repo stringclasses 1
value | github_id int64 1.27B 4.95B | github_node_id stringlengths 18 24 | number int64 8 14.3k | html_url stringlengths 49 53 | api_url stringlengths 59 63 | title stringlengths 1 402 | body stringlengths 1 62.9k ⌀ | state stringclasses 2
values | state_reason stringclasses 4
values | locked bool 2
classes | comments_count int64 0 99 | labels listlengths 0 5 | assignees listlengths 0 5 | created_at stringdate 2022-06-09 16:28:35 2026-07-22 10:32:58 | updated_at stringdate 2022-06-12 22:18:01 2026-07-22 10:32:58 | closed_at stringdate 2022-06-12 22:18:01 2026-07-21 13:17:28 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 149
values | extracted_at stringdate 2026-04-07 13:34:13 2026-07-22 10:35:16 | author_login stringlengths 3 28 | author_id int64 1.54k 296M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 1,280,876,848 | I_kwDOHa8MBc5MWKEw | 31 | https://github.com/huggingface/diffusers/issues/31 | https://api.github.com/repos/huggingface/diffusers/issues/31 | How to use multiple GPUs? | The example script shows how to use torch distributed to launch across multiple machines.
What about using multiple GPUs? Does huggingface automatically use up all GPUs on the current machine? How do we control how much it uses? | closed | completed | false | 6 | [] | [] | 2022-06-22T20:48:59Z | 2023-02-23T14:44:55Z | 2022-06-23T18:50:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | richardrl | 5,181,990 | MDQ6VXNlcjUxODE5OTA= | User | false |
huggingface/diffusers | 1,667,482,945 | I_kwDOHa8MBc5jY8VB | 3,100 | https://github.com/huggingface/diffusers/issues/3100 | https://api.github.com/repos/huggingface/diffusers/issues/3100 | `set_attn_processor` to AttnProcessor2_0 on CPU. | Currently, setting the Attention Processor to `AttnProcessor2_0` doesn't use the torch.nn.functional.scaled_dot_product_attention op.
I tried running the `torch.nn.functional.scaled_dot_product_attention` on the CPU for fp32 tensor and it worked.
```
pipe = StableDiffusionPipeline.from_pretrained("stabilityai/s... | closed | completed | false | 8 | [
"stale"
] | [] | 2023-04-14T04:15:30Z | 2023-07-01T15:03:16Z | 2023-07-01T15:03:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pashu123 | 16,246,821 | MDQ6VXNlcjE2MjQ2ODIx | User | false |
huggingface/diffusers | 1,667,517,035 | I_kwDOHa8MBc5jZEpr | 3,101 | https://github.com/huggingface/diffusers/issues/3101 | https://api.github.com/repos/huggingface/diffusers/issues/3101 | Training example of controlNet yield error | ### Describe the bug
I try training controlnet in my dataset "https://huggingface.co/datasets/svjack/diffusiondb_100_canny_zh"
with small gpu memory config as following
### Reproduction
```bash
export MODEL_DIR="IDEA-CCNL/Taiyi-Stable-Diffusion-1B-Chinese-v0.1"
export OUTPUT_DIR="TSD_save"
accelerate launch tr... | closed | completed | false | 8 | [
"bug",
"stale"
] | [
"williamberman"
] | 2023-04-14T05:07:32Z | 2023-05-22T15:03:08Z | 2023-05-22T15:03:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | svjack | 27,874,014 | MDQ6VXNlcjI3ODc0MDE0 | User | false |
huggingface/diffusers | 1,667,553,031 | I_kwDOHa8MBc5jZNcH | 3,102 | https://github.com/huggingface/diffusers/issues/3102 | https://api.github.com/repos/huggingface/diffusers/issues/3102 | Diffusers Dreambooth training does want to load safetensors during verification image generation | Hello,
im running the diffusors dreambooth script, and when it does try to do verification images it does try to load unet\diffusion_pytorch_model.safetensors and fails as this file is not there. I got unet\diffusion_pytorch_model.bin in that folder.
I do not know or understand why it would not try to .bin file but i... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-04-14T05:53:24Z | 2023-06-12T15:03:06Z | 2023-06-12T15:03:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | osi1880vr | 87,379,616 | MDQ6VXNlcjg3Mzc5NjE2 | User | false |
huggingface/diffusers | 1,667,564,990 | I_kwDOHa8MBc5jZQW- | 3,103 | https://github.com/huggingface/diffusers/issues/3103 | https://api.github.com/repos/huggingface/diffusers/issues/3103 | Inpainting height and width | ```
import PIL
import requests
import torch
from io import BytesIO
from diffusers import StableDiffusionInpaintPipeline
def download_image(url):
response = requests.get(url)
return PIL.Image.open(BytesIO(response.content)).convert("RGB")
img_url = "https://raw.githubusercontent.com/CompVis/la... | closed | completed | false | 12 | [] | [] | 2023-04-14T06:04:25Z | 2023-05-18T02:23:33Z | 2023-05-18T02:23:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Amazingldl | 54,729,177 | MDQ6VXNlcjU0NzI5MTc3 | User | false |
huggingface/diffusers | 1,667,595,736 | I_kwDOHa8MBc5jZX3Y | 3,104 | https://github.com/huggingface/diffusers/issues/3104 | https://api.github.com/repos/huggingface/diffusers/issues/3104 | controlnet inpaint with multiple conditions | I tried to use multiple conditions on controlnet inpaint (https://github.com/timegate/diffusers, https://github.com/huggingface/diffusers/commit/aab478100fce9fea6103f192a21065df11ffa0ed),
but it (seg + depth) showed worse results than just one condition (either seg or depth)
Is there anybody working on this? | closed | completed | false | 5 | [] | [] | 2023-04-14T06:28:54Z | 2025-03-06T03:35:50Z | 2023-04-29T09:14:15Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | timegate | 16,134,605 | MDQ6VXNlcjE2MTM0NjA1 | User | false |
huggingface/diffusers | 1,668,826,040 | I_kwDOHa8MBc5jeEO4 | 3,109 | https://github.com/huggingface/diffusers/issues/3109 | https://api.github.com/repos/huggingface/diffusers/issues/3109 | Generate small images with StableDiffusion, 128x128 possibly, and leave it to me to upscale later. | **Is your feature request related to a problem? Please describe.**
The images are often too large to run quickly. And the memory profile too large.
**Describe the solution you'd like**
I would like to be able to generate an image that is 128x128, and decide later if I want to upscale.
**Describe alternatives yo... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-14T19:15:14Z | 2024-02-19T08:03:33Z | 2023-05-23T15:03:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MatthewWaller | 5,520,521 | MDQ6VXNlcjU1MjA1MjE= | User | false |
huggingface/diffusers | 1,358,473,180 | I_kwDOHa8MBc5Q-Kfc | 311 | https://github.com/huggingface/diffusers/issues/311 | https://api.github.com/repos/huggingface/diffusers/issues/311 | [Community Discussion] Scheduler design | [This](https://github.com/patil-suraj/stable-diffusion-jax/pull/8) is a very nice PR by @pcuenca showing what changes need to be done to the PNDM/PMLS scheduler to make it work with JAX/XLA - it's actually more then anticipated and shows that the scheduler now substantially differs from the original implementation that... | closed | completed | false | 9 | [
"stale"
] | [] | 2022-09-01T08:40:56Z | 2022-11-02T15:04:02Z | 2022-11-02T15:04:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,669,100,751 | I_kwDOHa8MBc5jfHTP | 3,111 | https://github.com/huggingface/diffusers/issues/3111 | https://api.github.com/repos/huggingface/diffusers/issues/3111 | Custom encoder for conditioning instead of text-based encoder. | **Is your feature request related to a problem? Please describe.**
Is there anyway that we can load custom encoder instead of text encoder?
**Describe the solution you'd like**
I guess essentially, we can potentially have an abstract class that people can build a custom encoder on top of that, for example, class e... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-04-15T01:46:28Z | 2023-09-24T23:31:14Z | 2023-05-23T15:03:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nanlliu | 45,443,761 | MDQ6VXNlcjQ1NDQzNzYx | User | false |
huggingface/diffusers | 1,669,127,659 | I_kwDOHa8MBc5jfN3r | 3,112 | https://github.com/huggingface/diffusers/issues/3112 | https://api.github.com/repos/huggingface/diffusers/issues/3112 | Multi controlnet inpainting pipeline | ### Model/Pipeline/Scheduler description
@patrickvonplaten @sayakpaul Given that controlnet v1.1 has released an inpainting model, is it possible to use. multi controlnet with the inpainting model? The @haofanwang directory, which is the current goto repo for inpainting pipeline, doesn't work. And neither does the co... | closed | completed | false | 3 | [] | [] | 2023-04-15T03:35:14Z | 2023-04-17T02:39:02Z | 2023-04-15T04:30:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghpkishore | 26,677,859 | MDQ6VXNlcjI2Njc3ODU5 | User | false |
huggingface/diffusers | 1,669,253,150 | I_kwDOHa8MBc5jfsge | 3,113 | https://github.com/huggingface/diffusers/issues/3113 | https://api.github.com/repos/huggingface/diffusers/issues/3113 | Training-Free Layout Control with Cross-Attention Guidance | ### Model/Pipeline/Scheduler description
Hey Team,
I have added a [diffusers-based pipeline](https://github.com/nipunjindal/diffusers-layout-guidance) for [Training-Free Layout Control with Cross-Attention Guidance](https://arxiv.org/abs/2304.03373) based on original work [here](https://github.com/silent-chen/layou... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-15T09:00:53Z | 2023-05-23T15:03:29Z | 2023-05-23T15:03:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nipunjindal | 6,430,864 | MDQ6VXNlcjY0MzA4NjQ= | User | false |
huggingface/diffusers | 1,669,477,305 | I_kwDOHa8MBc5jgjO5 | 3,115 | https://github.com/huggingface/diffusers/issues/3115 | https://api.github.com/repos/huggingface/diffusers/issues/3115 | Save base model in ControlNet training scripts when saving trained model or pushing to hub | This came up in the PR to provide ControlNet support to Apple's `ml-stable-diffusion`: https://github.com/apple/ml-stable-diffusion/pull/153. I've also seen people training ControlNet models for Stable Diffusion v2 and v2.1, so it could be useful to save this as a metadata field for users of the trained models.
This... | closed | completed | false | 5 | [] | [
"pcuenca"
] | 2023-04-15T16:36:03Z | 2023-04-27T14:47:48Z | 2023-04-27T14:47:47Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,669,498,314 | I_kwDOHa8MBc5jgoXK | 3,116 | https://github.com/huggingface/diffusers/issues/3116 | https://api.github.com/repos/huggingface/diffusers/issues/3116 | KDPM2DiscreteScheduler produces latent noise in image2image when called with certain inference steps | ### Describe the bug
When the KPM2DiscreteScheduler is used in an image2image generation, under certain combinations of strength and num_inference_steps the resulting image will be latent noise rather than the expected image. The issue appears to occur when `img2img_pipeline.scheduler.set_timesteps` builds a `timestep... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-04-15T17:33:10Z | 2023-04-17T10:52:42Z | 2023-04-17T10:52:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lstein | 111,189 | MDQ6VXNlcjExMTE4OQ== | User | false |
huggingface/diffusers | 1,669,824,783 | I_kwDOHa8MBc5jh4EP | 3,117 | https://github.com/huggingface/diffusers/issues/3117 | https://api.github.com/repos/huggingface/diffusers/issues/3117 | how to access progressbar? | hello, I would like to yield the current progress of the pipe but there's no way to access the printed progressbar. How can I give live updates on the progress in my fastapi app | closed | completed | false | 7 | [] | [] | 2023-04-16T09:55:05Z | 2024-03-30T07:15:47Z | 2023-04-17T11:57:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,669,873,362 | I_kwDOHa8MBc5jiD7S | 3,120 | https://github.com/huggingface/diffusers/issues/3120 | https://api.github.com/repos/huggingface/diffusers/issues/3120 | The controlnet trained by diffusers scripts produce always same result no matter what the input images is | ### Describe the bug
I train a controlnet with the base model Chilloutmix-Ni and datasets Abhilashvj/vto_hd_train using the train_controlnet.py script provided in diffuses repo
After training I got a controlnet model.
When I inference the image with the model, if I use the same prompt and seed, no matter how I cha... | closed | completed | false | 6 | [
"bug",
"stale"
] | [
"williamberman",
"yiyixuxu"
] | 2023-04-16T11:16:58Z | 2023-07-08T15:03:12Z | 2023-07-08T15:03:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | garyhxfang | 84,261,039 | MDQ6VXNlcjg0MjYxMDM5 | User | false |
huggingface/diffusers | 1,670,328,298 | I_kwDOHa8MBc5jjy_q | 3,124 | https://github.com/huggingface/diffusers/issues/3124 | https://api.github.com/repos/huggingface/diffusers/issues/3124 | Stable Diffusion models convert doesn't work | ### Describe the bug
In order to convert the sd models to diffusers, I open the site https://huggingface.co/spaces/diffusers/sd-to-diffusers.
And upload the model download from civitai. Finally, the model does not change to diffusers.
![image](https://user-images.githubusercontent.com/129836021/232367466-c8a497f9-... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-04-17T03:11:11Z | 2023-04-27T13:12:32Z | 2023-04-17T10:21:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | IDSdevops2023 | 129,836,021 | U_kgDOB70j9Q | User | false |
huggingface/diffusers | 1,358,717,378 | I_kwDOHa8MBc5Q_GHC | 313 | https://github.com/huggingface/diffusers/issues/313 | https://api.github.com/repos/huggingface/diffusers/issues/313 | Push to Hub Design | It would be nice to discuss a bit the push to hub design of the library.
IMO we have two different use cases for `push_to_hub`.
1. The complement of `from_pretrained(...)`. Every model, scheduler, and diffusion pipeline should eventually have a `push_to_hub(...)` method that allows one to push the class to the ... | closed | completed | false | 10 | [
"wip"
] | [
"sayakpaul"
] | 2022-09-01T11:48:17Z | 2024-11-21T11:26:53Z | 2024-11-21T11:26:53Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,671,539,867 | I_kwDOHa8MBc5joayb | 3,134 | https://github.com/huggingface/diffusers/issues/3134 | https://api.github.com/repos/huggingface/diffusers/issues/3134 | Paella v3 | ### Model/Pipeline/Scheduler description
Moving outdated pull request: https://github.com/huggingface/diffusers/pull/2058 here.
The amazing @dome272 published a new paella model which is significantly more performant than the previous models while being very fast.
https://laion.ai/blog/paella/
I think it wo... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-04-17T16:13:56Z | 2023-07-01T15:03:14Z | 2023-07-01T15:03:14Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,672,224,064 | I_kwDOHa8MBc5jrB1A | 3,139 | https://github.com/huggingface/diffusers/issues/3139 | https://api.github.com/repos/huggingface/diffusers/issues/3139 | How to use safetensors file for image generation? | Hi, Thanks for sharing this library for using stable diffusion.
There is one questions I want to ask.
I want to use safetensors file. (like civitai checkpoint, LoRA..)
do you share doc about using safetensors file?
1. how to convert safetensors file? (like [this](https://civitai.com/models/4823/deliberate))
2.... | closed | completed | false | 5 | [] | [] | 2023-04-18T02:14:18Z | 2023-06-28T13:55:27Z | 2023-04-19T00:31:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DevJunghun | 28,581,778 | MDQ6VXNlcjI4NTgxNzc4 | User | false |
huggingface/diffusers | 1,672,579,269 | I_kwDOHa8MBc5jsYjF | 3,140 | https://github.com/huggingface/diffusers/issues/3140 | https://api.github.com/repos/huggingface/diffusers/issues/3140 | ImportError: cannot import name 'StableDiffusionControlNetPipeline' | ### Describe the bug
I try to generate text-to-image using ControlNet described in documentation [here](https://huggingface.co/docs/diffusers/v0.15.0/en/api/pipelines/stable_diffusion/controlnet), But got import error when try to import StableDiffusionControlNetPipeline. Could anyone help me out? Thank you!
### Rep... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-04-18T08:06:14Z | 2023-09-16T18:37:19Z | 2023-04-19T06:48:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JiaojiaoYe1994 | 38,762,818 | MDQ6VXNlcjM4NzYyODE4 | User | false |
huggingface/diffusers | 1,672,590,835 | I_kwDOHa8MBc5jsbXz | 3,141 | https://github.com/huggingface/diffusers/issues/3141 | https://api.github.com/repos/huggingface/diffusers/issues/3141 | About automatic hyperparameter optimization | Hi,Hyperparameter optimization refers to the process of finding the configuration of hyperparameters for optimal performance. This process can be automated using existing optimization methods. Automated hyperparameter optimization is a brilliant technique. I found that you don't seem to use automatic hyperparameter opt... | closed | completed | false | 0 | [] | [] | 2023-04-18T08:13:36Z | 2023-04-19T00:35:23Z | 2023-04-19T00:35:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Vickie0710 | 97,725,167 | U_kgDOBdMq7w | User | false |
huggingface/diffusers | 1,672,830,629 | I_kwDOHa8MBc5jtV6l | 3,142 | https://github.com/huggingface/diffusers/issues/3142 | https://api.github.com/repos/huggingface/diffusers/issues/3142 | Safety checker outside pipeline | ### Describe the bug
Safety checker works well like this,
```
pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5",torch_dtype=torch.float16).to("cuda")
prompt = 'close-up photo of a knot in a rope'
result = pipe(prompt=[prompt).images
```
I want to take advantage of function "run... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-04-18T10:33:40Z | 2023-05-29T15:03:40Z | 2023-05-29T15:03:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | adhikjoshi | 11,740,719 | MDQ6VXNlcjExNzQwNzE5 | User | false |
huggingface/diffusers | 1,673,064,453 | I_kwDOHa8MBc5juPAF | 3,143 | https://github.com/huggingface/diffusers/issues/3143 | https://api.github.com/repos/huggingface/diffusers/issues/3143 | Transfer ControlNet to custom SD1.5 Models | Hi, do I need to transfer ControlNet to custom models based on SD 1.5 (e.g. Dreamshaper, Deliberate, Anything) as described here [https://github.com/lllyasviel/ControlNet/discussions/12](url) when using Diffusers?
If yes - what is currently the best method? In the thread they mention that tool_transfer_control.py f... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-04-18T12:50:51Z | 2023-05-26T15:03:15Z | 2023-05-26T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | korzen | 1,381,860 | MDQ6VXNlcjEzODE4NjA= | User | false |
huggingface/diffusers | 1,673,846,826 | I_kwDOHa8MBc5jxOAq | 3,147 | https://github.com/huggingface/diffusers/issues/3147 | https://api.github.com/repos/huggingface/diffusers/issues/3147 | Reducing model loads on pipeline switches | **What API design would you like to have changed or added to the library? Why?**
When switching between txt2img and img2img, it is possible to use the constructor directly:
```
StableDiffusionImg2ImgPipeline(**prev_pipeline.pipe.components)
```
I would like to enable similar patterns when using ControlNet. ... | closed | completed | false | 3 | [] | [] | 2023-04-18T21:19:10Z | 2024-01-04T05:32:55Z | 2023-04-23T22:18:15Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pdoane | 1,621,950 | MDQ6VXNlcjE2MjE5NTA= | User | false |
huggingface/diffusers | 1,673,849,723 | I_kwDOHa8MBc5jxOt7 | 3,148 | https://github.com/huggingface/diffusers/issues/3148 | https://api.github.com/repos/huggingface/diffusers/issues/3148 | Promoting community ControlNet pipelines | **What API design would you like to have changed or added to the library? Why?**
Recently the txt2img ControlNet equivalent has been made standard, but the remaining variants are community supported. Given how useful ControlNet is, I think it makes sense for the whole group to be promoted.
**What use case would ... | closed | completed | false | 7 | [
"stale"
] | [] | 2023-04-18T21:22:31Z | 2023-05-30T15:03:28Z | 2023-05-30T15:03:28Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pdoane | 1,621,950 | MDQ6VXNlcjE2MjE5NTA= | User | false |
huggingface/diffusers | 1,674,108,210 | I_kwDOHa8MBc5jyN0y | 3,149 | https://github.com/huggingface/diffusers/issues/3149 | https://api.github.com/repos/huggingface/diffusers/issues/3149 | Diffusers 0.14.0 Produces Non-Deterministic Images | ### Describe the bug
In diffusers 0.14.0, when I generate two images with the same set of parameters, they look exactly the same to the naked eye, however, their bytes are different. I am working on a feature where I was planning on using the sha256 of each images' bytes to serve as a unique identifier, hence I came u... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-04-19T03:20:47Z | 2023-04-19T11:34:50Z | 2023-04-19T11:34:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | eolszewski | 3,613,197 | MDQ6VXNlcjM2MTMxOTc= | User | false |
huggingface/diffusers | 1,674,259,703 | I_kwDOHa8MBc5jyyz3 | 3,150 | https://github.com/huggingface/diffusers/issues/3150 | https://api.github.com/repos/huggingface/diffusers/issues/3150 | Cannot load a pretrained CLIPTextModel while using SD2.1 | I'm tring to write a stable-diffusion pipeline on my own, and I read some tutorials about it, such as https://huggingface.co/blog/stable_diffusion
This tutorials is based on "CompVis/stable-diffusion-v1-4" and I tried it on colab, it works well.
But when I tried to change the model to "stabilityai/stable-diffusio... | closed | completed | false | 3 | [] | [] | 2023-04-19T06:36:35Z | 2024-12-10T11:06:18Z | 2023-04-20T06:56:25Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DrDavidS | 20,372,610 | MDQ6VXNlcjIwMzcyNjEw | User | false |
huggingface/diffusers | 1,674,377,225 | I_kwDOHa8MBc5jzPgJ | 3,151 | https://github.com/huggingface/diffusers/issues/3151 | https://api.github.com/repos/huggingface/diffusers/issues/3151 | What is the format of the training data | Hello，I'm training Lora, but I don't know what the data format looks like,
The error is as follows:
--caption_column' value 'text' needs to be one of: image
What is the data format? | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-19T07:51:16Z | 2023-08-04T10:20:18Z | 2023-05-28T15:03:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WGS-note | 58,907,127 | MDQ6VXNlcjU4OTA3MTI3 | User | false |
huggingface/diffusers | 1,674,381,911 | I_kwDOHa8MBc5jzQpX | 3,152 | https://github.com/huggingface/diffusers/issues/3152 | https://api.github.com/repos/huggingface/diffusers/issues/3152 | cannot import name '_pywrap_analyzer_wrapper' from 'tensorflow.lite.python.analyzer_wrapper' | ### Describe the bug
RuntimeError: Failed to import transformers.models.clip.image_processing_clip
because of the following error (look up to see its traceback):
cannot import name '_pywrap_analyzer_wrapper' from
'tensorflow.lite.python.analyzer_wrapper' (unknown location)
### Reproduction
RuntimeError: Failed ... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-04-19T07:54:42Z | 2023-05-28T15:03:02Z | 2023-05-28T15:03:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tensor-song | 9,344,500 | MDQ6VXNlcjkzNDQ1MDA= | User | false |
huggingface/diffusers | 1,674,771,504 | I_kwDOHa8MBc5j0vww | 3,154 | https://github.com/huggingface/diffusers/issues/3154 | https://api.github.com/repos/huggingface/diffusers/issues/3154 | Model not offloading to disk when RAM is full | ### Describe the bug
Looks like accelerate is not offloading models to disk when RAM is occupied. Am I missing something?
### Reproduction
Ran in machine with 16GB RAM
```
from transformers import AutoModelForCausalLM
import torch
checkpoint = "facebook/opt-6.7b"
model = AutoModelForCausalLM.from_pretra... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-04-19T12:07:27Z | 2023-04-25T15:08:33Z | 2023-04-25T15:08:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hari10599 | 37,787,894 | MDQ6VXNlcjM3Nzg3ODk0 | User | false |
huggingface/diffusers | 1,674,821,328 | I_kwDOHa8MBc5j077Q | 3,155 | https://github.com/huggingface/diffusers/issues/3155 | https://api.github.com/repos/huggingface/diffusers/issues/3155 | Loading Lora/Textual inversion and applying custom names to them | Hi, I want to prevent name conflicts so how can I update the name of the concept trained and load it in stable diffusion pipeline? | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-19T12:39:34Z | 2023-05-30T15:03:27Z | 2023-05-30T15:03:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,675,742,881 | I_kwDOHa8MBc5j4c6h | 3,158 | https://github.com/huggingface/diffusers/issues/3158 | https://api.github.com/repos/huggingface/diffusers/issues/3158 | How can I get seeds to reproduce images after use num_images_per_prompt > 1? | when I set num_images_per_prompt to 4 and provide a seed for generator, I got 4 different images, of course I can reproduce first image by the seed I provided before. But how can I reproduce the other 3 images?
there is my sketch:
`
pipeline = StableDiffusionPipeline.from_pretrained(repo_id, custom_pipeline="Ala... | closed | completed | false | 5 | [] | [] | 2023-04-19T22:50:21Z | 2023-04-23T01:04:25Z | 2023-04-23T01:04:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bubooK90 | 127,664,027 | U_kgDOB5v_mw | User | false |
huggingface/diffusers | 1,675,747,363 | I_kwDOHa8MBc5j4eAj | 3,159 | https://github.com/huggingface/diffusers/issues/3159 | https://api.github.com/repos/huggingface/diffusers/issues/3159 | Typo in code in documentation docs/source/en/using-diffusers/write_own_pipeline.mdx | ### Describe the bug
Docs has typo in step 6 of "Deconstruct a basic pipeline" https://github.com/huggingface/diffusers/blob/main/docs/source/en/using-diffusers/write_own_pipeline.mdx
image = Image.fromarray((image * 255)).round().astype("uint8")
should be
image = Image.fromarray((image * 255).round().astype... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-04-19T22:57:11Z | 2023-06-07T08:56:49Z | 2023-05-21T10:16:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ambisinister | 16,823,221 | MDQ6VXNlcjE2ODIzMjIx | User | false |
huggingface/diffusers | 1,676,094,437 | I_kwDOHa8MBc5j5yvl | 3,162 | https://github.com/huggingface/diffusers/issues/3162 | https://api.github.com/repos/huggingface/diffusers/issues/3162 | Expected 'learned_classifier_free_sampling_embeddings' but not passed | ### Describe the bug
Hello.
The following snippet causes this error:
```
ValueError: Pipeline <class 'diffusers.pipelines.vq_diffusion.pipeline_vq_diffusion.VQDiffusionPipeline'> expected {'text_encoder', 'transformer', 'scheduler', 'vqvae', 'tokenizer', 'learned_classifier_free_sampling_embeddings'}, but only ... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-04-20T06:39:48Z | 2023-05-31T15:03:38Z | 2023-05-31T15:03:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jamarju-fp | 130,987,646 | U_kgDOB862fg | User | false |
huggingface/diffusers | 1,676,271,297 | I_kwDOHa8MBc5j6d7B | 3,164 | https://github.com/huggingface/diffusers/issues/3164 | https://api.github.com/repos/huggingface/diffusers/issues/3164 | diffusers result is different from webui | The same prompt input effect is much worse than webui, I think it is the problem of text encoder, especially when the prompt is longer, is there a plan to do this? | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-20T08:39:02Z | 2023-05-30T15:03:25Z | 2023-05-30T15:03:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dingjingzhen | 21,244,263 | MDQ6VXNlcjIxMjQ0MjYz | User | false |
huggingface/diffusers | 1,676,575,051 | I_kwDOHa8MBc5j7oFL | 3,167 | https://github.com/huggingface/diffusers/issues/3167 | https://api.github.com/repos/huggingface/diffusers/issues/3167 | 0.15.1 version changes the weight shapes of unet !!! | ### Describe the bug
There is a horrible bug that the weight shapes of unet are changed after the version update from 0.13.1 to 0.15.1. I found it when I tried to use controlnet pipeline, the 0.13.1 has not supported controlnet yet, so I updated the diffusers. After that I tried on controlnet pipeline, and I found the... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-04-20T11:56:09Z | 2023-06-13T15:03:15Z | 2023-06-13T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CharlieLzy | 37,946,631 | MDQ6VXNlcjM3OTQ2NjMx | User | false |
huggingface/diffusers | 1,676,824,935 | I_kwDOHa8MBc5j8lFn | 3,171 | https://github.com/huggingface/diffusers/issues/3171 | https://api.github.com/repos/huggingface/diffusers/issues/3171 | LPW prompt for controlnet. | **Is your feature request related to a problem? Please describe.**
Currently, controlnet pipelines donot support community pipelines, I was hoping this feature can be implemented ? I really liked the LPW feature from the community pipeline and have hacked up a solution following it:
https://github.com/aravind-h-v/d... | closed | completed | false | 5 | [] | [] | 2023-04-20T14:17:31Z | 2023-05-06T05:37:56Z | 2023-05-06T05:37:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aravind-h-v | 122,849,931 | U_kgDOB1KKiw | User | false |
huggingface/diffusers | 1,677,094,997 | I_kwDOHa8MBc5j9nBV | 3,173 | https://github.com/huggingface/diffusers/issues/3173 | https://api.github.com/repos/huggingface/diffusers/issues/3173 | Random constant unconditional output on cfg breaks nccl when using multi-gpu accelerate | ### Describe the bug
Writing a custom encoder wrapped around accelerate to run with `--multi_gpu` on `nccl` backend I was getting an `nccl` error that I couldn't identify it's origin. The error wasn't appearing when running on 1 gpu (ofc without `--multi_gpu`). I managed to pin it down and it came from something rat... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-04-20T17:01:21Z | 2023-05-30T15:03:24Z | 2023-05-30T15:03:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ysig | 28,439,529 | MDQ6VXNlcjI4NDM5NTI5 | User | false |
huggingface/diffusers | 1,677,123,371 | I_kwDOHa8MBc5j9t8r | 3,174 | https://github.com/huggingface/diffusers/issues/3174 | https://api.github.com/repos/huggingface/diffusers/issues/3174 | New features on LORA break lora+dreambooth model loading | ### Describe the bug
Unable to load the attn_proc on the freshly trained dreambooth+lora. Just using the supplied example. It was working on 7b0ba4820a7546520da4b099fc6c523d5b6d3383
### Reproduction
Train a model and save the new weights in `"pytorch_lora_weights.bin"`. That works fine, then try to create infere... | closed | completed | false | 13 | [
"bug"
] | [] | 2023-04-20T17:21:30Z | 2023-05-02T03:09:51Z | 2023-04-28T06:06:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tcapelle | 18,441,985 | MDQ6VXNlcjE4NDQxOTg1 | User | false |
huggingface/diffusers | 1,677,259,573 | I_kwDOHa8MBc5j-PM1 | 3,175 | https://github.com/huggingface/diffusers/issues/3175 | https://api.github.com/repos/huggingface/diffusers/issues/3175 | Train Multi face with train_dreambooth.py on SD model | ### Describe the bug
Hi there!
I have an issue with training the SD model with train_dreambooth.py.
I am doing these steps:
**Step 1:**
Downloaded the Stable Diffusion from hugging face and using from train_dreambooth.py script to start to train my own face with some parameters and special SEED.
**Step 2:**... | closed | completed | false | 10 | [
"bug"
] | [] | 2023-04-20T19:03:29Z | 2023-05-01T09:14:59Z | 2023-04-28T09:25:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hosein-moayedi | 38,086,978 | MDQ6VXNlcjM4MDg2OTc4 | User | false |
huggingface/diffusers | 1,677,279,237 | I_kwDOHa8MBc5j-UAF | 3,176 | https://github.com/huggingface/diffusers/issues/3176 | https://api.github.com/repos/huggingface/diffusers/issues/3176 | TensorRT demo error | ### Describe the bug
TensorRT demo error:
<img width="1242" alt="image" src="https://user-images.githubusercontent.com/1241824/233465104-501d0b4c-035f-4742-9249-a97ee01833fe.png">
### Reproduction
Demo code for TensorRT:
https://github.com/huggingface/diffusers/blob/main/examples/community/README.md#tensorrt-tex... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-04-20T19:18:03Z | 2023-05-29T15:03:35Z | 2023-05-29T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gaoxiao | 1,241,824 | MDQ6VXNlcjEyNDE4MjQ= | User | false |
huggingface/diffusers | 1,677,303,988 | I_kwDOHa8MBc5j-aC0 | 3,177 | https://github.com/huggingface/diffusers/issues/3177 | https://api.github.com/repos/huggingface/diffusers/issues/3177 | train_dreambooth.py DeepSpeed offloading stage 3 seams broken | ### Describe the bug
Unfortionatly training with train_dreambooth.py @ commit 3045fb276352681f6b9075956e599dd8ef571872 fails with
```
Traceback (most recent call last):
File "/media/sharedHome/machine-lerning/Diffuserspayground/rwomen/../train_dreambooth.py", line 1039, in <module>
Traceback (most recent call... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-04-20T19:36:27Z | 2023-06-24T15:03:02Z | 2023-06-24T15:03:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | IMbackK | 13,803,414 | MDQ6VXNlcjEzODAzNDE0 | User | false |
huggingface/diffusers | 1,677,495,826 | I_kwDOHa8MBc5j_I4S | 3,178 | https://github.com/huggingface/diffusers/issues/3178 | https://api.github.com/repos/huggingface/diffusers/issues/3178 | VersatileDiffusionPipeline bug | ### Describe the bug
Hello! I try to use VersatileDiffusionPipeline. Following instructions I get "get_down_block() got an unexpected keyword argument 'resnet_skip_time_act'" error.
In diffusers/pipelines/versatile_diffusion/modeling_text_unet.py line 422 we really have "resnet_skip_time_act=resnet_skip_time_act"... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-04-20T22:08:57Z | 2023-05-29T15:03:33Z | 2023-05-29T15:03:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | messimm | 60,918,406 | MDQ6VXNlcjYwOTE4NDA2 | User | false |
huggingface/diffusers | 1,358,800,917 | I_kwDOHa8MBc5Q_agV | 318 | https://github.com/huggingface/diffusers/issues/318 | https://api.github.com/repos/huggingface/diffusers/issues/318 | Grad-TTS | ### Model/Pipeline/Scheduler description
Grad TTS model: https://github.com/huawei-noah/Speech-Backbones/blob/main/Grad-TTS/README.md
### Open source status
- [X] The model implementation is available
- [X] The model weights are available (Only relevant if addition is not a scheduler).
### Provide useful links for ... | closed | completed | false | 6 | [
"New pipeline/model"
] | [] | 2022-09-01T12:53:08Z | 2026-01-09T17:50:52Z | 2026-01-09T17:50:52Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,677,826,317 | I_kwDOHa8MBc5kAZkN | 3,181 | https://github.com/huggingface/diffusers/issues/3181 | https://api.github.com/repos/huggingface/diffusers/issues/3181 | TensorRT Example TypeError | ### Describe the bug
Running the example in the repo for TensorRT receiving this error:
```
File "/root/.cache/huggingface/modules/diffusers_modules/git/stable_diffusion_tensorrt_txt2img.py", line 706, in to
super().to(torch_device, silence_dtype_warnings)
File "/root/.pyenv/versions/3.9.16/lib/python3.9/si... | closed | completed | false | 14 | [
"bug",
"stale"
] | [] | 2023-04-21T05:16:49Z | 2023-06-04T15:03:09Z | 2023-06-04T15:03:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | djj0s3 | 1,120,027 | MDQ6VXNlcjExMjAwMjc= | User | false |
huggingface/diffusers | 1,677,834,221 | I_kwDOHa8MBc5kAbft | 3,182 | https://github.com/huggingface/diffusers/issues/3182 | https://api.github.com/repos/huggingface/diffusers/issues/3182 | custom_pipeline="lpw_stable_diffusion" won't work with pipe.enable_model_cpu_offload() | ### Describe the bug
As per title, everything works until adding the line to use long prompt weighting pipeline when calling enable_model_cpu_offload()
### Reproduction
use enable_model_cpu_offload() with custom_pipeline="lpw_stable_diffusion"
```
import torch
from diffusers import StableDiffusionPipeline
... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-04-21T05:25:30Z | 2023-05-30T15:03:22Z | 2023-05-30T15:03:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cian0 | 1,121,189 | MDQ6VXNlcjExMjExODk= | User | false |
huggingface/diffusers | 1,678,192,537 | I_kwDOHa8MBc5kBy-Z | 3,184 | https://github.com/huggingface/diffusers/issues/3184 | https://api.github.com/repos/huggingface/diffusers/issues/3184 | Cannot use generator list with community pipeline `lpw_stable_diffusion` | ### Describe the bug
When I add `lpw_stable_diffusion` in StableDiffusionPipeline and use a generator list, program crashed. Use just one generator, every thing goes fine.
### Reproduction
from diffusers import DiffusionPipeline
import torch
pipe = DiffusionPipeline.from_pretrained(
"hakurei/waifu-diffusion... | closed | not_planned | false | 0 | [
"bug"
] | [] | 2023-04-21T09:19:36Z | 2023-04-21T12:02:29Z | 2023-04-21T12:02:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bubooK90 | 127,664,027 | U_kgDOB5v_mw | User | false |
huggingface/diffusers | 1,678,419,095 | I_kwDOHa8MBc5kCqSX | 3,185 | https://github.com/huggingface/diffusers/issues/3185 | https://api.github.com/repos/huggingface/diffusers/issues/3185 | The training of Stable-diffusion-upscale-4x model | I try to fine-tune the Stable-diffusion-upscale-4x model on custom data, but get very wired results, much worse than original one. Can anyone help? LoRA is used and part of my training code is as follows (input image is already normalized to [-1, 1]):
```python
for epoch in range(first_epoch, args.num_train_epochs):
... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-04-21T12:08:45Z | 2023-09-22T14:27:16Z | 2023-05-30T15:03:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WUMINGCHAzero | 42,505,610 | MDQ6VXNlcjQyNTA1NjEw | User | false |
huggingface/diffusers | 1,358,803,632 | I_kwDOHa8MBc5Q_bKw | 319 | https://github.com/huggingface/diffusers/issues/319 | https://api.github.com/repos/huggingface/diffusers/issues/319 | VQ-Diffusion | ### Model/Pipeline/Scheduler description
VQ-Diffusion is based on a VQ-VAE whose latent space is modeled by a conditional variant of the recently developed Denoising Diffusion Probabilistic Model (DDPM). It produces significantly better text-to-image generation results when compared with Autoregressive models with sim... | closed | completed | false | 20 | [
"New pipeline/model"
] | [] | 2022-09-01T12:55:15Z | 2023-04-26T23:46:19Z | 2023-02-13T05:25:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,678,833,296 | I_kwDOHa8MBc5kEPaQ | 3,190 | https://github.com/huggingface/diffusers/issues/3190 | https://api.github.com/repos/huggingface/diffusers/issues/3190 | Missing docs about how to fine-tune models | ### Describe the bug
Currently there are no docs about how to adapt existing models for fine-tuning. E.g. fine-tuning a text-to-image model for inpainting or other tasks.
It's actually quite easy to initialize a unet from pretrained weights while having a different archicture for some part of the model as shown he... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-04-21T17:03:53Z | 2023-05-10T23:02:49Z | 2023-05-10T23:02:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,678,973,676 | I_kwDOHa8MBc5kExrs | 3,191 | https://github.com/huggingface/diffusers/issues/3191 | https://api.github.com/repos/huggingface/diffusers/issues/3191 | Dreambooth + Lora is broken | ### Describe the bug
The loading of the LoRA weights is not working anymore. Reverting to previous commit fixes the issue:
<img width="1155" alt="image" src="https://user-images.githubusercontent.com/18441985/233715934-7eb0c9e5-af23-423e-9ac9-7fa5794cdc85.png">
https://wandb.ai/capecape/dreambooth-lora/reports/Dre... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-04-21T19:10:39Z | 2023-04-28T13:06:00Z | 2023-04-28T13:06:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tcapelle | 18,441,985 | MDQ6VXNlcjE4NDQxOTg1 | User | false |
huggingface/diffusers | 1,679,302,736 | I_kwDOHa8MBc5kGCBQ | 3,192 | https://github.com/huggingface/diffusers/issues/3192 | https://api.github.com/repos/huggingface/diffusers/issues/3192 | reuse pipline instance for later inference | when i reuse the same stable-diffision text-to-image instance after pipe() called, it throws not all data in same device execption.
does it request a newly created pipeline, when calling pipe()? | closed | completed | false | 5 | [] | [] | 2023-04-22T03:26:50Z | 2023-04-25T07:24:49Z | 2023-04-25T07:23:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | junjiexv | 24,380,723 | MDQ6VXNlcjI0MzgwNzIz | User | false |
huggingface/diffusers | 1,679,331,699 | I_kwDOHa8MBc5kGJFz | 3,193 | https://github.com/huggingface/diffusers/issues/3193 | https://api.github.com/repos/huggingface/diffusers/issues/3193 | Batch load of textual inversion embeddings | **Is your feature request related to a problem? Please describe.**
Calls to load_textual_inversion() take about 0.4 seconds each on my test machine and I have many in a directory that are being loaded. It would be more acceptable if this was app startup time, but as described in #3147, I frequently need to recreate t... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-04-22T04:45:16Z | 2023-05-26T02:06:18Z | 2023-05-26T02:06:18Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pdoane | 1,621,950 | MDQ6VXNlcjE2MjE5NTA= | User | false |
huggingface/diffusers | 1,679,411,069 | I_kwDOHa8MBc5kGcd9 | 3,194 | https://github.com/huggingface/diffusers/issues/3194 | https://api.github.com/repos/huggingface/diffusers/issues/3194 | PackageNotFoundError: No package metadata was found for bitsandbytes | ### Describe the bug
I have a working configuration to load a text2img diffuser model with diffuser 0.12.1.
When I switch of diffuser version >= 0.13.0, I have an error:
PackageNotFoundError: No package metadata was found for bitsandbytes
Some dependencies are changed?
### Reproduction
- a working configurati... | closed | completed | false | 21 | [
"bug",
"stale"
] | [] | 2023-04-22T08:14:28Z | 2025-11-08T22:48:22Z | 2023-06-13T15:03:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | EnricoBeltramo | 22,332,651 | MDQ6VXNlcjIyMzMyNjUx | User | false |
huggingface/diffusers | 1,679,484,049 | I_kwDOHa8MBc5kGuSR | 3,195 | https://github.com/huggingface/diffusers/issues/3195 | https://api.github.com/repos/huggingface/diffusers/issues/3195 | Addition of "num_cycles" parameter to the cosine_with_restarts scheduler in train_unconditional.py | In the [train_unconditional.py ](https://github.com/huggingface/diffusers/blob/main/examples/unconditional_image_generation/train_unconditional.py) script, I noticed that even after choosing the cosine_with_restarts learning rate scheduler. it's acts just like a normal cosine scheduler. That's because the [get_schedule... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-22T11:10:55Z | 2023-07-20T13:11:55Z | 2023-05-30T15:03:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lukewardford | 118,485,668 | U_kgDOBw_ypA | User | false |
huggingface/diffusers | 1,679,500,012 | I_kwDOHa8MBc5kGyLs | 3,196 | https://github.com/huggingface/diffusers/issues/3196 | https://api.github.com/repos/huggingface/diffusers/issues/3196 | I can't load safetensors model | The code is:
```
from diffusers import ControlNetModel
control1 = ControlNetModel.from_pretrained("./model/ControlNet/sd-controlnet-canny")
control2 = ControlNetModel.from_pretrained("./model/ControlNet/sd-controlnet-seg")
```
The models are local:
 where `EasyNegative.pt` loads fine but `EasyNegative.safetensors` causes a crash. I'm not sure if that's a problem with the embedding itself or a potential bug.
### R... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-04-22T14:49:38Z | 2023-04-26T13:05:58Z | 2023-04-26T13:05:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sakistriker | 115,505,099 | U_kgDOBuJ3yw | User | false |
huggingface/diffusers | 1,679,575,012 | I_kwDOHa8MBc5kHEfk | 3,199 | https://github.com/huggingface/diffusers/issues/3199 | https://api.github.com/repos/huggingface/diffusers/issues/3199 | Custom pipelines are broken | ### Describe the bug
It worked a couple of days ago, but today it stopped to. I wish I have never restarted none my space. Downgrading to diffusers 0.11.0, 0.12.0, 0.14.0 etc didn't help.
Please help me!
### Reproduction
```python
from diffusers import DiffusionPipeline, EulerAncestralDiscreteScheduler
im... | closed | completed | false | 29 | [
"bug"
] | [] | 2023-04-22T14:55:19Z | 2023-08-24T08:51:04Z | 2023-04-22T21:19:38Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nickkolok | 5,008,131 | MDQ6VXNlcjUwMDgxMzE= | User | false |
huggingface/diffusers | 1,282,809,908 | I_kwDOHa8MBc5MdiA0 | 32 | https://github.com/huggingface/diffusers/issues/32 | https://api.github.com/repos/huggingface/diffusers/issues/32 | cifar10 quality much worse than DDPM paper | Here are my cifar10 32x32 results, with 7x NVIDIA GeForce RTX 2080 Ti with 11GB VRAM trained for ~10 hours with:
` python3 -m torch.distributed.run --nproc_per_node 7 train_unconditional.py --dataset="cifar10" --resolution=32 --output_dir="cifar10-ddpm-" --batch_size=16 --num_epochs=100 --gradient_accumulation_... | closed | completed | false | 4 | [] | [
"anton-l"
] | 2022-06-23T18:52:40Z | 2022-07-21T19:04:25Z | 2022-07-21T19:04:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | richardrl | 5,181,990 | MDQ6VXNlcjUxODE5OTA= | User | false |
huggingface/diffusers | 1,358,806,524 | I_kwDOHa8MBc5Q_b38 | 320 | https://github.com/huggingface/diffusers/issues/320 | https://api.github.com/repos/huggingface/diffusers/issues/320 | Notes2Audio | ### Model/Pipeline/Scheduler description
https://github.com/magenta/music-spectrogram-diffusion
Very cool model that can transform notes to music.
**Note**: Requires translating the code from Flax to PyTorch.
### Open source status
- [X] The model implementation is available
- [X] The model weights are availabl... | closed | completed | false | 4 | [
"New pipeline/model"
] | [] | 2022-09-01T12:57:21Z | 2023-03-23T13:25:21Z | 2023-03-23T13:25:21Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,679,828,329 | I_kwDOHa8MBc5kICVp | 3,202 | https://github.com/huggingface/diffusers/issues/3202 | https://api.github.com/repos/huggingface/diffusers/issues/3202 | AttributeError: 'StableDiffusionControlNetPipeline' object has no attribute 'enable_vae_tiling' | ### Describe the bug
I am trying to enable vae tiling an a Control Net Pipeline but get this issue. This option seems to work on StableDiffusionPipeline but does not on ControlNetPipeline. Please advice.
### Reproduction
```
pipeline = StableDiffusionControlNetPipeline.from_pretrained(models[model_index]["path"],... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-04-23T02:50:37Z | 2023-04-25T22:46:24Z | 2023-04-25T22:46:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alejobrainz | 6,945,663 | MDQ6VXNlcjY5NDU2NjM= | User | false |
huggingface/diffusers | 1,679,953,873 | I_kwDOHa8MBc5kIg_R | 3,203 | https://github.com/huggingface/diffusers/issues/3203 | https://api.github.com/repos/huggingface/diffusers/issues/3203 | UNet1DModel not converging on single batch | Hello,
I'm currently trying to implement a latent generative diffuser using a custom-trained VAE which maps 3x256x256 to 1x256 and a UNet1DModel with the following parameters :
UNet1DModel(in_channels = 1, out_channels=1,down_block_types =
('DownBlock1D', 'DownBlock1D', 'DownBlock1D','AttnDownB... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-04-23T09:52:35Z | 2023-05-31T15:03:36Z | 2023-05-31T15:03:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DeepHunter317 | 131,586,681 | U_kgDOB9faeQ | User | false |
huggingface/diffusers | 1,680,026,758 | I_kwDOHa8MBc5kIyyG | 3,204 | https://github.com/huggingface/diffusers/issues/3204 | https://api.github.com/repos/huggingface/diffusers/issues/3204 | Can't load local model | ```
import torch
from diffusers import StableDiffusionControlNetPipeline, ControlNetModel, UniPCMultistepScheduler,ModelMixin
from io import BytesIO
from PIL import Image
import torch.multiprocessing as mp
import Loras
import random
controlnet = ControlNetModel.from_pretrained("lllyasviel/control_... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-04-23T13:23:09Z | 2023-05-31T15:03:35Z | 2023-05-31T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,680,179,936 | I_kwDOHa8MBc5kJYLg | 3,205 | https://github.com/huggingface/diffusers/issues/3205 | https://api.github.com/repos/huggingface/diffusers/issues/3205 | google colab tracebacks become un-useable after installing | after pip install diffusers, the traceback format is swapped with a multi colored comic sans ms style font without hyperlink pointers to the exception line, with nothing but the line number where an exception occured or .py files with the exception are no longe clickable from within the exception traceback. I think it ... | closed | completed | false | 7 | [
"stale"
] | [] | 2023-04-23T21:14:44Z | 2023-06-04T15:03:06Z | 2023-06-04T15:03:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fractaldna22 | 37,323,518 | MDQ6VXNlcjM3MzIzNTE4 | User | false |
huggingface/diffusers | 1,680,545,669 | I_kwDOHa8MBc5kKxeF | 3,207 | https://github.com/huggingface/diffusers/issues/3207 | https://api.github.com/repos/huggingface/diffusers/issues/3207 | About the attention implementation with torch < 2.0 | ### Describe the bug
I tried to run `train_unconditional.py` with torch 1.12.1 but failed.
The bug seems in https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/attention.py#L94
I add a code `batch_size = batch_size // head_size` right after L90, and the program seems to work well for now. But I... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-04-24T06:28:34Z | 2023-04-25T08:02:53Z | 2023-04-25T08:02:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tyshiwo1 | 34,150,970 | MDQ6VXNlcjM0MTUwOTcw | User | false |
huggingface/diffusers | 1,680,955,287 | I_kwDOHa8MBc5kMVeX | 3,209 | https://github.com/huggingface/diffusers/issues/3209 | https://api.github.com/repos/huggingface/diffusers/issues/3209 | AutoencoderKL encoder outputs NaN for large images | ### Describe the bug
AutoEncoderKL encoder loaded from runwayml/stable-diffusion-v1-5 outputs NaN for large images. I observe this behavior for image sizes starting from around 1500x1500 with vae_tiling disabled. I tried with float32, float16, with and without xFormers. Is it an expected behavior ?
I would have ... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-04-24T10:37:08Z | 2023-06-02T15:03:24Z | 2023-06-02T15:03:24Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | antoine-scenario | 129,726,301 | U_kgDOB7t3XQ | User | false |
huggingface/diffusers | 1,358,808,177 | I_kwDOHa8MBc5Q_cRx | 321 | https://github.com/huggingface/diffusers/issues/321 | https://api.github.com/repos/huggingface/diffusers/issues/321 | Blended Diffusion | ### Model/Pipeline/Scheduler description
https://github.com/omriav/blended-diffusion
Cool model for in-painting
### Open source status
- [X] The model implementation is available
- [X] The model weights are available (Only relevant if addition is not a scheduler).
### Provide useful links for the implementation
... | closed | completed | false | 11 | [
"New pipeline/model"
] | [] | 2022-09-01T12:58:40Z | 2026-01-09T17:52:44Z | 2026-01-09T17:52:44Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,681,595,862 | I_kwDOHa8MBc5kOx3W | 3,211 | https://github.com/huggingface/diffusers/issues/3211 | https://api.github.com/repos/huggingface/diffusers/issues/3211 | AttributeError: 'StableDiffusionMegaPipeline' object has no attribute 'requires_safety_checker' | ### Describe the bug
Sample code in the readme throws an error
[stable_diffusion_mega](https://github.com/huggingface/diffusers/blob/7e6886f5e93ca9bb1e6d4beece46fe1e43b819c2/examples/community/README.md#stable-diffusion-mega)
AttributeError: 'StableDiffusionMegaPipeline' object has no attribute 'requires_safety_... | closed | completed | false | 7 | [
"bug"
] | [] | 2023-04-24T15:59:37Z | 2023-04-24T21:24:34Z | 2023-04-24T20:41:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KevinColemanInc | 2,381,771 | MDQ6VXNlcjIzODE3NzE= | User | false |
huggingface/diffusers | 1,681,721,223 | I_kwDOHa8MBc5kPQeH | 3,212 | https://github.com/huggingface/diffusers/issues/3212 | https://api.github.com/repos/huggingface/diffusers/issues/3212 | Add clip skip for diffusion pipeline | ### Introduction
clip skip is a trick to feed the early-stopped features encoded by `CLIPTextModel` into the cross-attention. If `clip_skip = 2`, it means that we want to use the features from the layer before the last of the clip text encoder to guide our image generation. And our current diffusion pipeline can be ... | closed | completed | false | 34 | [] | [] | 2023-04-24T17:22:07Z | 2024-10-15T13:25:08Z | 2023-09-20T08:04:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NormXU | 33,339,685 | MDQ6VXNlcjMzMzM5Njg1 | User | false |
huggingface/diffusers | 1,681,983,123 | I_kwDOHa8MBc5kQQaT | 3,213 | https://github.com/huggingface/diffusers/issues/3213 | https://api.github.com/repos/huggingface/diffusers/issues/3213 | `lpw_stable_diffusion` pipeline is broken. Again. | ### Describe the bug
Just as in #3199 , `lpw_stable_diffusion` pipeline has suddenly broke today on an ordinary `Factory reboot` of my space.
### Reproduction
I've created a public HF space to reproduce the bug:
https://huggingface.co/spaces/NickKolok/test-20230424-bug-reprod/
Feel free to duplicate it and play ... | closed | completed | false | 16 | [
"bug"
] | [] | 2023-04-24T20:27:50Z | 2023-05-03T14:53:52Z | 2023-05-03T14:53:52Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nickkolok | 5,008,131 | MDQ6VXNlcjUwMDgxMzE= | User | false |
huggingface/diffusers | 1,682,272,187 | I_kwDOHa8MBc5kRW-7 | 3,218 | https://github.com/huggingface/diffusers/issues/3218 | https://api.github.com/repos/huggingface/diffusers/issues/3218 | excessive graph breaks on `attention.py` and `attention_processor.py` for control_net on `torch.compile` | ### Describe the bug
I tried to run the controlnet example from this blog [post](https://huggingface.co/blog/controlnet) and it turned out that the `BasicTransformerBlock` is causing a large number of graph breaks (>100) on a single controlnet pipeline. Ideally the whole `BasicTransformerBlock.forward` should be inclu... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-04-25T01:32:38Z | 2023-05-26T18:43:53Z | 2023-05-26T18:43:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shingjan | 11,846,349 | MDQ6VXNlcjExODQ2MzQ5 | User | false |
huggingface/diffusers | 1,682,312,763 | I_kwDOHa8MBc5kRg47 | 3,219 | https://github.com/huggingface/diffusers/issues/3219 | https://api.github.com/repos/huggingface/diffusers/issues/3219 | Docs for LoRA are confusing | Apologies if API Design is not the right tag,
I was reading the docs for this page https://huggingface.co/docs/diffusers/training/lora and for a number of reasons it seemed very confusing to me as a end-user:
1. At the top of the page "Currently, LoRA is only supported for the attention layers of the UNet2DCondit... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-25T02:09:39Z | 2023-06-02T15:03:22Z | 2023-06-02T15:03:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | oxysoft | 3,945,277 | MDQ6VXNlcjM5NDUyNzc= | User | false |
huggingface/diffusers | 1,358,814,496 | I_kwDOHa8MBc5Q_d0g | 322 | https://github.com/huggingface/diffusers/issues/322 | https://api.github.com/repos/huggingface/diffusers/issues/322 | V-Diffusion from Katherine | ### Model/Pipeline/Scheduler description
Katherine's trained models in v-diffusion are very powerful especially if used with CLIP-guidance.
Would be great to have them included here as well!
https://github.com/crowsonkb/v-diffusion-pytorch
### Open source status
- [X] The model implementation is available
- [X... | closed | completed | false | 5 | [
"New pipeline/model"
] | [] | 2022-09-01T13:03:14Z | 2023-10-02T16:45:32Z | 2023-10-02T16:45:32Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,682,365,003 | I_kwDOHa8MBc5kRtpL | 3,220 | https://github.com/huggingface/diffusers/issues/3220 | https://api.github.com/repos/huggingface/diffusers/issues/3220 | trigger | diffuser support adding "trigger" key in diffuser pipelines?how? | closed | completed | false | 2 | [
"stale"
] | [] | 2023-04-25T03:25:15Z | 2023-06-02T15:03:21Z | 2023-06-02T15:03:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wuliebucha | 31,274,177 | MDQ6VXNlcjMxMjc0MTc3 | User | false |
huggingface/diffusers | 1,682,574,782 | I_kwDOHa8MBc5kSg2- | 3,221 | https://github.com/huggingface/diffusers/issues/3221 | https://api.github.com/repos/huggingface/diffusers/issues/3221 | pipe.unet.load_attn_procs not working in diffusers` version: 0.16.0 | ### Describe the bug
I use unet.load_attn_procs but not working after update diffusers project.
what is load_lora_weights?
what is the difference between new and old lora format?
How can I know it is new or old?
### Reproduction
When
```python
pipe.load_lora_weights("./testlora/pytorch_lora_weights.bin... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-04-25T07:14:28Z | 2023-05-04T09:28:47Z | 2023-05-04T09:28:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sing817 | 37,936,950 | MDQ6VXNlcjM3OTM2OTUw | User | false |
huggingface/diffusers | 1,682,770,983 | I_kwDOHa8MBc5kTQwn | 3,222 | https://github.com/huggingface/diffusers/issues/3222 | https://api.github.com/repos/huggingface/diffusers/issues/3222 | no " pytorch_custom_diffusion_weights.bin" after training in custom_diffusion. | ### Describe the bug
"Error no file named pytorch_custom_diffusion_weights.bin found in directory xxx"
after training, I get these files:
<img width="195" alt="image" src="https://user-images.githubusercontent.com/94051337/234231865-fe46ec9e-f962-418d-a79e-73c7a1e7b528.png">
which one should I load fot inference?
... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-04-25T09:18:38Z | 2024-02-14T08:39:22Z | 2023-06-02T15:03:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yuxu915 | 94,051,337 | U_kgDOBZscCQ | User | false |
huggingface/diffusers | 1,682,886,288 | I_kwDOHa8MBc5kTs6Q | 3,223 | https://github.com/huggingface/diffusers/issues/3223 | https://api.github.com/repos/huggingface/diffusers/issues/3223 | RAM not cleared even after moving stable diffusion pipeline to cuda | ### Describe the bug
Moving the SD model to cuda after loading it, doesn't clear the RAM.
Before loading SD model,
- GPU usage: 0MiB / 49140MiB
- RAM usage: 2.232330322265625/41.46642303466797
After loading SD model,
- GPU usage: 3755MiB / 49140MiB
- RAM usage: 5.694705963134766/37.9942741394043
Why i... | closed | completed | false | 11 | [
"bug",
"stale"
] | [] | 2023-04-25T10:25:34Z | 2023-06-02T15:03:19Z | 2023-06-02T15:03:19Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hari10599 | 37,787,894 | MDQ6VXNlcjM3Nzg3ODk0 | User | false |
huggingface/diffusers | 1,682,895,097 | I_kwDOHa8MBc5kTvD5 | 3,224 | https://github.com/huggingface/diffusers/issues/3224 | https://api.github.com/repos/huggingface/diffusers/issues/3224 | Learning rate scaling duplicated in Dreambooth LoRa training script | ### Describe the bug
The learning rate scaling is applied twice in the Dreambooth Lora training script, both in line 749 and in line 759. I assume this is not intentional:
https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora.py#L749
```
if args.scale_lr:
args.lea... | closed | completed | false | 1 | [
"bug"
] | [
"sayakpaul"
] | 2023-04-25T10:31:58Z | 2023-05-03T04:43:07Z | 2023-05-03T04:43:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stefan-canva | 90,706,016 | MDQ6VXNlcjkwNzA2MDE2 | User | false |
huggingface/diffusers | 1,683,211,397 | I_kwDOHa8MBc5kU8SF | 3,227 | https://github.com/huggingface/diffusers/issues/3227 | https://api.github.com/repos/huggingface/diffusers/issues/3227 | Winograd Convolution & other Speed Is All You Need optimizations | **Is your feature request related to a problem? Please describe.**
Right now, Stable Diffusion, and other diffusers models are really optimized, but a recent preprint from Google researchers found that Winograd Convolutions could reduce latency more than using FlashAttention.
The paper in question: https://arxiv.org/... | open | null | false | 4 | [
"wip"
] | [] | 2023-04-25T13:47:09Z | 2023-06-20T17:37:58Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Conobi | 5,563,277 | MDQ6VXNlcjU1NjMyNzc= | User | false |
huggingface/diffusers | 1,683,277,631 | I_kwDOHa8MBc5kVMc_ | 3,228 | https://github.com/huggingface/diffusers/issues/3228 | https://api.github.com/repos/huggingface/diffusers/issues/3228 | [instructpixel2pixel]RuntimeError: For now only a certain number of K values are supported. Let us know if you hit this and we will fix it | ### Describe the bug
RuntimeError: For now only a certain number of K values are supported. Let us know if you hit this and we will fix it
### Reproduction
accelerate launch --mixed_precision="fp16" train_instruct_pix2pix.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--dataset_name=$DATASET_ID \
... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-04-25T14:23:02Z | 2023-06-02T15:03:17Z | 2023-06-02T15:03:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lsabrinax | 17,122,734 | MDQ6VXNlcjE3MTIyNzM0 | User | false |
huggingface/diffusers | 1,683,991,414 | I_kwDOHa8MBc5kX6t2 | 3,231 | https://github.com/huggingface/diffusers/issues/3231 | https://api.github.com/repos/huggingface/diffusers/issues/3231 | null on exported json config results in custom model not being loaded | Hi,
I have created my own CustomImage pipeline using the `save_pretrained` method using a `CustomEncoder` imported from a local file. The saved general config file looks like this:
```
{
"_class_name": "CustomImagePipeline",
"_diffusers_version": "0.14.0",
"customencoder": [
null,
"CustomEncod... | closed | completed | false | 10 | [
"stale"
] | [] | 2023-04-25T23:09:08Z | 2023-07-01T15:03:13Z | 2023-07-01T15:03:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ysig | 28,439,529 | MDQ6VXNlcjI4NDM5NTI5 | User | false |
huggingface/diffusers | 1,684,173,708 | I_kwDOHa8MBc5kYnOM | 3,232 | https://github.com/huggingface/diffusers/issues/3232 | https://api.github.com/repos/huggingface/diffusers/issues/3232 | How to user safetensors file in pipe | Hi, I'm looking for an example to directly use safetensors file (e.g. v1-5-pruned.safetensors) in pipe such as
```
DiffusionPipeline.from_pretrained("E:\\aimodels\\v1-5-pruned.safetensors", use_safetensors=True)
```
but what I can find is something like
```
DiffusionPipeline.from_pretrained(pretrained_model_n... | closed | completed | false | 7 | [
"stale"
] | [] | 2023-04-26T02:48:09Z | 2023-06-04T15:03:03Z | 2023-06-04T15:03:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | VmFree | 3,304,239 | MDQ6VXNlcjMzMDQyMzk= | User | false |
huggingface/diffusers | 1,684,209,992 | I_kwDOHa8MBc5kYwFI | 3,233 | https://github.com/huggingface/diffusers/issues/3233 | https://api.github.com/repos/huggingface/diffusers/issues/3233 | Convert trained LoRA to safetensor | ### Describe the bug
I trained LoRA with locally downloaded stable-diffusion-v1-5 base model, following the tutorial. It was well trained and no error popped. I got checkpoints like:
```
|- checkpoint-500
- optimizer.bin
- pytorch_model.bin
- random_states_0.pkl
- scheduler.bin
|- ....
```
I downloa... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-04-26T03:37:31Z | 2023-06-04T15:03:02Z | 2023-06-04T15:03:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gitlabspy | 61,307,585 | MDQ6VXNlcjYxMzA3NTg1 | User | false |
huggingface/diffusers | 1,684,502,459 | I_kwDOHa8MBc5kZ3e7 | 3,235 | https://github.com/huggingface/diffusers/issues/3235 | https://api.github.com/repos/huggingface/diffusers/issues/3235 | Custom Diffusion weights with biases wouldn't load | ### Describe the bug
It's possible to train Custom Diffusion with biases but the inference pipeline doesn't correctly load weights with biases.
A quick dirty fix by setting `bias=True` in the [relevant lines](https://github.com/huggingface/diffusers/blob/e51f19aee82c8dd874b715a09dbc521d88835d68/src/diffusers/models... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-04-26T08:11:40Z | 2023-06-04T15:03:01Z | 2023-06-04T15:03:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | guaneec | 24,762,404 | MDQ6VXNlcjI0NzYyNDA0 | User | false |
huggingface/diffusers | 1,684,681,113 | I_kwDOHa8MBc5kajGZ | 3,238 | https://github.com/huggingface/diffusers/issues/3238 | https://api.github.com/repos/huggingface/diffusers/issues/3238 | Convert LoRA safetensor to diffusers form | ### Describe the bug
The error occured when converting the custom LoRA weight in safetensor to diffuesrs form. I used this [script](https://github.com/huggingface/diffusers/blob/main/scripts/convert_lora_safetensor_to_diffusers.py)
I was converting the LoRA weight from [civitai](https://civitai.com/models/23780/bur... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-04-26T09:51:10Z | 2023-06-04T15:03:00Z | 2023-06-04T15:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | smko77 | 55,199,527 | MDQ6VXNlcjU1MTk5NTI3 | User | false |
huggingface/diffusers | 1,685,506,606 | I_kwDOHa8MBc5kdsou | 3,247 | https://github.com/huggingface/diffusers/issues/3247 | https://api.github.com/repos/huggingface/diffusers/issues/3247 | test_unclip.py is failing | ### Describe the bug
test_unclip.py is failing and I think because of which PR checks are also failing
### Reproduction
pytest tests/pipelines/unclip/test_unclip.py
### Logs
```shell
FAILED tests/pipelines/unclip/test_unclip.py::UnCLIPPipelineFastTests::test_attention_slicing_forward_pass - RuntimeError: shape '[... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-04-26T18:23:44Z | 2023-04-27T06:01:18Z | 2023-04-27T06:01:18Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | init-22 | 34,376,531 | MDQ6VXNlcjM0Mzc2NTMx | User | false |
huggingface/diffusers | 1,685,749,407 | I_kwDOHa8MBc5ken6f | 3,249 | https://github.com/huggingface/diffusers/issues/3249 | https://api.github.com/repos/huggingface/diffusers/issues/3249 | upsample_size not an argument for AttnUpBlock2D | ### Describe the bug
The argument `upsample_size` is passed to each upsampling block [here](https://github.com/huggingface/diffusers/blob/e0a2bd15f9a1eb0d48a69973a9c7ddb4eabb1a27/src/diffusers/models/unet_2d_condition.py#L779). However, it looks like `AttnUpBlock2D` doesn't accept this [argument](https://github.com/hu... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-04-26T21:27:15Z | 2023-05-05T18:50:42Z | 2023-05-05T18:50:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | will-rice | 25,072,137 | MDQ6VXNlcjI1MDcyMTM3 | User | false |
huggingface/diffusers | 1,685,964,781 | I_kwDOHa8MBc5kfcft | 3,250 | https://github.com/huggingface/diffusers/issues/3250 | https://api.github.com/repos/huggingface/diffusers/issues/3250 | Can't run StableDiffusionKDiffusionPipeline | ### Describe the bug
I would like to use StableDiffusionKDiffusionPipeline, but when I run it directly, an error occurs with the message "AttributeError: 'StableDiffusionKDiffusionPipeline' object has no attribute 'sampler'".
### Reproduction
```python
from diffusers import StableDiffusionKDiffusionPipeline
... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-04-27T01:57:22Z | 2023-04-27T03:36:36Z | 2023-04-27T03:36:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sanbuphy | 96,160,062 | U_kgDOBbtJPg | User | false |
huggingface/diffusers | 1,685,983,646 | I_kwDOHa8MBc5kfhGe | 3,251 | https://github.com/huggingface/diffusers/issues/3251 | https://api.github.com/repos/huggingface/diffusers/issues/3251 | Controlnet 1.1 Shuffle should only be used on CFG side, even when not in guess mode | ### Describe the bug
Most ControlNet models should run on both conditional and unconditional sides, _except_ in guess mode where they only run on the conditional side.
Controlnet 1.1 Shuffle model should only run on the conditional side _all the time_, whether in guess mode or not. It looks like the current control... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2023-04-27T02:24:12Z | 2023-06-06T15:03:30Z | 2023-06-06T15:03:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hafriedlander | 114,769 | MDQ6VXNlcjExNDc2OQ== | User | false |
huggingface/diffusers | 1,686,036,137 | I_kwDOHa8MBc5kft6p | 3,253 | https://github.com/huggingface/diffusers/issues/3253 | https://api.github.com/repos/huggingface/diffusers/issues/3253 | Why webui and diffuser chose the same prompt and sampler but cannot get the same results? | Hello, I would like to ask why the WebUI of Auto111 and all Diffusers on github cannot get the same results? I want to know the reason, thank you very much. | closed | completed | false | 12 | [] | [] | 2023-04-27T03:39:27Z | 2023-12-15T03:03:30Z | 2023-04-28T01:48:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sanbuphy | 96,160,062 | U_kgDOBbtJPg | User | false |
huggingface/diffusers | 1,686,153,401 | I_kwDOHa8MBc5kgKi5 | 3,254 | https://github.com/huggingface/diffusers/issues/3254 | https://api.github.com/repos/huggingface/diffusers/issues/3254 | Conversion of Inpainting CKPT to Diffusers format Results an error | Hello @patrickvonplaten, I have been trying to use this https://www.reddit.com/r/StableDiffusion/comments/11bdzsc/how_to_convert_inpanting_ckpt_model_to_diffusers/ methodology proposed here, I have experimented with two models and results in the same error
1. I have used the runwayml stable diffusion inpainting ckpt ... | closed | completed | false | 15 | [] | [] | 2023-04-27T05:43:24Z | 2023-05-11T15:20:31Z | 2023-05-01T05:39:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | satwiksunnam19 | 83,942,187 | MDQ6VXNlcjgzOTQyMTg3 | User | false |
huggingface/diffusers | 1,686,275,198 | I_kwDOHa8MBc5kgoR- | 3,255 | https://github.com/huggingface/diffusers/issues/3255 | https://api.github.com/repos/huggingface/diffusers/issues/3255 | (bug) audioldm ckpt to diffuser script is not working at medium or large size ckpt. |
HI. i tried to convert diffusers model from audioldm.
small size ckpt like as audioldm-full-s-v2.ckpt audioldm-s-text-ft.ckpt are well,
but middle or large size are not woking (audioldm-m-text-ft.ckpt, audioldm-m-full.ckpt)below
shape torch.Size([960]) from checkpoint, the shape in current model is
torch.Size(... | closed | completed | false | 4 | [] | [] | 2023-04-27T07:25:52Z | 2023-05-06T11:00:43Z | 2023-05-06T11:00:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | circuluspibo | 50,613,069 | MDQ6VXNlcjUwNjEzMDY5 | User | false |
huggingface/diffusers | 1,686,284,725 | I_kwDOHa8MBc5kgqm1 | 3,256 | https://github.com/huggingface/diffusers/issues/3256 | https://api.github.com/repos/huggingface/diffusers/issues/3256 | Bugs about modifying the corresponding value of text_encoder.text_model.embeddings.token_embedding.weight | ### Describe the bug
Hi. The problem is shown below:
I want to set a trainable parameter with MyFunction to modify the embeddings of text_encoder.text_model.embeddings.token_embedding.weight.
However, I got an Error: RuntimeError: a view of a leaf Variable that requires grad is being used in an in-place operatio... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-04-27T07:32:25Z | 2023-04-27T11:56:23Z | 2023-04-27T11:56:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ShianDu | 78,642,067 | MDQ6VXNlcjc4NjQyMDY3 | User | false |
huggingface/diffusers | 1,686,358,238 | I_kwDOHa8MBc5kg8je | 3,258 | https://github.com/huggingface/diffusers/issues/3258 | https://api.github.com/repos/huggingface/diffusers/issues/3258 | convert_lora_safetensor_to_diffusers OSError: No such device (os error 19) | ```shell
python scripts/convert_lora_safetensor_to_diffusers.py \
--base_model_path=stabilityai/stable-diffusion-2-1-base \
--checkpoint_path="my_lora_output_pytorch_lora_weights_file" \
--dump_path="./testoutput" \
```
get:
File "conda/envs/diffusers/lib/python3.10/site-packages/safetensors/torch.py", line... | closed | completed | false | 20 | [
"stale"
] | [] | 2023-04-27T08:19:10Z | 2024-08-05T02:54:45Z | 2023-06-04T15:02:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sing817 | 37,936,950 | MDQ6VXNlcjM3OTM2OTUw | User | false |
huggingface/diffusers | 1,686,527,657 | I_kwDOHa8MBc5khl6p | 3,261 | https://github.com/huggingface/diffusers/issues/3261 | https://api.github.com/repos/huggingface/diffusers/issues/3261 | Revisit Conversion Space | - [ ] No way to specify a model if the user does not have any models in their account: https://huggingface.co/spaces/diffusers/sd-to-diffusers/discussions/7#644a3407f3ff2cb84eb4147d
- [ ] Make it compatible with 2.1 (not base) models.
- [ ] Make it clearer how users can specify arbitrary models (hard to see "Other" w... | closed | completed | false | 3 | [] | [] | 2023-04-27T10:02:24Z | 2023-04-29T12:20:46Z | 2023-04-29T12:20:46Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,686,884,432 | I_kwDOHa8MBc5ki9BQ | 3,263 | https://github.com/huggingface/diffusers/issues/3263 | https://api.github.com/repos/huggingface/diffusers/issues/3263 | DeepFloyd/IF-I-XL-v1.0 404 error | https://huggingface.co/DeepFloyd/IF-I-XL-v1.0
| closed | completed | false | 3 | [] | [] | 2023-04-27T13:42:19Z | 2023-04-27T22:28:56Z | 2023-04-27T22:28:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rupeshs | 3,255,994 | MDQ6VXNlcjMyNTU5OTQ= | User | false |
huggingface/diffusers | 1,687,047,434 | I_kwDOHa8MBc5kjk0K | 3,264 | https://github.com/huggingface/diffusers/issues/3264 | https://api.github.com/repos/huggingface/diffusers/issues/3264 | Converting Vae ckpt to diffusers | Hello,
I have been trying to finetune the VAE and since there does not seem to be a example for this using the diffusers library so far, i have reverted to the original latent diffusion github (https://github.com/CompVis/latent-diffusion). My assumption so far is that this one is not really differ in any significant... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-04-27T15:03:14Z | 2023-07-18T09:06:38Z | 2023-06-05T15:03:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | flixmk | 66,331,106 | MDQ6VXNlcjY2MzMxMTA2 | User | false |
huggingface/diffusers | 1,687,614,616 | I_kwDOHa8MBc5klvSY | 3,267 | https://github.com/huggingface/diffusers/issues/3267 | https://api.github.com/repos/huggingface/diffusers/issues/3267 | How to invert image into noise using class diffusers.DDIMInverseScheduler? | I wonder how to use class `diffusers.DDIMInverseScheduler` to invert image into image noise? I did not see an example of how to use it. Thanks
| closed | completed | false | 10 | [
"stale"
] | [] | 2023-04-27T22:05:56Z | 2024-01-21T16:57:25Z | 2023-06-05T15:03:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | g-jing | 44,223,191 | MDQ6VXNlcjQ0MjIzMTkx | User | false |
huggingface/diffusers | 1,687,797,279 | I_kwDOHa8MBc5kmb4f | 3,269 | https://github.com/huggingface/diffusers/issues/3269 | https://api.github.com/repos/huggingface/diffusers/issues/3269 | Is there any way to enhance the prompt rules for Diffuser? | Hi, I was wondering if there is a way to improve the prompt rules for Diffuser in order to generate embedding vectors similar to those of A111 web UI?
| closed | completed | false | 7 | [
"stale"
] | [] | 2023-04-28T02:31:46Z | 2023-06-05T15:03:04Z | 2023-06-05T15:03:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sanbuphy | 96,160,062 | U_kgDOBbtJPg | User | false |
huggingface/diffusers | 1,687,816,955 | I_kwDOHa8MBc5kmgr7 | 3,270 | https://github.com/huggingface/diffusers/issues/3270 | https://api.github.com/repos/huggingface/diffusers/issues/3270 | Is there a way to run train_text_to_image.py with training text_encoder? | null | closed | completed | false | 3 | [] | [] | 2023-04-28T02:55:11Z | 2023-04-28T05:26:52Z | 2023-04-28T05:26:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DogsHeadZ | 36,128,818 | MDQ6VXNlcjM2MTI4ODE4 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.