environ["CUDA_VISIBLE_DEVICES"] = "2". However, after starting the training, an error message indicates that GPU0 is being used, and the memory of my GPU0 is already insufficient.
Traceback (most recent call l... | closed | completed | false | 13 | [
"bug"
] | [] | 2024-12-11T14:12:14Z | 2024-12-14T02:23:46Z | 2024-12-14T02:23:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zarato2122 | 167,994,972 | U_kgDOCgNmXA | User | false |
huggingface/diffusers | 2,733,764,450 | I_kwDOHa8MBc6i8e9i | 10,194 | https://github.com/huggingface/diffusers/issues/10194 | https://api.github.com/repos/huggingface/diffusers/issues/10194 | Full support for Flux attention masking | **Is your feature request related to a problem? Please describe.**
SimpleTuner/Kohya allow T5 attention masked training, however this is not currently supported natively in diffusers
**Describe the solution you'd like.**
Already implemented and used in Simpletuner and Kohya: https://github.com/bghira/SimpleTuner/b... | open | null | false | 3 | [
"stale"
] | [] | 2024-12-11T19:07:38Z | 2025-05-10T14:43:46Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | squewel | 97,603,184 | U_kgDOBdFOcA | User | false |
huggingface/diffusers | 2,735,345,535 | I_kwDOHa8MBc6jCg9_ | 10,196 | https://github.com/huggingface/diffusers/issues/10196 | https://api.github.com/repos/huggingface/diffusers/issues/10196 | How to finetune Flux-dev full params, 80G OOM ... | I am using the [train_dreambooth_flux](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_flux.py) script to fine-tune the `flux-dev` model with full parameters using DeepSpeed Stage 2. However, I am still encountering out-of-memory issues on an 80GB GPU. Are there any solutions ava... | open | null | false | 21 | [
"training"
] | [
"linoytsaban"
] | 2024-12-12T09:24:18Z | 2025-08-20T13:19:20Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | huangjun12 | 22,365,664 | MDQ6VXNlcjIyMzY1NjY0 | User | false |
huggingface/diffusers | 2,735,512,109 | I_kwDOHa8MBc6jDJot | 10,198 | https://github.com/huggingface/diffusers/issues/10198 | https://api.github.com/repos/huggingface/diffusers/issues/10198 | Flux.1-dev weight mismatch | ### Describe the bug
From [black forest code](https://github.com/black-forest-labs/flux/blob/805da8571a0b49b6d4043950bd266a65328c243b/src/flux/modules/layers.py#L247) and [diffusers code](https://github.com/huggingface/diffusers/blob/8170dc368d278ec40d27bf04f58bff140cebd99e/src/diffusers/models/transformers/transforme... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-12-12T10:35:33Z | 2024-12-12T11:02:35Z | 2024-12-12T11:02:35Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gameofdimension | 32,255,912 | MDQ6VXNlcjMyMjU1OTEy | User | false |
huggingface/diffusers | 2,736,508,121 | I_kwDOHa8MBc6jG8zZ | 10,202 | https://github.com/huggingface/diffusers/issues/10202 | https://api.github.com/repos/huggingface/diffusers/issues/10202 | Flux Control Lora not unloaded correctly | ### Describe the bug
Hi,
There is a bug while switching pipeline from Flux dev when lora control has been loaded:
### Reproduction
```python
import torch
from controlnet_aux import CannyDetector
from diffusers import FluxControlPipeline
from diffusers.utils import load_image
from diffusers import FluxI... | closed | completed | false | 14 | [
"bug",
"lora"
] | [] | 2024-12-12T17:32:58Z | 2025-01-08T09:46:20Z | 2025-01-08T09:46:20Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,736,855,782 | I_kwDOHa8MBc6jIRrm | 10,204 | https://github.com/huggingface/diffusers/issues/10204 | https://api.github.com/repos/huggingface/diffusers/issues/10204 | FluxFillPipeline quality deteriorates over multiple iterations of inpainting | ### Describe the bug
When the FluxFillPipeline is run multiple times on the same image, the image quality deteriorates in each iteration.
### Reproduction
```python
import torch
from diffusers import FluxFillPipeline
from diffusers.utils import load_image
pipe = FluxFillPipeline.from_pretrained("black-... | open | null | false | 9 | [
"bug",
"stale"
] | [] | 2024-12-12T20:49:49Z | 2025-09-03T10:39:40Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | svenschultze | 4,834,610 | MDQ6VXNlcjQ4MzQ2MTA= | User | false |
huggingface/diffusers | 2,737,965,816 | I_kwDOHa8MBc6jMgr4 | 10,207 | https://github.com/huggingface/diffusers/issues/10207 | https://api.github.com/repos/huggingface/diffusers/issues/10207 | KolorsPipeline does not support from_single_file | from diffusers import KolorsPipeline
KolorsPipeline.from_single_file("models/kolrs-8steps.safetensors")
How does KolorsPipeline load a single file model? | open | null | false | 3 | [
"stale",
"single_file"
] | [] | 2024-12-13T09:44:46Z | 2025-01-12T15:02:46Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Thekey756 | 75,780,148 | MDQ6VXNlcjc1NzgwMTQ4 | User | false |
huggingface/diffusers | 2,737,987,212 | I_kwDOHa8MBc6jMl6M | 10,209 | https://github.com/huggingface/diffusers/issues/10209 | https://api.github.com/repos/huggingface/diffusers/issues/10209 | cogvideox-1.5-5b微调报错 | ### Describe the bug
你好，请问这个是什么问题？
### Reproduction
einops.EinopsError: Error while processing rearrange-reduction pattern "b (t o h p w q) c -> b (t h w) (c o p q)".
Input tensor shape: torch.Size([2, 1036800, 49]). Addi... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-12-13T09:55:46Z | 2024-12-16T03:51:36Z | 2024-12-16T03:51:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | linwenzhao1 | 51,796,215 | MDQ6VXNlcjUxNzk2MjE1 | User | false |
huggingface/diffusers | 1,425,531,773 | I_kwDOHa8MBc5U9-N9 | 1,021 | https://github.com/huggingface/diffusers/issues/1021 | https://api.github.com/repos/huggingface/diffusers/issues/1021 | (Re)implement model cards for the training examples | #748 didn't account for `diffusers.hub_utils.create_model_card()` (called by `hub_utils.push_to_hub()` which was replaced by `repo.push_to_hub()`) that was responsible for creating model cards like https://huggingface.co/anton-l/ddpm-ema-flowers-64 , so now they come out empty.
But since the model card generation lo... | closed | completed | false | 2 | [
"enhancement"
] | [
"anton-l",
"patil-suraj"
] | 2022-10-27T12:10:08Z | 2024-11-21T11:30:44Z | 2024-11-21T11:30:43Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anton-l | 26,864,830 | MDQ6VXNlcjI2ODY0ODMw | User | false |
huggingface/diffusers | 2,738,057,051 | I_kwDOHa8MBc6jM29b | 10,210 | https://github.com/huggingface/diffusers/issues/10210 | https://api.github.com/repos/huggingface/diffusers/issues/10210 | FluxImg2ImgPipeline produces output almost identical to the input image | ### Describe the bug
When providing an input image and a prompt, the resulting output image is nearly identical to the input image, regardless of how I adjust the parameters. The differences are so minor that they are almost imperceptible.
Even after setting the strength parameter to a very high value, the output... | closed | completed | false | 16 | [
"bug"
] | [] | 2024-12-13T10:28:38Z | 2025-02-28T05:39:36Z | 2025-02-27T21:33:00Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CyberVy | 72,680,847 | MDQ6VXNlcjcyNjgwODQ3 | User | false |
huggingface/diffusers | 2,739,618,480 | I_kwDOHa8MBc6jS0Kw | 10,223 | https://github.com/huggingface/diffusers/issues/10223 | https://api.github.com/repos/huggingface/diffusers/issues/10223 | Where should I obtain the lora-sdxl-dreambooth-id in Inference | ### Describe the bug
I tried to upload the download link from the README file generated during training, but an error indicated it was incorrect. Where should I obtain the lora-id for Inference?
### Reproduction
README.md:
---
base_model: /data/ziqiang/czc/diffusers/examples/dreambooth/model
library_name: diffuse... | open | null | false | 5 | [
"bug",
"stale"
] | [] | 2024-12-14T06:34:56Z | 2025-02-07T15:03:24Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zarato2122 | 167,994,972 | U_kgDOCgNmXA | User | false |
huggingface/diffusers | 2,740,688,048 | I_kwDOHa8MBc6jW5Sw | 10,226 | https://github.com/huggingface/diffusers/issues/10226 | https://api.github.com/repos/huggingface/diffusers/issues/10226 | Dreambooth Flux with Lora: Tensor Size Mismatch | ### Describe the bug
I'm fine-tuning a model with Flux using LoRA and mixed precision (fp16). Initially, I ran into an out-of-memory error, which I solved by upgrading to a larger GPU. Now, during the forward pass, I'm encountering a new issue in the apply_rotary_emb function, where two tensors have mismatched sizes... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-12-15T15:07:06Z | 2024-12-16T01:18:20Z | 2024-12-16T01:18:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dyedd | 38,277,217 | MDQ6VXNlcjM4Mjc3MjE3 | User | false |
huggingface/diffusers | 2,740,727,302 | I_kwDOHa8MBc6jXC4G | 10,227 | https://github.com/huggingface/diffusers/issues/10227 | https://api.github.com/repos/huggingface/diffusers/issues/10227 | NotImplementedError: Only LoRAs with input/output features higher than the current module's input/output features are currently supported | ### Describe the bug
Unable to merge LoRAs with input/output features higher than the current module's input/output features. See the example and logs below.
### Reproduction
```python
from diffusers import FluxFillPipeline, FluxTransformer2DModel
transformer = FluxTransformer2DModel.from_pretrained(
... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-12-15T15:54:23Z | 2024-12-17T10:53:34Z | 2024-12-17T10:53:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JakobLS | 32,125,321 | MDQ6VXNlcjMyMTI1MzIx | User | false |
huggingface/diffusers | 2,741,202,172 | I_kwDOHa8MBc6jY2z8 | 10,231 | https://github.com/huggingface/diffusers/issues/10231 | https://api.github.com/repos/huggingface/diffusers/issues/10231 | ImportError when running diffusers tutorial code in Ubuntu 20.04 | ### Describe the bug
When I try to run:
`from diffusers import DDPMPipeline
ddpm = DDPMPipeline.from_pretrained("google/ddpm-cat-256", use_safetensors=True).to("cuda")
image = ddpm(num_inference_steps=25).images[0]
image`
> (habitat) /home/zhangyiqing/IROS2025/habitat-lab $ /home/zhangyiqing/miniconda3/envs/h... | open | null | false | 6 | [
"bug",
"stale"
] | [] | 2024-12-16T03:06:31Z | 2025-02-21T15:03:44Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Benson722 | 55,974,171 | MDQ6VXNlcjU1OTc0MTcx | User | false |
huggingface/diffusers | 2,741,379,838 | I_kwDOHa8MBc6jZiL- | 10,233 | https://github.com/huggingface/diffusers/issues/10233 | https://api.github.com/repos/huggingface/diffusers/issues/10233 | Unstable flux transformer prediction | ### Describe the bug
When feeding the same input (same noise, same prompt embed, same guidance scale and same timestep)to the flux.transformer, i got different output. I tried to fix the seed but nothing changed.
### Reproduction
import torch
import math
from diffusers import FluxPipeline
fr... | open | null | false | 9 | [
"bug",
"stale"
] | [] | 2024-12-16T05:35:00Z | 2025-02-20T15:04:21Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YihanHu-2022 | 58,056,486 | MDQ6VXNlcjU4MDU2NDg2 | User | false |
huggingface/diffusers | 2,742,568,954 | I_kwDOHa8MBc6jeEf6 | 10,239 | https://github.com/huggingface/diffusers/issues/10239 | https://api.github.com/repos/huggingface/diffusers/issues/10239 | HunyuanVideoPipeline not support enable_sequential_cpu_offload()？ | ### Describe the bug
when use **pipe.enable_sequential_cpu_offload()** for purpose of low vram.
but get error below and got the black video
```
diffusers\image_processor.py:147: RuntimeWarning: invalid value encountered in cast
images = (images * 255).round().astype("uint8")
```
_the similar issue：htt... | closed | completed | false | 13 | [
"bug"
] | [] | 2024-12-16T14:39:20Z | 2024-12-18T14:08:58Z | 2024-12-17T21:52:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyun0071 | 35,762,050 | MDQ6VXNlcjM1NzYyMDUw | User | false |
huggingface/diffusers | 2,742,756,779 | I_kwDOHa8MBc6jeyWr | 10,241 | https://github.com/huggingface/diffusers/issues/10241 | https://api.github.com/repos/huggingface/diffusers/issues/10241 | Sana issues | ### Describe the bug
Following `SanaPAGPipeline` implementation in #9982,
i cannot get decent output in more than 1% of runs at best.
- most of runs result in what appears to be image with a lot of residual noise and then dc-ae decoder makes it look like sketch-like image with many circular artifacts (see first ... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-12-16T15:51:49Z | 2024-12-17T13:53:06Z | 2024-12-17T08:13:25Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vladmandic | 57,876,960 | MDQ6VXNlcjU3ODc2OTYw | User | false |
huggingface/diffusers | 2,743,389,574 | I_kwDOHa8MBc6jhM2G | 10,247 | https://github.com/huggingface/diffusers/issues/10247 | https://api.github.com/repos/huggingface/diffusers/issues/10247 | Downsample in the MidResTemporalBlock1D | ### Describe the bug
In decoding/model/autoencoders/unets/unet_1d_blocks.py:212-222
```py
def forward(self, hidden_states: torch.Tensor, temb: torch.Tensor) -> torch.Tensor:
hidden_states = self.resnets[0](hidden_states, temb)
for resnet in self.resnets[1:]:
hidden_states = resnet(... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-12-16T21:00:17Z | 2024-12-16T23:25:17Z | 2024-12-16T23:25:17Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | holmosaint | 32,117,229 | MDQ6VXNlcjMyMTE3MjI5 | User | false |
huggingface/diffusers | 2,744,054,090 | I_kwDOHa8MBc6jjvFK | 10,257 | https://github.com/huggingface/diffusers/issues/10257 | https://api.github.com/repos/huggingface/diffusers/issues/10257 | Let's make `UNet2DConditionModel` take in not only class labels, but also continuous labels | The current `UNet2DConditoinModel` can take in class labels as described [here](https://huggingface.co/docs/diffusers/v0.31.0/en/api/models/unet2d-cond#diffusers.UNet2DConditionModel.num_class_embeds).
However, in my case and also some other people are using it with **continuous labels** and in that case there needs... | open | null | false | 1 | [
"stale"
] | [] | 2024-12-17T06:34:36Z | 2025-01-16T15:03:05Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Foundsheep | 92,705,171 | U_kgDOBYaRkw | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.