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 | 1,554,918,523 | I_kwDOHa8MBc5crix7 | 2,089 | https://github.com/huggingface/diffusers/issues/2089 | https://api.github.com/repos/huggingface/diffusers/issues/2089 | CUDA out of memory while training DreamBooth using AltDiffusion | ### Describe the bug
Hi I'm running Dreambooth training code. It works good with default stable diffusion v1.4 command. https://github.com/huggingface/diffusers/blob/fc8afa3ab5eb840ab0da5aadb629bf671eef9a39/examples/dreambooth/README.md?plain=1#L103.
However when I try AltDiffuision, it reports cuda out of memory er... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-01-24T12:44:57Z | 2023-01-26T02:58:22Z | 2023-01-26T02:58:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | airkid | 8,197,569 | MDQ6VXNlcjgxOTc1Njk= | User | false |
huggingface/diffusers | 1,555,228,088 | I_kwDOHa8MBc5csuW4 | 2,091 | https://github.com/huggingface/diffusers/issues/2091 | https://api.github.com/repos/huggingface/diffusers/issues/2091 | Issue when resuming training from checkpoint with mullti-gpu pipeline `train_text_to_image.py` | ### Describe the bug
Currently, there seems to be an issue when resuming training from a checkpoint using the `--resume_from_checkpoint` argument:
`FileNotFoundError: [Errno 2] No such file or directory: 'output_model/checkpoint-10/random_states_1.pkl`
I am using the same multi-gpu configuration for the initi... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-01-24T15:50:17Z | 2023-03-04T15:03:45Z | 2023-03-04T15:03:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | philippe-ml6 | 115,533,022 | U_kgDOBuLk3g | User | false |
huggingface/diffusers | 1,555,604,078 | I_kwDOHa8MBc5cuKJu | 2,093 | https://github.com/huggingface/diffusers/issues/2093 | https://api.github.com/repos/huggingface/diffusers/issues/2093 | Image-to-Image Diffusion Models | Dear all,
Thanks for the awesome work!
Is there any tutorial about how to train Image-to-Image Diffusion Models (without text guidance)? E.g. palette https://iterative-refinement.github.io/palette/ | closed | completed | false | 2 | [] | [] | 2023-01-24T20:12:28Z | 2023-01-29T00:23:57Z | 2023-01-29T00:23:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JunMa11 | 19,947,331 | MDQ6VXNlcjE5OTQ3MzMx | User | false |
huggingface/diffusers | 1,555,959,851 | I_kwDOHa8MBc5cvhAr | 2,097 | https://github.com/huggingface/diffusers/issues/2097 | https://api.github.com/repos/huggingface/diffusers/issues/2097 | Problem using UNet1D | ### Describe the bug
The UNet1D model is giving channel shape mismatch errors for me during the upscaling phase. This seems to be caused in part by DownResnetBlock1D setting all output_states prior to the downsample step whereas DownBlock2D does the last one after the downsampling step. Changing this was not enough ... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-01-25T02:20:25Z | 2023-03-31T06:52:52Z | 2023-01-26T04:14:57Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lawfordp2017 | 24,662,465 | MDQ6VXNlcjI0NjYyNDY1 | User | false |
huggingface/diffusers | 1,555,993,598 | I_kwDOHa8MBc5cvpP- | 2,098 | https://github.com/huggingface/diffusers/issues/2098 | https://api.github.com/repos/huggingface/diffusers/issues/2098 | Call to Torch.randn throws: "randn() received an invalid combination of arguments" | ### Describe the bug
When calling the `__call__` method of the stable diffusion pipeline, with a list of generators. (works without list of generators)
An internal call in the diffusers package to torch.randn throws an exception regarding in compatible function stubs.
I followed these docs
https://huggingface... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-01-25T03:13:47Z | 2023-04-06T19:53:31Z | 2023-03-04T15:03:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NicholasKao1029 | 45,542,006 | MDQ6VXNlcjQ1NTQyMDA2 | User | false |
huggingface/diffusers | 1,556,017,725 | I_kwDOHa8MBc5cvvI9 | 2,099 | https://github.com/huggingface/diffusers/issues/2099 | https://api.github.com/repos/huggingface/diffusers/issues/2099 | dataloader in text_to_image example | Hi,
Thanks for sharing this out-of-the-box example for text-to-image generation.
https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py
I followed the guidance to prepare the dataset but the `load_dataset` cannot load the text information.
https://huggingface.co/doc... | closed | completed | false | 6 | [] | [
"williamberman",
"patil-suraj"
] | 2023-01-25T04:00:32Z | 2023-01-27T09:36:59Z | 2023-01-27T01:42:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JunMa11 | 19,947,331 | MDQ6VXNlcjE5OTQ3MzMx | User | false |
huggingface/diffusers | 1,556,127,188 | I_kwDOHa8MBc5cwJ3U | 2,100 | https://github.com/huggingface/diffusers/issues/2100 | https://api.github.com/repos/huggingface/diffusers/issues/2100 | ImportError: cannot import name 'UnCLIPImageVariationPipeline' from 'diffusers.pipelines.unclip' | ### Describe the bug
```py
┌─────────────────────────────── Traceback (most recent call last) ────────────────────────────────┐
│ C:\Users\PC\Desktop\sd\envs\sd\lib\runpy.py:196 in _run_module_as_main │
│ ... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-01-25T06:33:37Z | 2023-01-26T14:41:24Z | 2023-01-26T14:41:23Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | camenduru | 54,370,274 | MDQ6VXNlcjU0MzcwMjc0 | User | false |
huggingface/diffusers | 1,556,518,122 | I_kwDOHa8MBc5cxpTq | 2,105 | https://github.com/huggingface/diffusers/issues/2105 | https://api.github.com/repos/huggingface/diffusers/issues/2105 | How can I use `*.safetensors` modles in `diffusers` library? | I found a really great model a few days ago. Then I wanted to use it on a pure API server (without webui or other Gradio interface)
Here is the model's path in huggingface: [WarriorMama777/OrangeMixs](https://huggingface.co/WarriorMama777/OrangeMixs)
And I mainly refer to these two demos:
[conditional_image_genera... | closed | completed | false | 3 | [] | [] | 2023-01-25T11:52:40Z | 2023-01-26T18:46:45Z | 2023-01-26T18:46:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Yiximail | 15,323,975 | MDQ6VXNlcjE1MzIzOTc1 | User | false |
huggingface/diffusers | 1,556,672,122 | I_kwDOHa8MBc5cyO56 | 2,107 | https://github.com/huggingface/diffusers/issues/2107 | https://api.github.com/repos/huggingface/diffusers/issues/2107 | Image resolution lower than 512x512 results in glitchy or black images | ### Describe the bug
For even the most common prompt "a photo of an astronaut riding a horse on mars", any image resolution lower than 512x512 either outputs a glitchy image or a complete black (stating it's NSFW). Tried "iphone from 2050" which again generates similar result.
[potentialBug_diffuser_imageResolution.z... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-01-25T13:36:40Z | 2023-03-04T15:03:42Z | 2023-03-04T15:03:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | singh-robbie | 118,924,089 | U_kgDOBxajOQ | User | false |
huggingface/diffusers | 1,556,947,735 | I_kwDOHa8MBc5czSMX | 2,108 | https://github.com/huggingface/diffusers/issues/2108 | https://api.github.com/repos/huggingface/diffusers/issues/2108 | Does img2img and txt2img shares same weights of textual encoder? | Hi, thanks for your great work.
Does text guided img2img and txt2img shares same textual encoder for conditioning?
I trained stable diffusion on dream booth which fine-tunes txt2img. Does the fine-tune affect text guided img2img's weight, too? | closed | completed | false | 2 | [] | [] | 2023-01-25T16:34:52Z | 2023-01-27T09:27:08Z | 2023-01-27T09:27:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YJ-142150 | 62,554,000 | MDQ6VXNlcjYyNTU0MDAw | User | false |
huggingface/diffusers | 1,343,077,647 | I_kwDOHa8MBc5QDb0P | 211 | https://github.com/huggingface/diffusers/issues/211 | https://api.github.com/repos/huggingface/diffusers/issues/211 | Exception raised in `LDMTextToImagePipeline` | ### Describe the bug
Invoking a LDMTextToImagePipeline loaded from model `CompVis/ldm-text2im-large-256` results in an exception, in the current `main` branch.
I haven't checked the code yet, maybe I'm using the wrong model.
### Reproduction
```
from diffusers import LDMTextToImagePipeline
ldm = LDMTextTo... | closed | completed | false | 2 | [
"bug"
] | [] | 2022-08-18T13:07:52Z | 2022-08-18T15:55:16Z | 2022-08-18T15:55:16Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,557,520,436 | I_kwDOHa8MBc5c1eA0 | 2,113 | https://github.com/huggingface/diffusers/issues/2113 | https://api.github.com/repos/huggingface/diffusers/issues/2113 | sample_dpmpp_sde | Looking to see if there is still appetite to build out `sample_dpmpp_sde` sampler.
_Originally posted by @patrickvonplaten in https://github.com/huggingface/diffusers/issues/277#issuecomment-1371428672_
Builidng this out will help close the gap on feature parity with auto1111's repo!It's become one of the "use a... | closed | completed | false | 4 | [
"stale"
] | [
"patrickvonplaten"
] | 2023-01-26T01:59:57Z | 2023-03-05T00:20:03Z | 2023-03-05T00:20:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NicholasKao1029 | 45,542,006 | MDQ6VXNlcjQ1NTQyMDA2 | User | false |
huggingface/diffusers | 1,557,692,989 | I_kwDOHa8MBc5c2II9 | 2,115 | https://github.com/huggingface/diffusers/issues/2115 | https://api.github.com/repos/huggingface/diffusers/issues/2115 | Why is there a `permute(2, 0, 1)` in textual_inversion example | Hi @patil-suraj, thanks for the example on textual inversion!
While I was looking at the code, I struggle to understand this line:
```python
example["pixel_values"] = torch.from_numpy(image).permute(2, 0, 1)
```
in textual_inversion.py
The image is already in RGB format, this permute effectively convert the... | closed | completed | false | 3 | [] | [] | 2023-01-26T06:40:37Z | 2023-01-27T06:00:45Z | 2023-01-27T06:00:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | taoisu | 30,015,018 | MDQ6VXNlcjMwMDE1MDE4 | User | false |
huggingface/diffusers | 1,557,870,828 | I_kwDOHa8MBc5c2zjs | 2,117 | https://github.com/huggingface/diffusers/issues/2117 | https://api.github.com/repos/huggingface/diffusers/issues/2117 | How can I change alpha of LoRA? | I'm not that familiar with the implementation detail of LoRA and wondering how I can change alpha in inference time.
I found a field scale here:
https://github.com/huggingface/diffusers/blob/085613733791b4d67be058edcd0bd23da5c3f57c/src/diffusers/models/cross_attention.py#L260
so I though I can change it by implement... | closed | completed | false | 3 | [] | [] | 2023-01-26T09:46:53Z | 2023-01-27T01:17:53Z | 2023-01-27T01:17:52Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hysts | 25,161,192 | MDQ6VXNlcjI1MTYxMTky | User | false |
huggingface/diffusers | 1,558,374,780 | I_kwDOHa8MBc5c4ul8 | 2,121 | https://github.com/huggingface/diffusers/issues/2121 | https://api.github.com/repos/huggingface/diffusers/issues/2121 | [Community] Implement `prompt-to-prompt` pipelines | **Describe the solution you'd like**
Now that we have an official way to tweak cross attention https://github.com/huggingface/diffusers/pull/1639 , would be great to have a pipeline (be it official or community) for `prompt-to-prompt` and further implementations of the technique (such as EDICT).
**Describe alterna... | closed | completed | false | 44 | [
"good first issue",
"New pipeline/model"
] | [
"patrickvonplaten"
] | 2023-01-26T15:58:53Z | 2024-11-24T03:27:51Z | 2024-11-24T03:27:17Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 1,558,552,505 | I_kwDOHa8MBc5c5Z-5 | 2,124 | https://github.com/huggingface/diffusers/issues/2124 | https://api.github.com/repos/huggingface/diffusers/issues/2124 | Incompatibility between LoRA attn_procs and other attn_procs, and other LoRA inference issues | ### Describe the bug
Why this matters:
The current implementation of attn_procs isn't really compatible with changing/switching attn_procs.
IMO, this would be quite important to add, since the lightweight nature of LoRA layers makes it an ideal candidate to rapidly switch between different "models" on a deploymen... | closed | completed | false | 7 | [
"bug",
"stale"
] | [
"patrickvonplaten"
] | 2023-01-26T18:05:44Z | 2023-03-06T15:03:32Z | 2023-03-06T15:03:32Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jorgemcgomes | 3,987,574 | MDQ6VXNlcjM5ODc1NzQ= | User | false |
huggingface/diffusers | 1,558,981,574 | I_kwDOHa8MBc5c7CvG | 2,132 | https://github.com/huggingface/diffusers/issues/2132 | https://api.github.com/repos/huggingface/diffusers/issues/2132 | example/dreambooth/train_dreambooth_lora.py - AttributeError: 'Accelerator' object has no attribute 'trackers' | ### Describe the bug
Using this format: https://github.com/huggingface/diffusers/tree/main/examples/dreambooth#training
`AttributeError: 'Accelerator' object has no attribute 'trackers'`
Getting an error on this line: https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora.... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-01-27T00:34:24Z | 2023-04-17T12:12:13Z | 2023-02-03T15:50:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mathyouf | 19,615,875 | MDQ6VXNlcjE5NjE1ODc1 | User | false |
huggingface/diffusers | 1,559,102,588 | I_kwDOHa8MBc5c7gR8 | 2,133 | https://github.com/huggingface/diffusers/issues/2133 | https://api.github.com/repos/huggingface/diffusers/issues/2133 | VRAM usage of inference | What factors influence the VRAM usage?
I used to thnk it is in linear with the number of pixels of the generated images. But when the image size grows, VRAW usage grows faster. It seems that space complexity is not just O(n). (Supposing that n is the number of pixels of the generated images).
Resolution VRAM... | closed | completed | false | 2 | [] | [] | 2023-01-27T03:17:22Z | 2023-01-31T09:32:02Z | 2023-01-31T09:32:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chesha1 | 49,867,392 | MDQ6VXNlcjQ5ODY3Mzky | User | false |
huggingface/diffusers | 1,559,152,730 | I_kwDOHa8MBc5c7sha | 2,135 | https://github.com/huggingface/diffusers/issues/2135 | https://api.github.com/repos/huggingface/diffusers/issues/2135 | Adding out-of-the-box support for multilingual models like BAAI/AltDiffusion to diffusers | ## Description of the problem
I'm unable to load multilingual models like [BAAI/AltDiffusion](https://huggingface.co/BAAI/AltDiffusion/tree/main) with `diffusers`.
## The solution I'd like
I would like `diffusers` to have out-of-the-box support for models like [BAAI/AltDiffusion](https://huggingface.co/BAAI/Al... | closed | completed | false | 15 | [
"good first issue"
] | [] | 2023-01-27T04:28:47Z | 2024-07-31T09:37:51Z | 2024-07-31T09:37:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jslegers | 883,647 | MDQ6VXNlcjg4MzY0Nw== | User | false |
huggingface/diffusers | 1,559,161,383 | I_kwDOHa8MBc5c7uon | 2,136 | https://github.com/huggingface/diffusers/issues/2136 | https://api.github.com/repos/huggingface/diffusers/issues/2136 | Overcoming the 77 token limit in diffusers | ## Description of the problem
CLIP has a 77 token limit, which is much too small for many prompts.
Several GUIs have found a way to overcome this limit, but not the `diffusers` library.
## The solution I'd like
I would like `diffusers` to be able to run longer prompts and overcome the 77 token limit of CLI... | closed | completed | false | 29 | [
"stale"
] | [] | 2023-01-27T04:38:41Z | 2024-10-09T19:58:49Z | 2023-03-06T15:03:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jslegers | 883,647 | MDQ6VXNlcjg4MzY0Nw== | User | false |
huggingface/diffusers | 1,560,183,846 | I_kwDOHa8MBc5c_oQm | 2,142 | https://github.com/huggingface/diffusers/issues/2142 | https://api.github.com/repos/huggingface/diffusers/issues/2142 | SD 2.1 Dreambooth Trained Model Generating Brown Blurry Images | I just completed my first training of SD 2.1 (768) but all outputs are completely blurred. Generations start with lots of noise in the preview window and always resolve as brown blurry images, with the token or without it. Has anyone here encountered this issue?
I have the 768 .yaml file in place. I have tried with ... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-01-27T18:04:51Z | 2023-04-26T14:11:12Z | 2023-03-07T15:03:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sebalon80 | 116,579,894 | U_kgDOBvLeNg | User | false |
huggingface/diffusers | 1,560,642,337 | I_kwDOHa8MBc5dBYMh | 2,145 | https://github.com/huggingface/diffusers/issues/2145 | https://api.github.com/repos/huggingface/diffusers/issues/2145 | How to synthesize the files trained by ‘train_dreambooth_lora.py’ into ckpt files? | Hello, admirable developer
I used the ‘train_dreambooth_lora.py’ method to train and got the ‘. bin’ file. I need to convert it to ckpt for use in webui,How to convert the bin file to the ckpt file
| closed | completed | false | 4 | [
"stale"
] | [] | 2023-01-28T02:16:34Z | 2023-03-07T15:03:35Z | 2023-03-07T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kkiqq22 | 20,888,251 | MDQ6VXNlcjIwODg4MjUx | User | false |
huggingface/diffusers | 1,560,852,759 | I_kwDOHa8MBc5dCLkX | 2,148 | https://github.com/huggingface/diffusers/issues/2148 | https://api.github.com/repos/huggingface/diffusers/issues/2148 | Is it possible to retrain again dream booth's result? | Hi, thank for your great work.
I trained stable-diffusion-2-1 four times in a row. Like,
```
export MODEL_NAME="stabilityai/stable-diffusion-2-1"
export INSTANCE_DIR="teddy" --prompt "xyz"
export OUTPUT_DIR="save_model"
accelerate launch train_dreambooth.py
export MODEL_NAME="save_model"
export INSTANCE_DIR="a... | closed | completed | false | 4 | [] | [
"patil-suraj"
] | 2023-01-28T13:11:09Z | 2023-01-31T13:30:50Z | 2023-01-31T13:30:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YJ-142150 | 62,554,000 | MDQ6VXNlcjYyNTU0MDAw | User | false |
huggingface/diffusers | 1,560,918,323 | I_kwDOHa8MBc5dCbkz | 2,149 | https://github.com/huggingface/diffusers/issues/2149 | https://api.github.com/repos/huggingface/diffusers/issues/2149 | [Community Pipeline] Support a texture painting workflow | **Is your feature request related to a problem? Please describe.**
The current texture projection setup simply replaces the material on selected faces with a new one, which makes it hard to blend between two textures or only apply specific features of a new texture to an existing one.
**Describe the solution you'd ... | closed | completed | false | 3 | [
"community-examples"
] | [] | 2023-01-28T16:19:22Z | 2023-02-03T15:54:21Z | 2023-01-31T13:15:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dekker3D | 789,100 | MDQ6VXNlcjc4OTEwMA== | User | false |
huggingface/diffusers | 1,343,831,949 | I_kwDOHa8MBc5QGT-N | 215 | https://github.com/huggingface/diffusers/issues/215 | https://api.github.com/repos/huggingface/diffusers/issues/215 | "Killed" without error during "from_pretrained" | ### Describe the bug
While running:
```
DiffusionPipeline.from_pretrained("CompVis/ldm-text2im-large-256"
```
No error at all just "Killed" gets outputted and then python is exited
### Reproduction
_No response_
### Logs
_No response_
### System Info
```shell
python3.9
windows 11 64-bit wsl/ubuntu
NVCC V... | closed | completed | false | 3 | [
"stale"
] | [] | 2022-08-19T01:55:46Z | 2022-10-10T13:01:02Z | 2022-10-10T13:01:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kennethgoodman | 11,861,823 | MDQ6VXNlcjExODYxODIz | User | false |
huggingface/diffusers | 1,560,919,566 | I_kwDOHa8MBc5dCb4O | 2,150 | https://github.com/huggingface/diffusers/issues/2150 | https://api.github.com/repos/huggingface/diffusers/issues/2150 | Make texture projection work with img2img models | **Is your feature request related to a problem? Please describe.**
Right now, texture projection only works with depth2img models.
**Describe the solution you'd like**
An img2img model can do some pretty decent work for texture projection, if you give it an input image with unshaded basic textures... if no such te... | closed | completed | false | 2 | [] | [] | 2023-01-28T16:23:08Z | 2023-01-31T13:15:58Z | 2023-01-31T13:15:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dekker3D | 789,100 | MDQ6VXNlcjc4OTEwMA== | User | false |
huggingface/diffusers | 1,561,212,670 | I_kwDOHa8MBc5dDjb- | 2,152 | https://github.com/huggingface/diffusers/issues/2152 | https://api.github.com/repos/huggingface/diffusers/issues/2152 | AutoEncoderKL | Hello,
AutoencoderKL.latent_channels is 4 by default, how does one interpret 4 channels ? How was this choice motivated? I know we are talking about a low meaningful latent space but I'm trying to understand a bit more about it. Also, are there pre-trained 1 or 3 channel AutoEncoderKL's?
Thanks in advance | closed | completed | false | 2 | [
"stale"
] | [] | 2023-01-29T09:52:16Z | 2023-03-08T15:03:30Z | 2023-03-08T15:03:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Kageyoshi317 | 122,362,383 | U_kgDOB0saDw | User | false |
huggingface/diffusers | 1,561,219,465 | I_kwDOHa8MBc5dDlGJ | 2,153 | https://github.com/huggingface/diffusers/issues/2153 | https://api.github.com/repos/huggingface/diffusers/issues/2153 | Error running `float16` in GTX 1660 | Newbie here. I am interested in stable diffusion and wanted to practice and see how it performs. I came across https://github.com/huggingface/diffusers/tree/main#new--stable-diffusion-is-now-fully-compatible-with-diffusers which I followed the code:
```
import torch
from diffusers import StableDiffusionPipeline
p... | closed | completed | false | 12 | [] | [] | 2023-01-29T10:12:09Z | 2024-02-06T11:01:38Z | 2023-01-29T14:00:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cmmalana | 123,867,749 | U_kgDOB2ISZQ | User | false |
huggingface/diffusers | 1,561,334,431 | I_kwDOHa8MBc5dEBKf | 2,155 | https://github.com/huggingface/diffusers/issues/2155 | https://api.github.com/repos/huggingface/diffusers/issues/2155 | latest release breaks Flax pipes for Stable Diffusion when using bfloat16 and/or non-default scheduler | ### Describe the bug
using Colab TPU environment, and bfloat16, blows up pipes if using anything but a packaged-built pipeline and its built-in default scheduler, even then it can unpredictably produce NaNs instead of images (can't reliably replicate the NaN thing, but it happens more often than not - see logs/outpu... | closed | completed | false | 11 | [
"bug",
"stale"
] | [
"pcuenca"
] | 2023-01-29T15:23:19Z | 2023-03-28T07:17:22Z | 2023-03-11T15:02:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | krahnikblis | 84,637,076 | MDQ6VXNlcjg0NjM3MDc2 | User | false |
huggingface/diffusers | 1,561,379,657 | I_kwDOHa8MBc5dEMNJ | 2,156 | https://github.com/huggingface/diffusers/issues/2156 | https://api.github.com/repos/huggingface/diffusers/issues/2156 | Dreambooth example script only uses CPU for training instead of GPU | Hey team, I'm trying to execute the example dreambooth script to train, and when it initializes I see that it only uses the CPU as the device, instead of my GPU, which is a 4090. I checked the accelerate config and where I set `'use_cpu'`to `false`. I'm new to ML and Python in general, coming from a web / mobile eng b... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-01-29T17:15:43Z | 2023-03-10T15:04:10Z | 2023-03-10T15:04:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Hendrixer | 2,840,504 | MDQ6VXNlcjI4NDA1MDQ= | User | false |
huggingface/diffusers | 1,561,895,720 | I_kwDOHa8MBc5dGKMo | 2,159 | https://github.com/huggingface/diffusers/issues/2159 | https://api.github.com/repos/huggingface/diffusers/issues/2159 | clip_sample argument is unused during the initialisation of DDPMScheduler (scheduling_ddpm.py) | ### Describe the bug
File location: https://github.com/huggingface/diffusers/blob/main/src/diffusers/schedulers/scheduling_ddpm.py
* During the initialisation of DDPMScheduler, clip_sample argument is taken however, it is not being used in any place. Even if we set the clip_sample to False, it still clips the sampl... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-01-30T07:27:25Z | 2023-03-10T15:04:07Z | 2023-03-10T15:04:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KnHuq | 18,214,937 | MDQ6VXNlcjE4MjE0OTM3 | User | false |
huggingface/diffusers | 1,343,834,339 | I_kwDOHa8MBc5QGUjj | 216 | https://github.com/huggingface/diffusers/issues/216 | https://api.github.com/repos/huggingface/diffusers/issues/216 | Porting `img2img.py` from SD into the diffusers library | The [image-to-image script](https://github.com/CompVis/stable-diffusion/blob/main/scripts/img2img.py) in stable-diffusion is a very nice tool for modifying images.
Their implementation seems relatively straightforward, though they have only implemented it for the DDIM scheduler. I took a crack at porting it over int... | closed | completed | false | 4 | [] | [
"anton-l"
] | 2022-08-19T02:00:53Z | 2022-08-24T11:39:53Z | 2022-08-24T11:39:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | atarashansky | 16,548,075 | MDQ6VXNlcjE2NTQ4MDc1 | User | false |
huggingface/diffusers | 1,561,942,469 | I_kwDOHa8MBc5dGVnF | 2,160 | https://github.com/huggingface/diffusers/issues/2160 | https://api.github.com/repos/huggingface/diffusers/issues/2160 | Stable diffusion fine-tuning with LoRa error wth vae.config.scaling_factor | ### Describe the bug
I followed the readme, but it showed
AttributeError: 'FrozenDict' object has no attribute 'scaling_factor'
what is wrong with it..?
```
Traceback (most recent call last):
File "/home/user/gene/LoRa.py", line 840, in <module>
main()
File "/home/user/gene/LoRa.py", line 690, in m... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-01-30T08:10:08Z | 2023-01-31T03:24:54Z | 2023-01-31T03:24:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | treasuraid | 96,979,904 | U_kgDOBcfLwA | User | false |
huggingface/diffusers | 1,562,190,180 | I_kwDOHa8MBc5dHSFk | 2,161 | https://github.com/huggingface/diffusers/issues/2161 | https://api.github.com/repos/huggingface/diffusers/issues/2161 | LoRA training on an inpainting model. | ### Describe the bug
I want to train using lora+dreambooth to add a concept to an inpainting model and then use the in-painting pipeline for inference.
The train_dreambooth_lora.py gives the following error:
RuntimeError: Given groups=1, weight of size [320, 9, 3, 3], expected input[1, 4, 64, 64] to have 9 channel... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-01-30T10:47:05Z | 2023-03-02T18:18:49Z | 2023-03-02T18:18:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vitalyshalumov | 33,824,221 | MDQ6VXNlcjMzODI0MjIx | User | false |
huggingface/diffusers | 1,562,203,610 | I_kwDOHa8MBc5dHVXa | 2,162 | https://github.com/huggingface/diffusers/issues/2162 | https://api.github.com/repos/huggingface/diffusers/issues/2162 | Add an example on pivotal tuning | Pivotal tuning: https://arxiv.org/abs/2106.05744
Reference implementation: https://github.com/cloneofsimo/lora | closed | completed | false | 9 | [
"stale"
] | [] | 2023-01-30T10:56:09Z | 2023-06-13T15:03:40Z | 2023-06-13T15:03:40Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,562,291,886 | I_kwDOHa8MBc5dHq6u | 2,163 | https://github.com/huggingface/diffusers/issues/2163 | https://api.github.com/repos/huggingface/diffusers/issues/2163 | [examples] ]Error due to mismatches in the dtype of UNet while running `train_text_to_image.py` | ### Describe the bug
I am referring to this script: https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py.
To perform qualitative validation and following https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_lora.py, I have added ... | closed | completed | false | 18 | [
"bug",
"stale"
] | [
"patil-suraj"
] | 2023-01-30T11:48:43Z | 2023-03-13T05:28:15Z | 2023-03-13T05:28:14Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,562,521,949 | I_kwDOHa8MBc5dIjFd | 2,164 | https://github.com/huggingface/diffusers/issues/2164 | https://api.github.com/repos/huggingface/diffusers/issues/2164 | LDM Training | Hello,
I tried to adapt your example
"https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/training_example.ipynb#scrollTo=83158bea-273e-4088-a75a-e3c7cc86e0fc"
which generates buterflies to make it work in latent space with an
AutoencoderKL.from_pretrained("CompVis/stable-diffusion-... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-01-30T14:10:31Z | 2025-01-02T18:21:47Z | 2023-03-13T15:03:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Kageyoshi317 | 122,362,383 | U_kgDOB0saDw | User | false |
huggingface/diffusers | 1,562,600,874 | I_kwDOHa8MBc5dI2Wq | 2,165 | https://github.com/huggingface/diffusers/issues/2165 | https://api.github.com/repos/huggingface/diffusers/issues/2165 | Make prev_timestep of DDIMScheduler accessible | **Is your feature request related to a problem? Please describe.**
The DDIM-Scheduler works in theory with any subselection of the train timestep; the original authors propose for example a square schedule. However, the current implementation fixes the subselection to be a linear subselection. I would like to be able ... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-01-30T14:54:41Z | 2023-03-10T15:04:04Z | 2023-03-10T15:04:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jenuk | 22,528,125 | MDQ6VXNlcjIyNTI4MTI1 | User | false |
huggingface/diffusers | 1,563,444,029 | I_kwDOHa8MBc5dMEM9 | 2,167 | https://github.com/huggingface/diffusers/issues/2167 | https://api.github.com/repos/huggingface/diffusers/issues/2167 | Im using jupyter notebook and every time it stacks ckpt file but I don't know where it is | every time I try using diffusers, it downloads all .bin files and ckpt files but it piles up somewhere in the server.
i thought it got piled up in anaconda3/env but it wasn't.
where would it downloads the files be? my server its full of memory:(
![image](https://user-images.githubusercontent.com/82705312/2156... | closed | completed | false | 2 | [] | [] | 2023-01-31T00:58:18Z | 2023-02-02T02:51:10Z | 2023-02-02T02:51:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jakeyahn | 82,705,312 | MDQ6VXNlcjgyNzA1MzEy | User | false |
huggingface/diffusers | 1,563,765,208 | I_kwDOHa8MBc5dNSnY | 2,170 | https://github.com/huggingface/diffusers/issues/2170 | https://api.github.com/repos/huggingface/diffusers/issues/2170 | When converting "stabilityai/stable-diffusion-2-1" to onnx model, can it support opset version 13? | ### Describe the bug
For some reasons, I need to get the onnx model with opset version 13.But I encountered this problem during converting: "UnsupportedOperatorError: Exporting the operator ::triu to ONNX opset version 13 is not supported. Support for this operator was added in version 14, try exporting with this vers... | closed | completed | false | 3 | [
"bug",
"stale"
] | [
"anton-l",
"echarlaix"
] | 2023-01-31T06:58:47Z | 2023-03-10T15:04:02Z | 2023-03-10T15:04:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | labmanno004 | 87,518,566 | MDQ6VXNlcjg3NTE4NTY2 | User | false |
huggingface/diffusers | 1,564,251,908 | I_kwDOHa8MBc5dPJcE | 2,174 | https://github.com/huggingface/diffusers/issues/2174 | https://api.github.com/repos/huggingface/diffusers/issues/2174 | Make logging and saving both use steps in textual inversion | I'll make a pr for this today. There is an inconsitency where logging is done in epochs and saving is done in steps in the textual inversion script like so
```
if args.validation_prompt is not None and epoch % args.validation_epochs == 0:
```
for logging and
```
if global_step % args.save_steps == 0:
```
for sa... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-01-31T12:50:43Z | 2023-03-10T15:04:00Z | 2023-03-10T15:04:00Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | isamu-isozaki | 23,430,101 | MDQ6VXNlcjIzNDMwMTAx | User | false |
huggingface/diffusers | 1,564,592,829 | I_kwDOHa8MBc5dQcq9 | 2,177 | https://github.com/huggingface/diffusers/issues/2177 | https://api.github.com/repos/huggingface/diffusers/issues/2177 | Cannot load pretrained `AutoencoderKL` | ### Describe the bug
Despite `AutoencoderKL` "inheriting" from `ModelMixin`, attempting to load a pretrained model using `from_pretrained` seems to fail.
Currently, if I want to use the pretrained autoencoder alone, I need to load in the entire `DiffusionPipeline` into memory when I only need a small subset of it. ... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-01-31T16:01:30Z | 2023-01-31T18:26:33Z | 2023-01-31T18:26:33Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | vvvm23 | 44,398,246 | MDQ6VXNlcjQ0Mzk4MjQ2 | User | false |
huggingface/diffusers | 1,564,765,158 | I_kwDOHa8MBc5dRGvm | 2,179 | https://github.com/huggingface/diffusers/issues/2179 | https://api.github.com/repos/huggingface/diffusers/issues/2179 | `FlaxDDPMScheduler` produces black images | ### Describe the bug
`FlaxDDPMScheduler` produces black images, even after fixing #2176.
I checked in `bfloat16`, `bfloat32` and at revision f106ab40. It failed in all cases.
### Reproduction
```Python
import jax
import numpy as np
from flax.jax_utils import replicate
from flax.training.common_utils import s... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-01-31T17:54:20Z | 2023-05-03T15:26:51Z | 2023-03-12T15:03:17Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,564,768,909 | I_kwDOHa8MBc5dRHqN | 2,180 | https://github.com/huggingface/diffusers/issues/2180 | https://api.github.com/repos/huggingface/diffusers/issues/2180 | Make `FlaxLMSDiscreteScheduler` jittable | Please, see discussion in #2155. I was under the wrong assumption that `FlaxLMSDiscreteScheduler` had been made jittable as part of #1661, but it never was. | open | null | false | 8 | [
"good first issue"
] | [] | 2023-01-31T17:56:58Z | 2025-06-28T17:08:07Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,564,840,352 | I_kwDOHa8MBc5dRZGg | 2,182 | https://github.com/huggingface/diffusers/issues/2182 | https://api.github.com/repos/huggingface/diffusers/issues/2182 | Textual inversion logging doesn't see placeholder tokens | ### Describe the bug
Currently, the textual inversion script sets up pipelines as
```
pipeline = DiffusionPipeline.from_pretrained(
args.pretrained_model_name_or_path,
text_encoder=accelerator.unwrap_model(text_encoder),
revision=args.revision,
)
```
without the tokenizer. So when I was training on... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-01-31T18:56:46Z | 2023-02-03T17:17:26Z | 2023-02-03T17:17:25Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | isamu-isozaki | 23,430,101 | MDQ6VXNlcjIzNDMwMTAx | User | false |
huggingface/diffusers | 1,564,876,224 | I_kwDOHa8MBc5dRh3A | 2,185 | https://github.com/huggingface/diffusers/issues/2185 | https://api.github.com/repos/huggingface/diffusers/issues/2185 | config files missed for training naive text-to-image model | ### Describe the bug
I used the following command to train a naive text-to-image model and the training process goes well.
Here are files in the model folder
```bash
pytorch_model.bin optimizer.bin scaler.pt
random_states_0.pkl scheduler.bin
```
However, during inference, it raises errors on mi... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-01-31T19:23:30Z | 2023-05-06T11:18:21Z | 2023-03-12T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JunMa11 | 19,947,331 | MDQ6VXNlcjE5OTQ3MzMx | User | false |
huggingface/diffusers | 1,564,903,656 | I_kwDOHa8MBc5dRojo | 2,186 | https://github.com/huggingface/diffusers/issues/2186 | https://api.github.com/repos/huggingface/diffusers/issues/2186 | How to change default Adam optimizer to special version of Adam deepspeed.ops.adam.DeepSpeedCPUAdam? | I am having a 8GB VRAM GPU. I am using ShivamShrirao's diffuser to train. On his page, It is said that "Changing the default Adam optimizer to DeepSpeed's special version of Adam deepspeed.ops.adam.DeepSpeedCPUAdam gives a substantial speedup"
How to do that? and where can we get CUDA toolchain with the same version... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-01-31T19:48:25Z | 2023-03-17T15:03:43Z | 2023-03-17T15:03:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | runner22k | 48,834,280 | MDQ6VXNlcjQ4ODM0Mjgw | User | false |
huggingface/diffusers | 1,565,095,066 | I_kwDOHa8MBc5dSXSa | 2,189 | https://github.com/huggingface/diffusers/issues/2189 | https://api.github.com/repos/huggingface/diffusers/issues/2189 | Loading multiple LoRA weights during inference | During inference, is it possible to load multiple LoRA weights on top of the frozen model. Something like:
```
pipe = StableDiffusionPipeline.from_pretrained("CompVis/stable-diffusion-v1-4", torch_dtype=torch.float16)
pipe.unet.load_attn_procs("a.bin")
pipe.unet.load_attn_procs("b.bin")
pipe.to("cuda")
```
Thi... | closed | completed | false | 6 | [] | [] | 2023-01-31T22:32:12Z | 2023-02-07T21:40:55Z | 2023-02-07T21:40:55Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | asadm | 1,973,513 | MDQ6VXNlcjE5NzM1MTM= | User | false |
huggingface/diffusers | 1,565,287,976 | I_kwDOHa8MBc5dTGYo | 2,192 | https://github.com/huggingface/diffusers/issues/2192 | https://api.github.com/repos/huggingface/diffusers/issues/2192 | train_dreambooth_lora.py doesn't have --dataset_name and --caption_column arguments | So I'm trying to give it a try launching locally with an provided example from the announcements page:
```
accelerate launch --mixed_precision="fp16" train_text_to_image_lora.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--dataset_name=$DATASET_NAME --caption_column="text" \
--resolution=512 --random_... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-01T02:01:01Z | 2023-03-12T15:03:14Z | 2023-03-12T15:03:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Pirog17000 | 3,022,852 | MDQ6VXNlcjMwMjI4NTI= | User | false |
huggingface/diffusers | 1,565,537,598 | I_kwDOHa8MBc5dUDU- | 2,194 | https://github.com/huggingface/diffusers/issues/2194 | https://api.github.com/repos/huggingface/diffusers/issues/2194 | Shark | ### Model/Pipeline/Scheduler description
https://github.com/nod-ai/SHARK/tree/main/shark/examples/shark_inference/stable_diffusion
It's really fast into AMD (windows)
ONNX 50 steps 512x512: 2 min
Shark 50 steps 512x512: 15 sec
### Open source status
- [ ] The model implementation is available
- [ ] The model ... | closed | completed | false | 8 | [
"stale"
] | [] | 2023-02-01T06:47:15Z | 2023-03-03T15:08:31Z | 2023-03-03T15:08:31Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ForserX | 13,867,290 | MDQ6VXNlcjEzODY3Mjkw | User | false |
huggingface/diffusers | 1,565,641,851 | I_kwDOHa8MBc5dUcx7 | 2,195 | https://github.com/huggingface/diffusers/issues/2195 | https://api.github.com/repos/huggingface/diffusers/issues/2195 | Tune A Video | ### Model/Pipeline/Scheduler description
Tune a video's idea is to leverages Stable Diffusion to create short videos.
Code and weights are available here and already in the diffusers format:
https://github.com/showlab/Tune-A-Video
### Open source status
- [X] The model implementation is available
- [X] Th... | closed | completed | false | 1 | [
"New pipeline/model"
] | [
"patil-suraj"
] | 2023-02-01T08:13:12Z | 2023-03-20T03:30:20Z | 2023-03-20T03:30:20Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,565,645,254 | I_kwDOHa8MBc5dUdnG | 2,196 | https://github.com/huggingface/diffusers/issues/2196 | https://api.github.com/repos/huggingface/diffusers/issues/2196 | Attend and Excite | ### Model/Pipeline/Scheduler description
Based on Prompt-2-Prompt, Attend and Excite allows to generate better composite images by tweaking the text embeddings.
Code available: https://github.com/AttendAndExcite/Attend-and-Excite
### Open source status
- [X] The model implementation is available
- [X] The model ... | closed | completed | false | 10 | [
"New pipeline/model"
] | [
"yiyixuxu"
] | 2023-02-01T08:15:44Z | 2024-11-21T07:29:03Z | 2024-11-21T07:29:03Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,565,774,410 | I_kwDOHa8MBc5dU9JK | 2,199 | https://github.com/huggingface/diffusers/issues/2199 | https://api.github.com/repos/huggingface/diffusers/issues/2199 | How to load a custom fine-tuned embeds into the inpainting pipeline？ | Hi, thanks for the great work.
I want to know if there is any way to load the new concept obtained by fine-tuned (whether it is textual inversion, dreambooth or others) into the inpainting model? As shown in the textual inversion paper: replace the iphone in the hands of Steve Jobs with <cat-toy>.
I only saw exa... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-01T09:40:32Z | 2023-03-12T15:03:11Z | 2023-03-12T15:03:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anonymoussss | 12,672,787 | MDQ6VXNlcjEyNjcyNzg3 | User | false |
huggingface/diffusers | 1,566,120,730 | I_kwDOHa8MBc5dWRsa | 2,201 | https://github.com/huggingface/diffusers/issues/2201 | https://api.github.com/repos/huggingface/diffusers/issues/2201 | Padding difference in Stable Diffusion tokenizers: 1.x vs 2.x vs OpenCLIP | ### Describe the bug
There is a difference between the tokenizers in SD 1.x and 2.x, which seems to be introduced by a typo/bug. The consequence is that 1.x is padding with EOS (49407), while 2.x is padding with ! (0).
I know these two models use different CLIP models, but the tokenizer should be the same.
If we... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-02-01T13:29:50Z | 2023-03-12T15:03:10Z | 2023-03-12T15:03:10Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jorgemcgomes | 3,987,574 | MDQ6VXNlcjM5ODc1NzQ= | User | false |
huggingface/diffusers | 1,566,133,302 | I_kwDOHa8MBc5dWUw2 | 2,202 | https://github.com/huggingface/diffusers/issues/2202 | https://api.github.com/repos/huggingface/diffusers/issues/2202 | Is Openvino pipeline support in progress ? | ### Model/Pipeline/Scheduler description
1. Planning to support **Openvino pipeline** with converted **IR models** from stable-diffusion with model optimizer. Is the development for same in-progress ?
2. And also currently there is support only for **CPUExecutionProvide**r and there is no support to use **Openvino ... | closed | completed | false | 7 | [] | [] | 2023-02-01T13:38:08Z | 2023-03-03T15:12:46Z | 2023-02-14T07:25:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lalith-mcw | 84,698,958 | MDQ6VXNlcjg0Njk4OTU4 | User | false |
huggingface/diffusers | 1,566,223,530 | I_kwDOHa8MBc5dWqyq | 2,203 | https://github.com/huggingface/diffusers/issues/2203 | https://api.github.com/repos/huggingface/diffusers/issues/2203 | Why $MODEL_NAME is downloaded to my Home directory instead of dreambooth directory? | ### System Info
```Shell
- `Accelerate` version: 0.16.0
- Platform: Linux-5.15.0-58-generic-x86_64-with-glibc2.31
- Python version: 3.10.9
- Numpy version: 1.23.5
- PyTorch version (GPU?): 1.12.1 (True)
- `Accelerate` default config:
- compute_environment: LOCAL_MACHINE
- distributed_type: DEEPSPEED
- mi... | closed | completed | false | 3 | [] | [] | 2023-01-31T22:57:19Z | 2023-02-11T18:01:59Z | 2023-02-11T18:00:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | runner22k | 48,834,280 | MDQ6VXNlcjQ4ODM0Mjgw | User | false |
huggingface/diffusers | 1,345,057,211 | I_kwDOHa8MBc5QK_G7 | 221 | https://github.com/huggingface/diffusers/issues/221 | https://api.github.com/repos/huggingface/diffusers/issues/221 | Add safety check usage in the readme | It seems like a safety check is now required for stable diffusion. Could update the readme with a proper example on how to establish the pipeline? The current example is now outdated | closed | completed | false | 2 | [] | [] | 2022-08-20T05:01:48Z | 2022-08-22T20:09:07Z | 2022-08-22T20:09:07Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dblunk88 | 39,381,389 | MDQ6VXNlcjM5MzgxMzg5 | User | false |
huggingface/diffusers | 1,566,576,602 | I_kwDOHa8MBc5dYA_a | 2,210 | https://github.com/huggingface/diffusers/issues/2210 | https://api.github.com/repos/huggingface/diffusers/issues/2210 | Allow for LoRA to train the text encoder as well | **Is your feature request related to a problem? Please describe.**
The current LoRA training code only trains the unet
**Describe the solution you'd like**
Also train the text encoder, this is already possible with the non-LoRA Dreambooth, also popular diffusers based implementations of LoRA such as `cloneofsimo/l... | closed | completed | false | 2 | [] | [] | 2023-02-01T18:10:01Z | 2023-02-02T09:17:57Z | 2023-02-02T09:17:56Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 1,566,832,855 | I_kwDOHa8MBc5dY_jX | 2,212 | https://github.com/huggingface/diffusers/issues/2212 | https://api.github.com/repos/huggingface/diffusers/issues/2212 | Question on how to train a stable diffusion image variation model | Hello, I'm trying to come up with an image variation model, where the input would be an unstained pathology image and the model would output a stained version of the pathology image. Both input and output are png images of same size. The images look something like [this](https://www.google.com/search?sxsrf=AJOqlzWGwEsK... | closed | completed | false | 13 | [
"community-examples"
] | [] | 2023-02-01T21:08:15Z | 2023-03-08T08:54:54Z | 2023-02-20T20:02:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chokevin8 | 66,405,082 | MDQ6VXNlcjY2NDA1MDgy | User | false |
huggingface/diffusers | 1,566,969,138 | I_kwDOHa8MBc5dZg0y | 2,213 | https://github.com/huggingface/diffusers/issues/2213 | https://api.github.com/repos/huggingface/diffusers/issues/2213 | train_text_to_image No Distributed environment, code hangs on TPU v4-8 | ### Describe the bug
I am trying to run `train_text_to_image.py` on TPU v4-8 but when I start training, I see
```
02/01/2023 21:12:22 - INFO - train_text_to_image - Distributed environment: NO
Num processes: 1
Process index: 0
Local process index: 0
Device: cpu
```
So it doesn't seem to recognize tpu and r... | closed | completed | false | 12 | [
"bug",
"stale"
] | [
"williamberman",
"patil-suraj"
] | 2023-02-01T22:47:37Z | 2023-04-09T15:03:26Z | 2023-04-09T15:03:26Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ssusie | 8,882,373 | MDQ6VXNlcjg4ODIzNzM= | User | false |
huggingface/diffusers | 1,566,996,965 | I_kwDOHa8MBc5dZnnl | 2,214 | https://github.com/huggingface/diffusers/issues/2214 | https://api.github.com/repos/huggingface/diffusers/issues/2214 | NSFW filter broken on 2.1 | ### Describe the bug
stabilityai/stable-diffusion-2-1-base evades the NSFW filter
### Reproduction
Using
`pipe = StableDiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-2-1-base", torch_dtype=torch.float16)`
And generating images completely avoids the NSFW filter.
My understanding is that... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-02-01T23:10:28Z | 2023-03-13T15:03:46Z | 2023-03-13T15:03:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nbardy | 1,278,972 | MDQ6VXNlcjEyNzg5NzI= | User | false |
huggingface/diffusers | 1,567,367,043 | I_kwDOHa8MBc5dbB-D | 2,218 | https://github.com/huggingface/diffusers/issues/2218 | https://api.github.com/repos/huggingface/diffusers/issues/2218 | Getting overtrained models when training with diffusers | ### Describe the bug
I've been trying to train a model using diffusers with this base model https://huggingface.co/wavymulder/Analog-Diffusion, but I've been getting very overtrained models, even on a low amount of steps.
These are the parameters I'm passing in
--pretrained_model_name_or_path="/home/ubuntu/model... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-02-02T05:51:38Z | 2023-03-31T15:03:36Z | 2023-03-31T15:03:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | monis24 | 52,933,919 | MDQ6VXNlcjUyOTMzOTE5 | User | false |
huggingface/diffusers | 1,567,579,114 | I_kwDOHa8MBc5db1vq | 2,219 | https://github.com/huggingface/diffusers/issues/2219 | https://api.github.com/repos/huggingface/diffusers/issues/2219 | how to fine-tune the stable diffusion super resolution model | Thank you for you amazing work!
Could you please tell me how to finetune the stable diffusion super resolution model ? I don't know how to do that | closed | completed | false | 3 | [
"stale"
] | [] | 2023-02-02T08:40:17Z | 2023-05-18T07:53:00Z | 2023-03-13T15:03:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | liushanyuan18 | 46,879,442 | MDQ6VXNlcjQ2ODc5NDQy | User | false |
huggingface/diffusers | 1,345,238,387 | I_kwDOHa8MBc5QLrVz | 222 | https://github.com/huggingface/diffusers/issues/222 | https://api.github.com/repos/huggingface/diffusers/issues/222 | Training not working because of RuntimeError | ### Describe the bug
Hi,
unfortunately, I could not run the training due to an error in the scheduler.
Below you will find the error log.
Best,
Tristan
### Reproduction
```
accelerate launch train_unconditional.py \
--dataset_name="huggan/flowers-102-categories" \
--resolution=64 \
--output_dir=... | closed | completed | false | 4 | [
"bug"
] | [] | 2022-08-20T17:35:35Z | 2022-08-21T07:03:00Z | 2022-08-21T07:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AI-Guru | 32,195,399 | MDQ6VXNlcjMyMTk1Mzk5 | User | false |
huggingface/diffusers | 1,567,805,312 | I_kwDOHa8MBc5dcs-A | 2,220 | https://github.com/huggingface/diffusers/issues/2220 | https://api.github.com/repos/huggingface/diffusers/issues/2220 | Multiple GPU train dreambooth problems？ | **what happend？**
I used **a gpu A100** to train stable diffution(https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth.py), and change the train_batch_size from 1 to 2, 4.
The following is the experimental data：
GPU. batch size. train cost time. ... | closed | completed | false | 2 | [
"stale"
] | [
"williamberman",
"patil-suraj"
] | 2023-02-02T10:50:51Z | 2023-03-13T15:03:41Z | 2023-03-13T15:03:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wen020 | 54,690,997 | MDQ6VXNlcjU0NjkwOTk3 | User | false |
huggingface/diffusers | 1,568,236,494 | I_kwDOHa8MBc5deWPO | 2,221 | https://github.com/huggingface/diffusers/issues/2221 | https://api.github.com/repos/huggingface/diffusers/issues/2221 | colossalai error | **What happend？**
I used colossalai to train stable diffution(https://github.com/huggingface/diffusers/blob/main/examples/research_projects/colossalai/train_dreambooth_colossalai.py) and find the error:
Traceback (most recent call last):
File "/workspace/mutilObject/diffusers/examples/research_projects/colossalai/... | closed | completed | false | 3 | [] | [] | 2023-02-02T15:12:44Z | 2023-02-07T07:55:40Z | 2023-02-07T07:55:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wen020 | 54,690,997 | MDQ6VXNlcjU0NjkwOTk3 | User | false |
huggingface/diffusers | 1,568,439,339 | I_kwDOHa8MBc5dfHwr | 2,222 | https://github.com/huggingface/diffusers/issues/2222 | https://api.github.com/repos/huggingface/diffusers/issues/2222 | Diffusers API/Design Philosophy | 🚨🚨 **Diffusers Design Philosophy** 🚨🚨
If you're interested in contributing or understanding this library a bit better, please have a look at the new design philosophy doc:
https://huggingface.co/docs/diffusers/main/en/conceptual/philosophy
Feedback and comments are very welcome below | closed | completed | false | 4 | [
"stale"
] | [] | 2023-02-02T17:10:27Z | 2023-03-13T15:03:39Z | 2023-03-13T15:03:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,568,541,606 | I_kwDOHa8MBc5dfgum | 2,224 | https://github.com/huggingface/diffusers/issues/2224 | https://api.github.com/repos/huggingface/diffusers/issues/2224 | Another model on training Dreambooth | When I try to change the model on training Dreambooth to any other from Huggingface, it give me an error that says model_index.json cannot found.
I'm trying to change it from the "Settings and Run" section on Colab.
Firstly it cannot find the Revision Fp16 and when I delete Fp16, it gives me that model_index.json... | closed | completed | false | 0 | [] | [] | 2023-02-02T18:14:06Z | 2023-02-02T18:24:58Z | 2023-02-02T18:24:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xstrauss | 59,344,184 | MDQ6VXNlcjU5MzQ0MTg0 | User | false |
huggingface/diffusers | 1,568,606,744 | I_kwDOHa8MBc5dfwoY | 2,225 | https://github.com/huggingface/diffusers/issues/2225 | https://api.github.com/repos/huggingface/diffusers/issues/2225 | Using safety checker outside of stable diffusion pipeline | I am interested in flagging images I am analyzing as safe/unsafe and know that stable diffusion uses a safety checker. Is it possible to use this safety checker in isolation (i.e. without using it in the stable diffusion pipeline)? If so, I would really appreciate any examples or suggestions on how to do so! Here is wh... | closed | completed | false | 2 | [] | [] | 2023-02-02T18:52:58Z | 2023-02-03T18:28:56Z | 2023-02-03T18:28:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | preethiseshadri518 | 60,128,552 | MDQ6VXNlcjYwMTI4NTUy | User | false |
huggingface/diffusers | 1,568,743,366 | I_kwDOHa8MBc5dgR_G | 2,226 | https://github.com/huggingface/diffusers/issues/2226 | https://api.github.com/repos/huggingface/diffusers/issues/2226 | Rank 0 Higher Memory Usage with Xformers | ### Describe the bug
When training stable diffusion with xformers, rank 0 has substantially higher memory usage. When not using xformers, this behavior is not observed
<img width="379" alt="image" src="https://user-images.githubusercontent.com/17102158/216443051-267b11c3-4522-49d9-a9cf-7dd196285c84.png">
![image](... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-02-02T20:33:25Z | 2023-02-03T01:30:38Z | 2023-02-03T01:30:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mvpatel2000 | 17,102,158 | MDQ6VXNlcjE3MTAyMTU4 | User | false |
huggingface/diffusers | 1,568,983,349 | I_kwDOHa8MBc5dhMk1 | 2,228 | https://github.com/huggingface/diffusers/issues/2228 | https://api.github.com/repos/huggingface/diffusers/issues/2228 | Pipe doesnt get loaded into GPU StableDiffusionPipeline.from_pretrained | ### Describe the bug
My system nvidea 3080 ti (12 Gig) win 11 cuda 12 (even with the toolkit), with WSL2 ubuntu
Pytorch works fine with cuda in my other jupyterlab experiments.
I can login to huginglab, (i see in debug console), and have tried various models.
I tried with autocast and without, the fan's of the... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-02-03T00:37:56Z | 2023-03-13T15:03:38Z | 2023-03-13T15:03:38Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | PGTBoos | 23,136,027 | MDQ6VXNlcjIzMTM2MDI3 | User | false |
huggingface/diffusers | 1,569,490,037 | I_kwDOHa8MBc5djIR1 | 2,230 | https://github.com/huggingface/diffusers/issues/2230 | https://api.github.com/repos/huggingface/diffusers/issues/2230 | Do models should be downloaded before using dream booth? | Hi, thanks for your great work!
```
export MODEL_NAME="CompVis/stable-diffusion-v1-4"
export OUTPUT_DIR="path-to-save-model"
```
When using the code, does the pretrained model just automatically downloaded?
Or should I download the model separately and place it in same folder before using dream booth??
I'm aski... | closed | completed | false | 2 | [] | [] | 2023-02-03T09:15:02Z | 2023-02-03T09:28:29Z | 2023-02-03T09:28:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YJ-142150 | 62,554,000 | MDQ6VXNlcjYyNTU0MDAw | User | false |
huggingface/diffusers | 1,569,714,318 | I_kwDOHa8MBc5dj_CO | 2,233 | https://github.com/huggingface/diffusers/issues/2233 | https://api.github.com/repos/huggingface/diffusers/issues/2233 | Issues in Textual_Inversion example (Validation code and Gradient accumulation) | Hi,
I'm glad that you build this amazing project!
But there seems some issues in the Textual_Inversion example (https://github.com/huggingface/diffusers/blob/main/examples/textual_inversion/textual_inversion.py):
1. The validation pipeline forget to replace the tokenizer at Line783 (https://github.com/huggingfa... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-02-03T11:46:28Z | 2023-03-13T15:03:34Z | 2023-03-13T15:03:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yxding95 | 16,249,556 | MDQ6VXNlcjE2MjQ5NTU2 | User | false |
huggingface/diffusers | 1,570,343,183 | I_kwDOHa8MBc5dmYkP | 2,234 | https://github.com/huggingface/diffusers/issues/2234 | https://api.github.com/repos/huggingface/diffusers/issues/2234 | Can't finetune stable diffusion with --enable_xformers_memory_efficient_attention | ### Describe the bug
I'm trying to finetune stable diffusion, and I'm trying to reduce the memory footprint so I can train with a larger batch size (and thus fewer gradient accumulation steps, and thus faster).
Setting `--enable_xformers_memory_efficient_attention` results in numeric instability of some kind, I thi... | closed | completed | false | 14 | [
"bug"
] | [] | 2023-02-03T19:34:54Z | 2025-01-09T02:17:29Z | 2023-02-06T05:20:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LucasSloan | 102,313 | MDQ6VXNlcjEwMjMxMw== | User | false |
huggingface/diffusers | 1,570,456,444 | I_kwDOHa8MBc5dm0N8 | 2,236 | https://github.com/huggingface/diffusers/issues/2236 | https://api.github.com/repos/huggingface/diffusers/issues/2236 | Document disabling progress bar in docs for DiffusionPipeline | The [DiffusionPipeline docs](https://huggingface.co/docs/diffusers/main/en/api/diffusion_pipeline#diffusers.DiffusionPipeline) currently mention "enabling/disabling the progress bar for the denoising iteration" but do not document how to do this.
By reading the source code I figured out that I could disable it via:
... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-03T21:09:59Z | 2023-03-14T15:03:29Z | 2023-03-14T15:03:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | morrisalp | 8,263,996 | MDQ6VXNlcjgyNjM5OTY= | User | false |
huggingface/diffusers | 1,570,611,058 | I_kwDOHa8MBc5dnZ9y | 2,238 | https://github.com/huggingface/diffusers/issues/2238 | https://api.github.com/repos/huggingface/diffusers/issues/2238 | Is it possible to change the prompt on the fly? | E.g. if 20 inference steps are to be performed, is it possible to run the first 10 steps with one prompt and the second 10 steps with another? `lpw_stable_diffusion` seems to change the effective prompt every inference step, doesn't it?
This question is motivated by the observation that some words in the prompt affe... | closed | completed | false | 8 | [
"stale"
] | [] | 2023-02-03T23:22:25Z | 2023-04-14T15:03:38Z | 2023-04-14T15:03:38Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nickkolok | 5,008,131 | MDQ6VXNlcjUwMDgxMzE= | User | false |
huggingface/diffusers | 1,570,651,921 | I_kwDOHa8MBc5dnj8R | 2,239 | https://github.com/huggingface/diffusers/issues/2239 | https://api.github.com/repos/huggingface/diffusers/issues/2239 | Is unet architecture changed as diffusers upgrading? | ### Describe the bug
I want to export onnx for UNet2DConditionModel by torch.onnx.export. However, when I use diffusers==0.12.1, I get a model larger than 2GB, which outputs a bunch of weight files. When I use diffusers==0.7.2, I get a normal onnx file of 1.7GB.
### Reproduction
```python
import onnx
import torch
... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-02-04T00:15:52Z | 2023-03-21T15:03:37Z | 2023-03-21T15:03:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 1049451037 | 15,194,939 | MDQ6VXNlcjE1MTk0OTM5 | User | false |
huggingface/diffusers | 1,570,728,909 | I_kwDOHa8MBc5dn2vN | 2,240 | https://github.com/huggingface/diffusers/issues/2240 | https://api.github.com/repos/huggingface/diffusers/issues/2240 | keyError : Parameter containing: | Hello, thanks for your great work.
I tried multi_subject_dreambooth with DeepSpeed. But this error showed up.
```
Traceback (most recent call last):
File "train_multi_subject_dreambooth.py", line 869, in <module>
main(args)
File "train_multi_subject_dreambooth.py", line 694, in main
unet, text_encod... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-04T03:08:15Z | 2023-03-14T15:03:24Z | 2023-03-14T15:03:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YJ-142150 | 62,554,000 | MDQ6VXNlcjYyNTU0MDAw | User | false |
huggingface/diffusers | 1,570,878,400 | I_kwDOHa8MBc5dobPA | 2,242 | https://github.com/huggingface/diffusers/issues/2242 | https://api.github.com/repos/huggingface/diffusers/issues/2242 | Some code are missing | I don't find the Training was done using a [modified version of the original Stable Diffusion training code](https://huggingface.co/lambdalabs/sd-image-variations-diffusers/blob/main/(https://github.com/justinpinkney/stable-diffusion).
Something is empty.
Can you reload this code?
Thanks!!!
![image](https://user... | closed | completed | false | 2 | [] | [] | 2023-02-04T10:42:44Z | 2023-02-17T22:35:18Z | 2023-02-17T22:35:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pokameng | 47,957,015 | MDQ6VXNlcjQ3OTU3MDE1 | User | false |
huggingface/diffusers | 1,571,170,397 | I_kwDOHa8MBc5dpihd | 2,244 | https://github.com/huggingface/diffusers/issues/2244 | https://api.github.com/repos/huggingface/diffusers/issues/2244 | Dreambooth inference fails with runtime error "norma_cpu" not implemented for 'Half' | ### Describe the bug
In the dreambooth example, model inference fails with the runtime error `"norma_cpu" not implemented for 'Half'`. Is this a bug? How do I resolve this?
I trained my model using my own dataset for instance images and class images. I followed the instructions here for training - https://github.c... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-02-04T23:42:46Z | 2023-02-05T20:25:16Z | 2023-02-05T20:25:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kirit93 | 8,190,572 | MDQ6VXNlcjgxOTA1NzI= | User | false |
huggingface/diffusers | 1,571,227,567 | I_kwDOHa8MBc5dpwev | 2,246 | https://github.com/huggingface/diffusers/issues/2246 | https://api.github.com/repos/huggingface/diffusers/issues/2246 | ONNX GPU Freezing | ### Describe the bug
Image generated in ` 44%|####4 | 11/25 [00:21<00:26, 1.89s/it]`
What does the algorithm do the rest of the time? The image is really ready-made on the disk and does not change any... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-02-05T04:14:22Z | 2023-02-06T10:32:57Z | 2023-02-06T10:32:57Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ForserX | 13,867,290 | MDQ6VXNlcjEzODY3Mjkw | User | false |
huggingface/diffusers | 1,571,411,555 | I_kwDOHa8MBc5dqdZj | 2,247 | https://github.com/huggingface/diffusers/issues/2247 | https://api.github.com/repos/huggingface/diffusers/issues/2247 | Does LoRa affect memory usage and run-time during inference? | Hello,
I fine-tuned the Stable Diffusion using LoRa. Although the fine-tuning became much faster and required less memory, the inference was heavily affected. Now, inference has become very slow and needs more memory.
Is this normal? | closed | completed | false | 3 | [] | [] | 2023-02-05T12:33:57Z | 2023-02-13T11:17:36Z | 2023-02-13T11:17:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | FBehrad | 41,340,554 | MDQ6VXNlcjQxMzQwNTU0 | User | false |
huggingface/diffusers | 1,571,601,051 | I_kwDOHa8MBc5drLqb | 2,250 | https://github.com/huggingface/diffusers/issues/2250 | https://api.github.com/repos/huggingface/diffusers/issues/2250 | Dreambooth inference fails to generate image with prompt | ### Describe the bug
I trained my model using my own dataset for instance images and class images. I followed the instructions here for training - https://github.com/huggingface/diffusers/tree/main/examples/dreambooth#training-on-a-12gb-gpu.
I am using a SageMaker Notebook instance running on a ml.g4dn.xlarge image... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-02-05T20:29:15Z | 2023-02-06T01:28:01Z | 2023-02-06T01:28:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kirit93 | 8,190,572 | MDQ6VXNlcjgxOTA1NzI= | User | false |
huggingface/diffusers | 1,572,056,181 | I_kwDOHa8MBc5ds6x1 | 2,253 | https://github.com/huggingface/diffusers/issues/2253 | https://api.github.com/repos/huggingface/diffusers/issues/2253 | Use KerasCV fine-tuned checkpoints in Diffusers 🧨 | You can now fine-tune Stable Diffusion (v1) models and use them in Diffusers as standard `StableDiffusionPipeline`s. Check out the following document to learn more!
[Using KerasCV Stable Diffusion Checkpoints in Diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/kerascv) | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-06T07:22:13Z | 2023-03-09T12:14:11Z | 2023-03-09T12:14:10Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,572,155,049 | I_kwDOHa8MBc5dtS6p | 2,255 | https://github.com/huggingface/diffusers/issues/2255 | https://api.github.com/repos/huggingface/diffusers/issues/2255 | Is the train_dreambooth.py only trains the VAE part or also including diffusers models? | After training, only there are some scheduler, VAE, U net and Text related folders & bin files saved, so want to confirm this, thanks | closed | completed | false | 2 | [
"stale"
] | [] | 2023-02-06T08:45:13Z | 2023-03-16T15:03:04Z | 2023-03-16T15:03:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wyang22 | 17,822,943 | MDQ6VXNlcjE3ODIyOTQz | User | false |
huggingface/diffusers | 1,572,204,448 | I_kwDOHa8MBc5dte-g | 2,256 | https://github.com/huggingface/diffusers/issues/2256 | https://api.github.com/repos/huggingface/diffusers/issues/2256 | Is the train_dreambooth.py only trains the VAE part or also including diffusers models? | After training, only there are some scheduler, VAE, U net and Text related folders & bin files saved, and there is no bin file generated in the scheduler folder, so want to confirm this, thanks | closed | completed | false | 0 | [] | [] | 2023-02-06T09:19:01Z | 2023-02-06T09:21:45Z | 2023-02-06T09:20:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wyang22 | 17,822,943 | MDQ6VXNlcjE3ODIyOTQz | User | false |
huggingface/diffusers | 1,572,349,592 | I_kwDOHa8MBc5duCaY | 2,261 | https://github.com/huggingface/diffusers/issues/2261 | https://api.github.com/repos/huggingface/diffusers/issues/2261 | zwx | **Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
**Describe the solution you'd like**
A clear and concise description of what you want to happen.
**Describe alternatives you've considered**
A clear an... | closed | completed | false | 0 | [
"spam"
] | [] | 2023-02-06T10:57:44Z | 2023-02-08T15:05:43Z | 2023-02-06T12:40:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | no-static | 117,646,893 | U_kgDOBwMmLQ | User | false |
huggingface/diffusers | 1,572,377,929 | I_kwDOHa8MBc5duJVJ | 2,262 | https://github.com/huggingface/diffusers/issues/2262 | https://api.github.com/repos/huggingface/diffusers/issues/2262 | How to use safetensors LoRA model? | I tried to disassemble the model with local converters, but it didn't work out. | closed | completed | false | 3 | [
"stale"
] | [] | 2023-02-06T11:13:48Z | 2023-03-09T12:48:51Z | 2023-03-09T12:48:51Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ForserX | 13,867,290 | MDQ6VXNlcjEzODY3Mjkw | User | false |
huggingface/diffusers | 1,573,206,910 | I_kwDOHa8MBc5dxTt- | 2,264 | https://github.com/huggingface/diffusers/issues/2264 | https://api.github.com/repos/huggingface/diffusers/issues/2264 | Setting seed in StableDiffusionInstructPix2PixPipeline throws error | ### Describe the bug
Hello!
I tried to use this code to generate a pix2pix image with a given seed:
model_id = "timbrooks/instruct-pix2pix"
```
prompt = "turn the image into Retro comic style artwork style"
guidance_scale = 7
negative_prompt=""
image_guidance_scale=1.9
num_inference_steps=22
generator = t... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-02-06T20:11:33Z | 2023-02-09T09:41:36Z | 2023-02-08T16:00:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Marcophono2 | 22,599,855 | MDQ6VXNlcjIyNTk5ODU1 | User | false |
huggingface/diffusers | 1,573,452,000 | I_kwDOHa8MBc5dyPjg | 2,265 | https://github.com/huggingface/diffusers/issues/2265 | https://api.github.com/repos/huggingface/diffusers/issues/2265 | unconditional_image_generation/train_unconditional.py incompatibile with ONNXRuntime | ### Describe the bug
I am testing compatibility of ONNXRuntime with the diffusers unet and am running into an issue with a post-processing step in the train_unconditional.py example script.
In previous discussions, we discovered an issue where ONNXRuntime expects a plain numpy array as output of the forward pass ... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-02-06T23:30:14Z | 2023-04-22T15:03:53Z | 2023-04-22T15:03:53Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | prathikr | 31,260,940 | MDQ6VXNlcjMxMjYwOTQw | User | false |
huggingface/diffusers | 1,573,626,028 | I_kwDOHa8MBc5dy6Cs | 2,266 | https://github.com/huggingface/diffusers/issues/2266 | https://api.github.com/repos/huggingface/diffusers/issues/2266 | sequential cpu offload causes > 3x slowdown | Using StableDiffusionPipeline.enable_sequential_cpu_offload causes massive slowdown. e.g. from 6 it/s to 2.
### Reproduction
Standard pipeline:
```py
pipe = diffusers.StableDiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
revision="fp16", torch_dtype=torch.float16,
)
pipe.enabl... | closed | completed | false | 4 | [
"bug",
"stale"
] | [
"pcuenca"
] | 2023-02-07T03:41:59Z | 2023-04-09T15:03:21Z | 2023-04-09T15:03:21Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | keturn | 83,819 | MDQ6VXNlcjgzODE5 | User | false |
huggingface/diffusers | 1,573,981,081 | I_kwDOHa8MBc5d0QuZ | 2,268 | https://github.com/huggingface/diffusers/issues/2268 | https://api.github.com/repos/huggingface/diffusers/issues/2268 | colossalai problem | **What happend？**
I used colossalai to train stable diffution (https://github.com/huggingface/diffusers/blob/main/examples/research_projects/colossalai/train_dreambooth_colossalai.py) to train dreambooth and found to be slower than train_dreambooth.py (https://github.com/huggingface/diffusers/blob/main/examples/dream... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-02-07T09:28:45Z | 2023-03-17T15:03:33Z | 2023-03-17T15:03:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wen020 | 54,690,997 | MDQ6VXNlcjU0NjkwOTk3 | User | false |
huggingface/diffusers | 1,574,234,776 | I_kwDOHa8MBc5d1OqY | 2,270 | https://github.com/huggingface/diffusers/issues/2270 | https://api.github.com/repos/huggingface/diffusers/issues/2270 | Does it support negative prompt? | I tried pipe(negative_prompt="uncensored") but it seems not working.
Does it support negative prompts? | closed | completed | false | 0 | [] | [] | 2023-02-07T12:23:42Z | 2023-02-07T15:15:53Z | 2023-02-07T15:15:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JioChoi | 73,428,860 | MDQ6VXNlcjczNDI4ODYw | User | false |
huggingface/diffusers | 1,574,436,390 | I_kwDOHa8MBc5d1_4m | 2,272 | https://github.com/huggingface/diffusers/issues/2272 | https://api.github.com/repos/huggingface/diffusers/issues/2272 | convert_stable_diffusion_checkpoint_to_onnx.py doesn't convert | ### Describe the bug
convert_stable_diffusion_checkpoint_to_onnx.py doesn't convert .ckpt to .onnx and throws
```
raise EnvironmentError(f"It looks like the config file at '{config_file}' is not a valid JSON file.")
OSError: It looks like the config file at 'K:\dreamshaper_33.ckpt' is not a valid JSON file.
```
... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-02-07T14:27:00Z | 2023-03-21T15:03:34Z | 2023-03-21T15:03:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dingo64 | 13,681,463 | MDQ6VXNlcjEzNjgxNDYz | User | false |
huggingface/diffusers | 1,574,522,295 | I_kwDOHa8MBc5d2U23 | 2,273 | https://github.com/huggingface/diffusers/issues/2273 | https://api.github.com/repos/huggingface/diffusers/issues/2273 | Does it support vae and negative prompts? | I couldn't find any information about those. Could you give me some sample code? | closed | completed | false | 0 | [] | [] | 2023-02-07T15:17:43Z | 2023-02-07T15:18:22Z | 2023-02-07T15:18:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JioChoi | 73,428,860 | MDQ6VXNlcjczNDI4ODYw | User | false |
huggingface/diffusers | 1,574,706,690 | I_kwDOHa8MBc5d3B4C | 2,276 | https://github.com/huggingface/diffusers/issues/2276 | https://api.github.com/repos/huggingface/diffusers/issues/2276 | xformers not installing on dreambooth training colab notebook | ### Describe the bug
Error in the second step of SD Dreambooth Training Colab notebook installing xformers
Then, Out of Memory error on "run training"
Anyone has suggestions on how to make the notebook work?
Thanks!
### Reproduction
https://colab.research.google.com/github/huggingface/notebooks/blob/main/d... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-02-07T17:04:34Z | 2023-03-19T15:03:03Z | 2023-03-19T15:03:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | estelleflores | 83,784,489 | MDQ6VXNlcjgzNzg0NDg5 | User | false |
huggingface/diffusers | 1,574,791,580 | I_kwDOHa8MBc5d3Wmc | 2,277 | https://github.com/huggingface/diffusers/issues/2277 | https://api.github.com/repos/huggingface/diffusers/issues/2277 | UNet2DCondition model can be passed `attention_head_dim` of wrong length | ### Describe the bug
When passed `attention_head_dim` of wrong length, it will silently be used. The encoder will start indexing from the front of the list and the decoder will start indexing from the back.
### Reproduction
This test is an example of an incorrectly parameterized conditional unet. It has two bl... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-02-07T17:54:08Z | 2023-02-14T00:27:38Z | 2023-02-14T00:27:38Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | williamberman | 6,505,431 | MDQ6VXNlcjY1MDU0MzE= | User | false |
huggingface/diffusers | 1,346,071,403 | I_kwDOHa8MBc5QO2tr | 228 | https://github.com/huggingface/diffusers/issues/228 | https://api.github.com/repos/huggingface/diffusers/issues/228 | stable-diffusion-v1-4 link in release v0.2.3 is broken | ### Describe the bug
@anton-l the link (https://huggingface.co/CompVis/stable-diffusion-v1-4) in the [release v0.2.3](https://github.com/huggingface/diffusers/releases/tag/v0.2.3) returns a 404.
### Reproduction
_No response_
### Logs
_No response_
### System Info
```shell
N/A
```
| closed | completed | false | 12 | [
"question"
] | [] | 2022-08-22T09:07:27Z | 2022-08-22T20:53:00Z | 2022-08-22T19:21:18Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | leszekhanusz | 4,641,499 | MDQ6VXNlcjQ2NDE0OTk= | User | false |
huggingface/diffusers | 1,575,122,169 | I_kwDOHa8MBc5d4nT5 | 2,280 | https://github.com/huggingface/diffusers/issues/2280 | https://api.github.com/repos/huggingface/diffusers/issues/2280 | Script for converting stable diffusion to stable diffusion inpainting. | I have been trying to fine tune stable diffusion and convert that into inpainting model and resume training from there. Any script for model surgery / conversion pipeline would be great. | closed | completed | false | 3 | [] | [] | 2023-02-07T22:29:43Z | 2023-02-27T09:11:27Z | 2023-02-08T16:06:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | theja-vanka | 22,734,103 | MDQ6VXNlcjIyNzM0MTAz | User | false |
huggingface/diffusers | 1,575,686,516 | I_kwDOHa8MBc5d6xF0 | 2,283 | https://github.com/huggingface/diffusers/issues/2283 | https://api.github.com/repos/huggingface/diffusers/issues/2283 | When converting "runwayml/stable-diffusion-v1-5" to onnx model: AttributeError: 'CLIPVisionConfig' object has no attribute 'num_channels' | ### Describe the bug
When converting "runwayml/stable-diffusion-v1-5" to onnx model: AttributeError: 'CLIPVisionConfig' object has no attribute 'num_channels'
I tried to change the code in line 192 "clip_num_channels = safety_checker.config.vision_config.num_channels" to "clip_num_channels = 3"(according to /root/.ca... | closed | completed | false | 4 | [
"bug",
"stale"
] | [
"anton-l",
"echarlaix"
] | 2023-02-08T08:22:12Z | 2023-04-14T15:03:34Z | 2023-04-14T15:03:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | labmanno004 | 87,518,566 | MDQ6VXNlcjg3NTE4NTY2 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.