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 | 1,688,826,632 | I_kwDOHa8MBc5kqXMI | 3,276 | https://github.com/huggingface/diffusers/issues/3276 | https://api.github.com/repos/huggingface/diffusers/issues/3276 | Adding 'strength' parameter to StableDiffusionInpaintingPipeline | Currently in inpainting pipelines (e.g. StableDiffusionInpaintingPipeline), masked areas are expected to be 100% noise. During the diffussion sampling process the masked areas are then generated w.r.t the visible image. The problem with this process is that it assumes you have no estimate of the underlying masked area ... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-04-28T16:25:08Z | 2023-06-07T14:55:57Z | 2023-06-07T14:55:57Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rupertmenneer | 71,332,436 | MDQ6VXNlcjcxMzMyNDM2 | User | false |
huggingface/diffusers | 1,689,022,511 | I_kwDOHa8MBc5krHAv | 3,279 | https://github.com/huggingface/diffusers/issues/3279 | https://api.github.com/repos/huggingface/diffusers/issues/3279 | train_dreambooth_lora.py resume from checkpoint errors. | ### Describe the bug
I get the following issue when trying to resume from checkpoint.
`KeyError: 'unet.down_blocks.0.attentions.0.transformer_blocks.0.attn1.processor'`
There was also a naming issue where I had to change pytorch_lora_weights.bin to pytorch_model.bin in the checkpoint folders
### Reproduction
acce... | closed | completed | false | 22 | [
"bug",
"stale"
] | [] | 2023-04-28T19:16:24Z | 2023-11-11T03:30:53Z | 2023-06-08T15:03:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JaLnYn | 33,004,730 | MDQ6VXNlcjMzMDA0NzMw | User | false |
huggingface/diffusers | 1,689,286,003 | I_kwDOHa8MBc5ksHVz | 3,280 | https://github.com/huggingface/diffusers/issues/3280 | https://api.github.com/repos/huggingface/diffusers/issues/3280 | No non-default schedulers appear to work with DeepFloyd IF | ### Describe the bug
Attempting to use any non-default scheduler with DeepFloyd IF crashes out like this:
```py
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ <stdin>:1 in <module> │
... | closed | completed | false | 8 | [
"bug",
"New scheduler",
"Good second issue"
] | [] | 2023-04-29T00:45:46Z | 2024-11-21T15:38:58Z | 2024-11-21T15:38:58Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 1,689,345,961 | I_kwDOHa8MBc5ksV-p | 3,282 | https://github.com/huggingface/diffusers/issues/3282 | https://api.github.com/repos/huggingface/diffusers/issues/3282 | No operator found for `memory_efficient_attention_forward` with inputs | ### Describe the bug
Running Dreambooth inside a Docker container causes it to crash with the error
```
NotImplementedError: No operator found for `memory_efficient_attention_forward` with inputs:
...
`flshattF` is not supported because:
device=cpu (supported: {'cuda'})
dtype=torch.float32 (suppo... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-04-29T03:39:05Z | 2023-06-18T15:03:09Z | 2023-06-18T15:03:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gameveloster | 103,383,490 | U_kgDOBimBwg | User | false |
huggingface/diffusers | 1,689,347,742 | I_kwDOHa8MBc5ksWae | 3,283 | https://github.com/huggingface/diffusers/issues/3283 | https://api.github.com/repos/huggingface/diffusers/issues/3283 | How to convert LORA .safetensors to .bin? | I am trying to convert [a LORA safetensor](https://civitai.com/models/13941/epinoiseoffset) into a `.bin` so it can be used in the `StableDiffusionPipeline`, for example:
```py
from diffusers import StableDiffusionPipeline
pipe = StableDiffusionPipeline.from_pretrained(sd_model_path, torch_dtype=torch.float16).t... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-04-29T03:46:18Z | 2024-09-09T02:19:38Z | 2023-06-06T15:03:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | athenawisdoms | 6,615,285 | MDQ6VXNlcjY2MTUyODU= | User | false |
huggingface/diffusers | 1,689,403,298 | I_kwDOHa8MBc5ksj-i | 3,284 | https://github.com/huggingface/diffusers/issues/3284 | https://api.github.com/repos/huggingface/diffusers/issues/3284 | run train_dreambooth_lora.py failed with accelerate | ### Describe the bug
Thanks for this awesome project!
When I run the script "train_dreambooth_lora.py" without acceleration, it works fine. But when I use acceleration launch, it fails when the number of steps reaches "checkpointing_steps".
I am running the script in a Docker with 4 * 3090 vGPUs. And I ran acceler... | closed | completed | false | 23 | [
"bug",
"stale"
] | [] | 2023-04-29T05:40:24Z | 2023-08-03T15:03:55Z | 2023-08-03T15:03:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | webliupeng | 3,380,812 | MDQ6VXNlcjMzODA4MTI= | User | false |
huggingface/diffusers | 1,689,584,082 | I_kwDOHa8MBc5ktQHS | 3,287 | https://github.com/huggingface/diffusers/issues/3287 | https://api.github.com/repos/huggingface/diffusers/issues/3287 | Support OpenAI improved DDPM | **Is your feature request related to a problem? Please describe.**
Reduce the steps needed in the inference process is important for the practical deployment.
**Describe the solution you'd like**
We should support Variational bound loss when `variance_type=learned_range`.
**Additional context**
The original... | closed | completed | false | 13 | [
"stale"
] | [] | 2023-04-29T14:46:34Z | 2025-07-26T21:14:32Z | 2023-07-05T15:04:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dathudeptrai | 43,868,663 | MDQ6VXNlcjQzODY4NjYz | User | false |
huggingface/diffusers | 1,689,607,632 | I_kwDOHa8MBc5ktV3Q | 3,288 | https://github.com/huggingface/diffusers/issues/3288 | https://api.github.com/repos/huggingface/diffusers/issues/3288 | NameError: name 'str2optimizer8bit_blockwise' is not defined | I'm running dreambooth training on pytorch 2.0 and get this `bitsandbytes` error at the start of training (after caching latents and creating class images)
```
F.optimizer_update_8bit_blockwise(
File "/opt/conda/lib/python3.10/site-packages/bitsandbytes/functional.py", line 975, in optimizer_update_8bit_bloc... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-04-29T16:07:25Z | 2023-06-07T15:03:04Z | 2023-06-07T15:03:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gameveloster | 103,383,490 | U_kgDOBimBwg | User | false |
huggingface/diffusers | 1,689,616,914 | I_kwDOHa8MBc5ktYIS | 3,289 | https://github.com/huggingface/diffusers/issues/3289 | https://api.github.com/repos/huggingface/diffusers/issues/3289 | Super-resolution in DeepFloyd is bugged for any non-64x64 to 256x256 upscales | ### Describe the bug
Superresolution code is currently bugged because it does not accept height and width, and rather just assigns a height and width of 256x256. This results in squished images when you try to super resolution anything from the first stage model that is not 64x64.
https://github.com/huggingface/dif... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-04-29T16:40:34Z | 2023-05-17T02:21:08Z | 2023-05-17T02:21:08Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 1,359,915,048 | I_kwDOHa8MBc5RDqgo | 329 | https://github.com/huggingface/diffusers/issues/329 | https://api.github.com/repos/huggingface/diffusers/issues/329 | Support K-LMS for inpainting and img2img for stable diffusion | When it was in reference, it supported img2img, but now it doesn't.
| closed | completed | false | 3 | [] | [] | 2022-09-02T09:06:55Z | 2022-09-15T15:40:00Z | 2022-09-03T10:20:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Inkorak | 52,286,717 | MDQ6VXNlcjUyMjg2NzE3 | User | false |
huggingface/diffusers | 1,689,638,954 | I_kwDOHa8MBc5ktdgq | 3,290 | https://github.com/huggingface/diffusers/issues/3290 | https://api.github.com/repos/huggingface/diffusers/issues/3290 | AttributeError: 'NoneType' object has no attribute 'flush' | ### Describe the bug
I have a python program that I have converted into an exe file using PyInstaller. I have the latest versions of Diffusers and Transformers library which are 0.16.1 and 4.28.1 respectively. When I run the exe file, I get this error: AttributeError: 'NoneType' object has no attribute 'flush'. But wh... | closed | completed | false | 12 | [
"bug",
"stale"
] | [] | 2023-04-29T17:56:58Z | 2024-05-24T05:16:50Z | 2023-06-24T15:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Xvareon | 91,329,439 | MDQ6VXNlcjkxMzI5NDM5 | User | false |
huggingface/diffusers | 1,689,728,052 | I_kwDOHa8MBc5ktzQ0 | 3,291 | https://github.com/huggingface/diffusers/issues/3291 | https://api.github.com/repos/huggingface/diffusers/issues/3291 | Deepfloyd aspect ratios | **Is your feature request related to a problem? Please describe.**
I saw that the pre-release cherry picks had various aspect ratios. As far as I know, the current implementation only has a standard aspect ratio
**Describe the solution you'd like**
The ability to specify resolutions for both stage 1 and 2 | closed | completed | false | 11 | [
"stale"
] | [] | 2023-04-30T00:20:42Z | 2023-06-12T15:03:01Z | 2023-06-12T15:03:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dblunk88 | 39,381,389 | MDQ6VXNlcjM5MzgxMzg5 | User | false |
huggingface/diffusers | 1,689,920,734 | I_kwDOHa8MBc5kuiTe | 3,293 | https://github.com/huggingface/diffusers/issues/3293 | https://api.github.com/repos/huggingface/diffusers/issues/3293 | Adding input pertubation | Input perturbation is a very simple change proposed [here](https://arxiv.org/abs/2301.11706) where we add noise during training to make the trainer robust to the error accumulation during inference. This got state-of-the-art on the celeb dataset as can be seen [here](https://paperswithcode.com/sota/image-generation-on-... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-04-30T13:32:57Z | 2023-06-12T17:24:47Z | 2023-06-12T17:24:46Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | isamu-isozaki | 23,430,101 | MDQ6VXNlcjIzNDMwMTAx | User | false |
huggingface/diffusers | 1,689,943,076 | I_kwDOHa8MBc5kunwk | 3,296 | https://github.com/huggingface/diffusers/issues/3296 | https://api.github.com/repos/huggingface/diffusers/issues/3296 | saving and loading checkpoints do not work on train_dreambooth.py when using text_encoder | ### Describe the bug
There are 2 issues when trying to start from a checkpoint, when using --train_text_encoder.
First issue is #2480 . I wrote there how to fix it in train_dreambooth.py. It fixes how the checkpoint is saved.
Second issue is similar. It is in load_model_hook (how the text_encoder checkpoint is load... | closed | completed | false | 16 | [
"bug",
"stale"
] | [
"williamberman"
] | 2023-04-30T14:43:47Z | 2023-06-12T03:41:50Z | 2023-06-11T15:02:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | amitgurintelcom | 50,952,358 | MDQ6VXNlcjUwOTUyMzU4 | User | false |
huggingface/diffusers | 1,283,257,469 | I_kwDOHa8MBc5MfPR9 | 33 | https://github.com/huggingface/diffusers/issues/33 | https://api.github.com/repos/huggingface/diffusers/issues/33 | New modalities | One of the stated design decisions from the readme was to support arbitrary modalities, not just images.
I'm in the process of trying to adapt the code for 1D vectors (not H x W x C images).
this line:
` noisy_images = noise_scheduler.training_step(clean_inputs, noise_samples, timesteps)
`
a... | closed | completed | false | 10 | [
"stale"
] | [
"anton-l"
] | 2022-06-24T04:48:48Z | 2022-11-28T15:41:58Z | 2022-11-28T15:04:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | richardrl | 5,181,990 | MDQ6VXNlcjUxODE5OTA= | User | false |
huggingface/diffusers | 1,690,410,628 | I_kwDOHa8MBc5kwZ6E | 3,300 | https://github.com/huggingface/diffusers/issues/3300 | https://api.github.com/repos/huggingface/diffusers/issues/3300 | Diffusers on HuggingFace are broken | ### Describe the bug
Diffusers broken
### Reproduction
pipe = DiffusionPipeline.from_pretrained(
pretrained_model_name_or_path=model,
safety_checker=None,
custom_pipeline="lpw_stable_diffusion",
)
### Logs
```shell
Traceback (most recent call last):
File "/src/script/download-wei... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-05-01T05:42:13Z | 2023-06-09T15:03:02Z | 2023-06-09T15:03:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dpyy | 19,830,772 | MDQ6VXNlcjE5ODMwNzcy | User | false |
huggingface/diffusers | 1,690,589,722 | I_kwDOHa8MBc5kxFoa | 3,301 | https://github.com/huggingface/diffusers/issues/3301 | https://api.github.com/repos/huggingface/diffusers/issues/3301 | train_dreambooth_lora is not working! | ### Describe the bug
Hello!
I tried to use train_dreambooth_lora but after finishing training, I can't see my face on the result image that I trained!
**After making an image, that is a very random instance like a Car or train or another human face that is not me.
what is the problem here?!**
### Reproduction
**T... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-05-01T09:10:45Z | 2023-05-01T09:21:43Z | 2023-05-01T09:21:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hosein-moayedi | 38,086,978 | MDQ6VXNlcjM4MDg2OTc4 | User | false |
huggingface/diffusers | 1,690,774,588 | I_kwDOHa8MBc5kxyw8 | 3,303 | https://github.com/huggingface/diffusers/issues/3303 | https://api.github.com/repos/huggingface/diffusers/issues/3303 | Using Modelscope: Error caught was: No module named 'triton' | ### Describe the bug
When calling the Modelscope model, I get the `Error caught was: No module named 'triton'` error. As I understand it is triton a MacOS only lib, and properly shouldn't be attempted to be reached from Windows?
Btw. for the Modelscope approach is there also this set watermark free and anime friend... | closed | completed | false | 12 | [
"bug",
"stale"
] | [] | 2023-05-01T12:55:41Z | 2023-08-21T15:04:10Z | 2023-08-21T15:04:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 1,690,835,010 | I_kwDOHa8MBc5kyBhC | 3,305 | https://github.com/huggingface/diffusers/issues/3305 | https://api.github.com/repos/huggingface/diffusers/issues/3305 | Latent couples in diffusers | Hi, I have been wondering if there's an equivalent of latent couples in diffusers. For those who don't know, here's a video on it: https://www.youtube.com/watch?v=uR89wZMXiJ8
This could be very useful for applying loras in different areas separately. Is there anything like this yet? | closed | completed | false | 1 | [] | [] | 2023-05-01T13:52:17Z | 2023-05-02T22:01:46Z | 2023-05-02T22:01:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,691,056,563 | I_kwDOHa8MBc5ky3mz | 3,307 | https://github.com/huggingface/diffusers/issues/3307 | https://api.github.com/repos/huggingface/diffusers/issues/3307 | How to use predicted_variance in deepfloyd for textual inversion tuning? | Im confused with 3 input and 6 output channels and not sure how to calculate loss
I checked pipeline code and example for textual inversion.
https://github.com/huggingface/diffusers/blob/eade4308dabc7f7ba75eab508d386b66b3764513/src/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py#L927
Seems like channel... | closed | completed | false | 8 | [] | [
"williamberman"
] | 2023-05-01T16:54:08Z | 2023-09-20T02:21:32Z | 2023-05-22T17:45:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hadaev8 | 20,247,085 | MDQ6VXNlcjIwMjQ3MDg1 | User | false |
huggingface/diffusers | 1,691,176,884 | I_kwDOHa8MBc5kzU-0 | 3,309 | https://github.com/huggingface/diffusers/issues/3309 | https://api.github.com/repos/huggingface/diffusers/issues/3309 | [FR] Multiclass Finetuning with Dreambooth | **Is your feature request related to a problem? Please describe.**
Hi! first of all thanks for such a great tool and all your hard work. Well at this moment I'm working with a partner @fededemo on a thesis related to data augmentation using synthetic images with generative models. We were fintuning things with dream... | closed | completed | false | 3 | [] | [] | 2023-05-01T18:44:25Z | 2023-06-16T01:02:42Z | 2023-05-03T17:33:31Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mauricio-repetto | 11,300,196 | MDQ6VXNlcjExMzAwMTk2 | User | false |
huggingface/diffusers | 1,693,069,714 | I_kwDOHa8MBc5k6jGS | 3,317 | https://github.com/huggingface/diffusers/issues/3317 | https://api.github.com/repos/huggingface/diffusers/issues/3317 | Trying to make a new Pipeline, but the scheduler doesn't apply the denoising step | Hello, I am creating a new pipeline for latent couple which is essentially a way to specify prompts on parts of the picture. Here's the code I'm trying to replicate:
https://colab.research.google.com/drive/1UdElpQfKFjY5luch9v_LlmSdH7AmeiDe?usp=sharing
I copied the latest StableDiffusionPipeline class and modified i... | closed | completed | false | 9 | [] | [] | 2023-05-02T20:38:21Z | 2023-12-19T08:02:02Z | 2023-05-03T15:10:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,360,254,085 | I_kwDOHa8MBc5RE9SF | 332 | https://github.com/huggingface/diffusers/issues/332 | https://api.github.com/repos/huggingface/diffusers/issues/332 | how to export stable fusion to onnx? | how to export stable fusion to onnx? | closed | completed | false | 1 | [] | [] | 2022-09-02T14:12:29Z | 2022-09-02T14:42:58Z | 2022-09-02T14:42:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | luohao123 | 49,749,220 | MDQ6VXNlcjQ5NzQ5MjIw | User | false |
huggingface/diffusers | 1,693,897,295 | I_kwDOHa8MBc5k9tJP | 3,324 | https://github.com/huggingface/diffusers/issues/3324 | https://api.github.com/repos/huggingface/diffusers/issues/3324 | How can i delete lastest output from pipe command? | Hello! I'm writing a program in Colab to use SD without any webui. I have most of it done but i want to delete lastest output made from "pipe(prompt, etc)", It exactly shows a progress bar that i can't handle like a normal Widget from Ipywidgets (I mean, replacing that progress bar instead of generating another one b... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-05-03T11:32:59Z | 2023-06-11T15:02:49Z | 2023-06-11T15:02:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Mudereded401 | 65,833,857 | MDQ6VXNlcjY1ODMzODU3 | User | false |
huggingface/diffusers | 1,694,004,214 | I_kwDOHa8MBc5k-HP2 | 3,325 | https://github.com/huggingface/diffusers/issues/3325 | https://api.github.com/repos/huggingface/diffusers/issues/3325 | RuntimeError: CUDA error: invalid argument when train_dreambooth.py | ### Describe the bug
I followed the instructions to train the dreambooth model, but the RuntimeError: CUDA error: invalid argument occured at accelerator.backward(loss)
My device is RTX-3090, and I will appreciate it if some one can help me solve the problem.
### Reproduction
warnings.warn(
05/03/2023 20:30:07 -... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-05-03T12:43:27Z | 2023-10-11T09:36:19Z | 2023-06-10T15:02:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sharkDDD | 42,758,995 | MDQ6VXNlcjQyNzU4OTk1 | User | false |
huggingface/diffusers | 1,694,585,706 | I_kwDOHa8MBc5lAVNq | 3,327 | https://github.com/huggingface/diffusers/issues/3327 | https://api.github.com/repos/huggingface/diffusers/issues/3327 | Could we make `BaseOutput` picklable? | Currently we can pickle but cannot recover from the pickled bytes:
```
In [40]: bs =pickle.dumps(output)
In [41]: pickle.loads(bs)
╭─────────────────────────────── Traceback (most recent call last) ────────────────────────────────╮
│ in <module>:1 ... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-05-03T18:29:55Z | 2023-06-22T21:47:05Z | 2023-06-11T15:02:48Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | larme | 66,847 | MDQ6VXNlcjY2ODQ3 | User | false |
huggingface/diffusers | 1,695,542,869 | I_kwDOHa8MBc5lD-5V | 3,329 | https://github.com/huggingface/diffusers/issues/3329 | https://api.github.com/repos/huggingface/diffusers/issues/3329 | Can inpaint load t2i's models? Or do I have to load a specific inpaint model | Can inpaint load t2i's models? Or do I have to load a specific inpaint model | closed | completed | false | 4 | [
"stale"
] | [] | 2023-05-04T08:42:17Z | 2023-06-11T15:02:47Z | 2023-06-11T15:02:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dingjingzhen | 21,244,263 | MDQ6VXNlcjIxMjQ0MjYz | User | false |
huggingface/diffusers | 1,696,718,822 | I_kwDOHa8MBc5lId_m | 3,332 | https://github.com/huggingface/diffusers/issues/3332 | https://api.github.com/repos/huggingface/diffusers/issues/3332 | canot import StableDiffusionSafetyChecker from diffusers.pipelines.stable_diffusion | ### Describe the bug
File "D:\generate\fonc\tammy\tammy-main\tammy-main\tammy\stable_diffusion.py", line 15, in <module>
from diffusers.pipelines.stable_diffusion import StableDiffusionSafetyChecker
ImportError: cannot import name 'StableDiffusionSafetyChecker' from 'diffusers.pipelines.stable_diffusion' (C:\U... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-05-04T21:18:50Z | 2023-12-04T16:56:01Z | 2023-06-12T15:02:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | saladinlorenz | 34,999,656 | MDQ6VXNlcjM0OTk5NjU2 | User | false |
huggingface/diffusers | 1,696,818,481 | I_kwDOHa8MBc5lI2Ux | 3,336 | https://github.com/huggingface/diffusers/issues/3336 | https://api.github.com/repos/huggingface/diffusers/issues/3336 | DPM-Solver Multistep Inverse Scheduler | ### Model/Pipeline/Scheduler description
As zero-shot image editing and interpolation techniques based on T2I diffusion models continue to proliferate, the use of deterministic inverse schedulers for deriving noise maps from user-provided images continues to grow. Implementing a higher-order inverse scheduler will hel... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-05-04T23:17:54Z | 2023-06-13T15:03:09Z | 2023-06-13T15:03:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clarencechen | 8,482,341 | MDQ6VXNlcjg0ODIzNDE= | User | false |
huggingface/diffusers | 1,697,054,467 | I_kwDOHa8MBc5lJv8D | 3,338 | https://github.com/huggingface/diffusers/issues/3338 | https://api.github.com/repos/huggingface/diffusers/issues/3338 | Importing diffusers takes >10 seconds | ### Describe the bug
Importing the `diffusers` package takes a long time in a Docker container (~13-15s). I've provided the Dockerfile, the test script I used to measure the import time, the corresponding output from the script, as well as exact package versions.
### Reproduction
Dockerfile:
```
FROM nvcr.i... | closed | completed | false | 7 | [
"bug",
"enhancement"
] | [] | 2023-05-05T05:58:57Z | 2024-11-21T07:30:48Z | 2024-11-21T07:30:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Anirudh-Scale | 88,404,749 | MDQ6VXNlcjg4NDA0NzQ5 | User | false |
huggingface/diffusers | 1,697,276,458 | I_kwDOHa8MBc5lKmIq | 3,342 | https://github.com/huggingface/diffusers/issues/3342 | https://api.github.com/repos/huggingface/diffusers/issues/3342 | Traning both text_encoder and unet with deepspeed zero opt. | Hi there,
I am currently attempting to train a txt2img model (both encoder and unet) using deepspeed. I have made some modifications to the code, but I am encountering an error. The error message indicates that there may be an issue with the backward function. Could you please advise me on how to correct the code?
... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-05-05T09:12:04Z | 2023-06-12T15:02:52Z | 2023-06-12T15:02:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | uygnef | 13,539,441 | MDQ6VXNlcjEzNTM5NDQx | User | false |
huggingface/diffusers | 1,697,980,905 | I_kwDOHa8MBc5lNSHp | 3,346 | https://github.com/huggingface/diffusers/issues/3346 | https://api.github.com/repos/huggingface/diffusers/issues/3346 | Support for resume_from_checkpoint in examples/dreambooth/train_dreambooth_lora.py | ### Describe the bug
In train_dreambooth_lora.py we still have accelerator.load_state(os.path.join(args.output_dir, path)) for resuming from a checkpoint, while we are not saving the state. This results in the following error: FileNotFoundError: [Errno 2] No such file or directory:
'/home/ec2-user/ssl/Jupyter/exp/... | closed | completed | false | 4 | [
"bug",
"stale"
] | [
"williamberman",
"patrickvonplaten"
] | 2023-05-05T17:53:31Z | 2023-06-13T15:03:07Z | 2023-06-13T15:03:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | abhijitpal1247 | 69,110,711 | MDQ6VXNlcjY5MTEwNzEx | User | false |
huggingface/diffusers | 1,698,061,038 | I_kwDOHa8MBc5lNlru | 3,347 | https://github.com/huggingface/diffusers/issues/3347 | https://api.github.com/repos/huggingface/diffusers/issues/3347 | DiffusionPipeline with text prompt outputs a array of nan as image | ### Describe the bug
Diffusion pipeline outputs nan array as an image when following quicktour instructions from huggingface.co/docs/diffusers. The error occurs for different models as well.
### Reproduction
### package installs:
```
conda install mamba
mamba install accelerate
mamba install -c huggingface trans... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-05-05T19:01:36Z | 2023-05-06T08:45:45Z | 2023-05-06T08:45:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dragonjinx | 39,952,058 | MDQ6VXNlcjM5OTUyMDU4 | User | false |
huggingface/diffusers | 1,698,176,033 | I_kwDOHa8MBc5lOBwh | 3,348 | https://github.com/huggingface/diffusers/issues/3348 | https://api.github.com/repos/huggingface/diffusers/issues/3348 | Support for 'scale' parameter when xformers is used for lora models | ### Describe the bug
Facing the following error while using xformers attention in case of a lora model and making use of the inference parameter cross_attention_kwargs = {'scale':0.5}. `TypeError: XFormersAttnProcessor.__call__() got an unexpected keyword argument 'scale'`
### Reproduction
pipe is an instance of Sta... | closed | completed | false | 10 | [
"bug"
] | [] | 2023-05-05T20:44:37Z | 2023-05-16T07:11:50Z | 2023-05-16T07:11:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | abhijitpal1247 | 69,110,711 | MDQ6VXNlcjY5MTEwNzEx | User | false |
huggingface/diffusers | 1,698,260,356 | I_kwDOHa8MBc5lOWWE | 3,349 | https://github.com/huggingface/diffusers/issues/3349 | https://api.github.com/repos/huggingface/diffusers/issues/3349 | Install instructions missing transformers | The installing instructions say to install diffusers, but not anything else.
So the first time people try the example they are going to get this error:
ImportError:
StableDiffusionPipeline requires the transformers library but it was not found in your environment. You can install it with pip: `pip
install transf... | closed | completed | false | 1 | [] | [] | 2023-05-05T22:11:31Z | 2023-05-11T15:52:30Z | 2023-05-11T15:52:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wrexbe | 81,056,464 | MDQ6VXNlcjgxMDU2NDY0 | User | false |
huggingface/diffusers | 1,698,266,343 | I_kwDOHa8MBc5lOXzn | 3,350 | https://github.com/huggingface/diffusers/issues/3350 | https://api.github.com/repos/huggingface/diffusers/issues/3350 | Errors on using animov-512x | ### Describe the bug
There are new weights for textvideo out: https://huggingface.co/strangeman3107/animov-512x
I used the previous version strangeman3107/animov-0.1.1 through Diffusers, and the worked fine, but the new ones gives me a lot of errors. See log. Are they Diffusers related?
### Reproduction
```
... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-05-05T22:20:59Z | 2023-05-06T04:26:22Z | 2023-05-06T04:26:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 1,698,515,326 | I_kwDOHa8MBc5lPUl- | 3,352 | https://github.com/huggingface/diffusers/issues/3352 | https://api.github.com/repos/huggingface/diffusers/issues/3352 | How to apply multi-vector textureal inversion embeds | ### Describe the bug
When using the official textural inversion [training script](https://github.com/huggingface/diffusers/blob/main/examples/textual_inversion/textual_inversion.py) with multi-vector, the output of the training is in the form of
```
{
placeholder_token: learned_embeds
}
```
`learned_embeds` i... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-05-06T07:27:29Z | 2023-05-10T02:48:18Z | 2023-05-10T02:48:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | haowang1013 | 7,006,155 | MDQ6VXNlcjcwMDYxNTU= | User | false |
huggingface/diffusers | 1,698,619,272 | I_kwDOHa8MBc5lPt-I | 3,354 | https://github.com/huggingface/diffusers/issues/3354 | https://api.github.com/repos/huggingface/diffusers/issues/3354 | Unet being cast to fp16 in LoRa training script despite contradicting comment | ### Describe the bug
[This comment](https://github.com/huggingface/diffusers/blob/0ffac97933d8ca1487d4ae9c374504a0979f0c5d/examples/text_to_image/train_text_to_image_lora.py#L419) states that "For mixed precision training we cast the text_encoder and vae weights to half-precision", but a few lines later, the Unet is a... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-05-06T12:51:43Z | 2024-06-06T07:54:45Z | 2023-06-13T15:03:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stefsietz | 2,621,937 | MDQ6VXNlcjI2MjE5Mzc= | User | false |
huggingface/diffusers | 1,698,695,510 | I_kwDOHa8MBc5lQAlW | 3,355 | https://github.com/huggingface/diffusers/issues/3355 | https://api.github.com/repos/huggingface/diffusers/issues/3355 | Inconsistent behavior of loading .safetensors models when setting use_safetensors=True | ### Describe the bug
The use of function `StableDiffusionPipeline.from_ckpt` is not consistent with the document w.r.t. loading .safetensors models. Specifically, calling this function to load a .safetensors model *and in the same time* setting `use_safetensors=True` will result in a ValueError saying `safetensors` is... | closed | completed | false | 9 | [
"bug",
"stale"
] | [
"sayakpaul",
"patrickvonplaten"
] | 2023-05-06T16:34:03Z | 2023-06-20T11:11:51Z | 2023-06-20T11:11:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | arthur-x | 30,208,088 | MDQ6VXNlcjMwMjA4MDg4 | User | false |
huggingface/diffusers | 1,698,867,152 | I_kwDOHa8MBc5lQqfQ | 3,356 | https://github.com/huggingface/diffusers/issues/3356 | https://api.github.com/repos/huggingface/diffusers/issues/3356 | How to fine-tune DeepFloyd IF? | It might be related to #3307.
I would like to try DreamBooth on DeepFloyd IF, but I found its pipeline seems slightly different from Stable Diffusion Pipeline.
https://github.com/huggingface/diffusers/blob/eade4308dabc7f7ba75eab508d386b66b3764513/src/diffusers/pipelines/deepfloyd_if/pipeline_if_img2img.py#L927
In ad... | closed | completed | false | 11 | [] | [] | 2023-05-07T03:07:44Z | 2023-09-17T02:51:53Z | 2023-05-09T19:12:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xingzhehe | 28,001,407 | MDQ6VXNlcjI4MDAxNDA3 | User | false |
huggingface/diffusers | 1,698,876,204 | I_kwDOHa8MBc5lQsss | 3,357 | https://github.com/huggingface/diffusers/issues/3357 | https://api.github.com/repos/huggingface/diffusers/issues/3357 | Torch Compile Error: torch._dynamo.exc.Unsupported: hasattr: TensorVariable() | ### Describe the bug
Getting this error:
```
0%| | 0/100 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/david/PycharmProjects/local_diffusion/deepfloyd.py", line 61, in <module>
image = pipe(
File "/home/david/.local/lib/python3.10/site-packages/torch/utils/_contextlib.py", li... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2023-05-07T03:48:59Z | 2023-06-19T15:03:04Z | 2023-06-19T15:03:04Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dblunk88 | 39,381,389 | MDQ6VXNlcjM5MzgxMzg5 | User | false |
huggingface/diffusers | 1,698,992,070 | I_kwDOHa8MBc5lRI_G | 3,358 | https://github.com/huggingface/diffusers/issues/3358 | https://api.github.com/repos/huggingface/diffusers/issues/3358 | How to remove lora after loading it? | The title pretty much. I am trying to load and unload inside the denoising loop to apply the loras to specific parts of the image only | closed | completed | false | 6 | [] | [] | 2023-05-07T10:11:08Z | 2023-05-10T09:14:25Z | 2023-05-08T13:22:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,360,766,761 | I_kwDOHa8MBc5RG6cp | 336 | https://github.com/huggingface/diffusers/issues/336 | https://api.github.com/repos/huggingface/diffusers/issues/336 | scheduler leaky abstractions in pipelines | Excited by the opportunities that Stable Diffusion has made for use with consumer hardware, I've started working with the code. I was looking to get an idea of what I'd need to do for something like #277 and studying the StableDiffusionPipeline class.
I'm concerned by the use of ` isinstance(self.scheduler, LMSDiscr... | closed | completed | false | 6 | [] | [] | 2022-09-03T05:52:20Z | 2022-10-05T12:41:20Z | 2022-10-05T12:41:20Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | keturn | 83,819 | MDQ6VXNlcjgzODE5 | User | false |
huggingface/diffusers | 1,699,676,472 | I_kwDOHa8MBc5lTwE4 | 3,363 | https://github.com/huggingface/diffusers/issues/3363 | https://api.github.com/repos/huggingface/diffusers/issues/3363 | train_dreambooth_lora.py failed on two machines | ### Describe the bug
I have found two errors.
1. when process save checkpoint
```
Traceback (most recent call last):
File "/home/momistest/db/diffusers/examples/dreambooth/train_dreambooth_lora.py", line 1112, in <module>
main(args)
File "/home/momistest/db/diffusers/examples/dreambooth/train_dreambo... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2023-05-08T07:13:33Z | 2023-06-15T15:03:16Z | 2023-06-15T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bohong13 | 43,626,613 | MDQ6VXNlcjQzNjI2NjEz | User | false |
huggingface/diffusers | 1,700,259,376 | I_kwDOHa8MBc5lV-Yw | 3,366 | https://github.com/huggingface/diffusers/issues/3366 | https://api.github.com/repos/huggingface/diffusers/issues/3366 | `LoRALinearLayer` stops autocast from working | ### Describe the bug
I encountered an issue while finetuning a UNet2DConditionModel and attempting to use lora, resulting in a significant decrease in speed, around 2.6 times slower than the original processing time per step.
Upon investigation, I discovered that this slowdown was caused by the xformers library runni... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-05-08T13:28:43Z | 2023-06-15T15:03:13Z | 2023-06-15T15:03:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Teoge | 10,808,856 | MDQ6VXNlcjEwODA4ODU2 | User | false |
huggingface/diffusers | 1,700,758,607 | I_kwDOHa8MBc5lX4RP | 3,367 | https://github.com/huggingface/diffusers/issues/3367 | https://api.github.com/repos/huggingface/diffusers/issues/3367 | Exception: Error while deserializing header: HeaderTooLarge | ### Describe the bug
There may be multiple reasons for the `Exception: Error while deserializing header: HeaderTooLarge` error message when loading a LoRA, but one of them is caused by passing in `/pytorch_lora_weights.bin` instead of just the root directory.
Suggestion: add a check to make certain the user only pa... | closed | completed | false | 5 | [
"bug"
] | [
"sayakpaul",
"patrickvonplaten"
] | 2023-05-08T18:58:08Z | 2024-01-15T15:18:13Z | 2023-05-21T09:56:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jelling | 1,887,486 | MDQ6VXNlcjE4ODc0ODY= | User | false |
huggingface/diffusers | 1,701,068,093 | I_kwDOHa8MBc5lZD09 | 3,369 | https://github.com/huggingface/diffusers/issues/3369 | https://api.github.com/repos/huggingface/diffusers/issues/3369 | ValueError: cross_attention_dim must be specified for CrossAttnUpBlock2D | ### Describe the bug
This is new and happening after saving the pipeline and then attempting to load it from pretrained, when it tries to load the vae, it is throwing an error for something which didnt used to throw an error. If the dim isnt specified, its because it didnt need to be specified, and if it was specified... | closed | completed | false | 14 | [
"bug",
"stale"
] | [] | 2023-05-08T23:11:58Z | 2023-10-17T05:07:58Z | 2023-06-17T15:02:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fractaldna22 | 37,323,518 | MDQ6VXNlcjM3MzIzNTE4 | User | false |
huggingface/diffusers | 1,360,794,497 | I_kwDOHa8MBc5RHBOB | 337 | https://github.com/huggingface/diffusers/issues/337 | https://api.github.com/repos/huggingface/diffusers/issues/337 | Problem with img2img & text guided inpainting | ### Describe the bug
When I try img2img & text guided inpainting，I got the same problems as follows. The text to image code works perfectly in the same environment.
### Reproduction
I git clone this repo and run the image_to_image.py as follows:
```python
from torch import autocast
import requests
import to... | closed | completed | false | 1 | [
"bug"
] | [] | 2022-09-03T08:17:23Z | 2022-09-03T08:39:01Z | 2022-09-03T08:39:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KevinGoodman | 43,786,278 | MDQ6VXNlcjQzNzg2Mjc4 | User | false |
huggingface/diffusers | 1,701,689,936 | I_kwDOHa8MBc5lbbpQ | 3,371 | https://github.com/huggingface/diffusers/issues/3371 | https://api.github.com/repos/huggingface/diffusers/issues/3371 | Diffusers Controlnet pipeline with textual inversion tokens | I was wondering whether Textual Inversion tokens are already supported for Diffusers Controlnet pipelines?
Would this be a lot of changes to the original (Multi)controlnet pipeline?
I'm also wondering whether it's easy to load multiple Textual Inversion concepts for a Diffusers pipeline.
If this is not yet possibl... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-05-09T09:20:10Z | 2023-06-18T15:03:06Z | 2023-06-18T15:03:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bertChristiaens1 | 70,344,362 | MDQ6VXNlcjcwMzQ0MzYy | User | false |
huggingface/diffusers | 1,701,856,941 | I_kwDOHa8MBc5lcEat | 3,372 | https://github.com/huggingface/diffusers/issues/3372 | https://api.github.com/repos/huggingface/diffusers/issues/3372 | [feat] raise exceptions on all network errors | *This is not a bug report - please do not ask for updates on a HF outage here. This is a feature request for the `diffusers` library to change how to handles 50X network issues.*
*The May 9 HF outage is a known issue. Suggest checking Twitter for status updates https://twitter.com/huggingface or visit HF's discord*
... | closed | completed | false | 12 | [] | [] | 2023-05-09T11:00:09Z | 2023-05-23T13:22:44Z | 2023-05-23T13:22:44Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | psychedelicious | 4,822,129 | MDQ6VXNlcjQ4MjIxMjk= | User | false |
huggingface/diffusers | 1,702,183,242 | I_kwDOHa8MBc5ldUFK | 3,375 | https://github.com/huggingface/diffusers/issues/3375 | https://api.github.com/repos/huggingface/diffusers/issues/3375 | Make diffusers independent from server | Hello diffusers team, we appreciate the libraries and your dedication. We understand that you are doing your best with the server issue, but it makes no sense why we need a server to run things locally.
Maybe try to allow better ways to use models that don't involve downloading an entire repository with files we don... | closed | completed | false | 9 | [
"stale"
] | [] | 2023-05-09T14:40:58Z | 2023-06-16T15:02:50Z | 2023-06-16T15:02:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,703,022,774 | I_kwDOHa8MBc5lghC2 | 3,377 | https://github.com/huggingface/diffusers/issues/3377 | https://api.github.com/repos/huggingface/diffusers/issues/3377 | [Question] About SNR weighting optimization during diffusion training | Hi,
Currently SNR weighting is only implemented for full text to image fine-tuning (in [train_text_to_image.py](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py#L838C26-L850)). I wonder if it makes sense to implement that optimization for other fine-tuning strategies (... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-05-10T02:43:52Z | 2023-06-18T15:03:05Z | 2023-06-18T15:03:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | haowang1013 | 7,006,155 | MDQ6VXNlcjcwMDYxNTU= | User | false |
huggingface/diffusers | 1,703,184,316 | I_kwDOHa8MBc5lhIe8 | 3,378 | https://github.com/huggingface/diffusers/issues/3378 | https://api.github.com/repos/huggingface/diffusers/issues/3378 | DeepFloyd/IF-I-XL-v1.0 process stage2 error | When I use demo code from https://huggingface.co/DeepFloyd/IF-I-XL-v1.0:
```
#stage 2
image = stage_2(
image=image,
prompt_embeds=prompt_embeds,
negative_prompt_embeds=negative_embeds,
generator=generator,
output_type="pt",
).images
```
It throwed an exception like:
```
image = stage_2(
File "/usr/local... | closed | completed | false | 1 | [] | [] | 2023-05-10T06:01:50Z | 2023-05-11T01:43:04Z | 2023-05-11T01:43:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sp-ljw | 85,480,243 | MDQ6VXNlcjg1NDgwMjQz | User | false |
huggingface/diffusers | 1,703,300,019 | I_kwDOHa8MBc5lhkuz | 3,379 | https://github.com/huggingface/diffusers/issues/3379 | https://api.github.com/repos/huggingface/diffusers/issues/3379 | Dreambooth can not load intermedia checkpoint | ### Describe the bug
I am training dreambooth and want to use the intermedia checkpoint-500 following [this](https://huggingface.co/docs/diffusers/training/dreambooth#:~:text=v1%2D4%22%0A%0Aunet%20%3D-,UNet2DConditionModel,-.from_pretrained(%22/sddata).
However it cannot load the submodule unet:
### Reproducti... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-05-10T07:33:26Z | 2023-06-18T15:03:03Z | 2023-06-18T15:03:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | g-jing | 44,223,191 | MDQ6VXNlcjQ0MjIzMTkx | User | false |
huggingface/diffusers | 1,360,806,664 | I_kwDOHa8MBc5RHEMI | 338 | https://github.com/huggingface/diffusers/issues/338 | https://api.github.com/repos/huggingface/diffusers/issues/338 | [Tests] Add tests for LMS Discrete | Similar to other tests we should add tests for the LMS Discrete Scheduler in https://github.com/huggingface/diffusers/blob/main/tests/test_scheduler.py | closed | completed | false | 6 | [
"Good second issue"
] | [] | 2022-09-03T09:24:21Z | 2023-10-02T16:45:39Z | 2023-10-02T16:45:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,703,308,786 | I_kwDOHa8MBc5lhm3y | 3,380 | https://github.com/huggingface/diffusers/issues/3380 | https://api.github.com/repos/huggingface/diffusers/issues/3380 | embedding_size is hard coded and might be wrong | https://github.com/huggingface/diffusers/blob/94a0c644a8ce5b05a969859e0814ef4883ac870e/src/diffusers/models/unet_1d.py#L100 | closed | not_planned | false | 3 | [] | [] | 2023-05-10T07:39:33Z | 2023-06-06T16:58:51Z | 2023-06-06T16:58:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JayL0321 | 31,190,549 | MDQ6VXNlcjMxMTkwNTQ5 | User | false |
huggingface/diffusers | 1,703,482,808 | I_kwDOHa8MBc5liRW4 | 3,381 | https://github.com/huggingface/diffusers/issues/3381 | https://api.github.com/repos/huggingface/diffusers/issues/3381 | Discoloured output in StableDiffusionInpaintPipeline batch inference + possible docs error | ### Describe the bug
When I do batch inference using the StableDiffusionInpaintPipeline (inputting the images/masks as multidimensional tensors), I get discolored/pale output and the inpainted area seems to bear little relation to the inputted mask. Example:
Input image:
<img src="https://github.com/huggingface/... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-05-10T09:20:21Z | 2023-06-17T15:02:47Z | 2023-06-17T15:02:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | BenjaminIrwin | 59,652,119 | MDQ6VXNlcjU5NjUyMTE5 | User | false |
huggingface/diffusers | 1,703,610,175 | I_kwDOHa8MBc5liwc_ | 3,382 | https://github.com/huggingface/diffusers/issues/3382 | https://api.github.com/repos/huggingface/diffusers/issues/3382 | train_text_to_image.py multi_gpu training cuda out of memory error but sufficient memory when using single GPU | ### Describe the bug
When using the `train_text_to_image.py` example script on a single NVIDIA A10G GPU the example script works great. However, when using 4xNVIDIA A10G if I use the same input arguments but use the `--multi_gpu` accelerate flag all 4 of the GPUs run out of memory before the first step is complete.
... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-05-10T10:32:29Z | 2023-07-24T19:41:32Z | 2023-05-11T11:26:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TrueWheelProgramming | 133,107,666 | U_kgDOB-8P0g | User | false |
huggingface/diffusers | 1,703,761,986 | I_kwDOHa8MBc5ljVhC | 3,383 | https://github.com/huggingface/diffusers/issues/3383 | https://api.github.com/repos/huggingface/diffusers/issues/3383 | pipeline.from_pretrained is broken when the pipeline is partially downloaded | ### Describe the bug
Hi,
There's a bug in [pipeline_utils.py](https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/pipeline_utils.py) which causes `pipeline.from_pretrained` to fail if the pipeline was partially downloaded. Specifically the code doesn't handle missing components in the `feat... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2023-05-10T11:56:15Z | 2023-06-18T15:03:02Z | 2023-06-18T15:03:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | haowang1013 | 7,006,155 | MDQ6VXNlcjcwMDYxNTU= | User | false |
huggingface/diffusers | 1,704,362,326 | I_kwDOHa8MBc5lloFW | 3,385 | https://github.com/huggingface/diffusers/issues/3385 | https://api.github.com/repos/huggingface/diffusers/issues/3385 | DeepFloyd-IF in JAX | ### Model/Pipeline/Scheduler description
The new [DeepFloyd-IF model](https://huggingface.co/DeepFloyd/IF-I-XL-v1.0) looks really good! Hope to see it implemented in Jax to run on TPUs :)
### Open source status
- [X] The model implementation is available
- [X] The model weights are available (Only relevant if additi... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-05-10T17:33:00Z | 2023-07-01T15:03:11Z | 2023-07-01T15:03:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yuvalkirstain | 57,996,478 | MDQ6VXNlcjU3OTk2NDc4 | User | false |
huggingface/diffusers | 1,360,806,863 | I_kwDOHa8MBc5RHEPP | 339 | https://github.com/huggingface/diffusers/issues/339 | https://api.github.com/repos/huggingface/diffusers/issues/339 | [Tests] Add Tests for Karras VE | Similar to other tests we should add tests for the Karras VE Scheduler in https://github.com/huggingface/diffusers/blob/main/tests/test_scheduler.py | closed | completed | false | 4 | [
"good first issue"
] | [
"anton-l"
] | 2022-09-03T09:25:33Z | 2023-03-18T18:41:42Z | 2023-03-18T18:41:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,705,010,847 | I_kwDOHa8MBc5loGaf | 3,390 | https://github.com/huggingface/diffusers/issues/3390 | https://api.github.com/repos/huggingface/diffusers/issues/3390 | Can diffuser load local safetensors files? | ### Describe the bug
I want to directly load a stablediffusion base safetensors model locally , but I found that it seems to only support the repository format. Is there any way to make it load the local model?
... | closed | completed | false | 27 | [
"bug",
"stale"
] | [] | 2023-05-11T03:43:28Z | 2024-06-28T04:36:21Z | 2023-06-18T15:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sanbuphy | 96,160,062 | U_kgDOBbtJPg | User | false |
huggingface/diffusers | 1,705,084,058 | I_kwDOHa8MBc5loYSa | 3,391 | https://github.com/huggingface/diffusers/issues/3391 | https://api.github.com/repos/huggingface/diffusers/issues/3391 | DeepFloyd super-resolution pipeline generates overexposure image for PIL.Image low-res input | ### Describe the bug
Super-resolution pipeline for deepfloyd model generates overexposure image when the low-res input image is an `PIL.Image` object.
Low-resolution input:
Super-resolution result:
![image... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-05-11T05:17:57Z | 2023-05-27T22:44:44Z | 2023-05-27T22:44:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Question406 | 15,529,162 | MDQ6VXNlcjE1NTI5MTYy | User | false |
huggingface/diffusers | 1,705,131,336 | I_kwDOHa8MBc5loj1I | 3,392 | https://github.com/huggingface/diffusers/issues/3392 | https://api.github.com/repos/huggingface/diffusers/issues/3392 | How do we use multiple GPUs to generate a single image? | I am trying to use multiple GPUs to generate a single image. Not batch images in parallel like #2977
I want my images to generate fast and not be bottlenecked by memory constants when I generate larger images or attempt in/out painting. I tried to use deepspeed; however, wsl2 is insanely slow and deepspeed just doesn... | closed | completed | false | 14 | [
"stale"
] | [] | 2023-05-11T06:08:26Z | 2025-08-28T00:13:37Z | 2023-08-08T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JemiloII | 5,192,788 | MDQ6VXNlcjUxOTI3ODg= | User | false |
huggingface/diffusers | 1,705,271,996 | I_kwDOHa8MBc5lpGK8 | 3,393 | https://github.com/huggingface/diffusers/issues/3393 | https://api.github.com/repos/huggingface/diffusers/issues/3393 | How can I convert A111 .safetensors format lora to .bin format lora? | Due to current issues loading A111 safetensors, https://github.com/huggingface/diffusers/issues/3064
I want to load it at runtime by converting it to the bin format. I used the script (https://github.com/haofanwang/Lora-for-Diffusers) but found that it could not fully convert the weights to bin format, resulting in a ... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-05-11T07:55:16Z | 2023-06-18T15:02:59Z | 2023-06-18T15:02:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sanbuphy | 96,160,062 | U_kgDOBbtJPg | User | false |
huggingface/diffusers | 1,705,315,755 | I_kwDOHa8MBc5lpQ2r | 3,394 | https://github.com/huggingface/diffusers/issues/3394 | https://api.github.com/repos/huggingface/diffusers/issues/3394 | Which script is for traing latent_diffusion_uncond？ | I want to training unconditional image generation model of latent diffusion. I find the script "/examples/unconditional_image_generation/train_unconditional.py". However, it seems train Unet by raw images but not latent features.
Which script is for traing latent_diffusion_uncond？
Thank you. | closed | completed | false | 1 | [] | [] | 2023-05-11T08:23:00Z | 2023-05-11T14:35:56Z | 2023-05-11T14:35:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shenwenxiao | 48,246,620 | MDQ6VXNlcjQ4MjQ2NjIw | User | false |
huggingface/diffusers | 1,705,589,240 | I_kwDOHa8MBc5lqTn4 | 3,395 | https://github.com/huggingface/diffusers/issues/3395 | https://api.github.com/repos/huggingface/diffusers/issues/3395 | [Configuration] Allowing to load configs with AutoConfig | `transformers` exposes [`AutoConfig` classes](https://huggingface.co/docs/transformers/v4.29.0/en/model_doc/auto#transformers.AutoConfig) that allow users to load a model configuration and reuse that:
```python
from transformers import AutoConfig
config = AutoConfig.from_pretrained("bert-base-uncased")
config
... | closed | completed | false | 5 | [] | [
"sayakpaul"
] | 2023-05-11T11:07:06Z | 2023-05-16T13:11:56Z | 2023-05-16T13:11:56Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,705,592,972 | I_kwDOHa8MBc5lqUiM | 3,396 | https://github.com/huggingface/diffusers/issues/3396 | https://api.github.com/repos/huggingface/diffusers/issues/3396 | Make Loras also modify text embeddings | Apparently, A1111 also makes Loras modify the embeddings as you can see in the example here: https://civitai.com/models/8252/gal-gadotlora the trigger word is gldot which isn't part of the general training data of SD1.5. Also, @takuma104 confirmed that to be the case. Now, I'll be completely honest. I have no idea what... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-05-11T11:08:48Z | 2023-07-05T15:04:00Z | 2023-07-05T15:04:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,706,345,190 | I_kwDOHa8MBc5ltMLm | 3,401 | https://github.com/huggingface/diffusers/issues/3401 | https://api.github.com/repos/huggingface/diffusers/issues/3401 | class_labels should be provided when num_class_embeds > 0 | ### Describe the bug
Running into this issue and not entirely sure what i am doing wrong, do I need a metadata file or something?
### Reproduction
!python3 train_dreambooth_lora.py \
--pretrained_model_name_or_path="DeepFloyd/IF-II-L-v1.0" \
--output_dir="/content/drive/MyDrive/stable_diffusion_weights/zwx23" ... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-05-11T18:37:44Z | 2023-07-25T01:16:51Z | 2023-07-25T01:16:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stpg06 | 96,386,710 | U_kgDOBb6-lg | User | false |
huggingface/diffusers | 1,706,558,692 | I_kwDOHa8MBc5luATk | 3,403 | https://github.com/huggingface/diffusers/issues/3403 | https://api.github.com/repos/huggingface/diffusers/issues/3403 | TorchDynamo with Pipeline loading | ### Describe the bug
Hi! I am using accelerate + diffusers for fine-tuning of stable diffusion using this script:
https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py
I have pytorch 2.0 and accelerate recently supports torch dynamo for improving speed of training.
Howeve... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-05-11T21:06:52Z | 2023-06-19T15:02:59Z | 2023-06-19T15:02:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thepowerfuldeez | 11,796,343 | MDQ6VXNlcjExNzk2MzQz | User | false |
huggingface/diffusers | 1,706,843,165 | I_kwDOHa8MBc5lvFwd | 3,405 | https://github.com/huggingface/diffusers/issues/3405 | https://api.github.com/repos/huggingface/diffusers/issues/3405 | A bug happen in controlnet training process | ### Describe the bug
Exception has occurred: AssertionError
exception: no description
File "/root/autodl-tmp/diffusers_per_exp_5/train_control.py", line 122, in log_validation
image = pipeline(
File "/root/autodl-tmp/diffusers_per_exp_5/train_control.py", line 1125, in main
image_logs = log_validation... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-05-12T02:59:56Z | 2023-05-14T09:21:40Z | 2023-05-14T09:21:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LeoooooMM | 88,709,183 | MDQ6VXNlcjg4NzA5MTgz | User | false |
huggingface/diffusers | 1,706,983,060 | I_kwDOHa8MBc5lvn6U | 3,406 | https://github.com/huggingface/diffusers/issues/3406 | https://api.github.com/repos/huggingface/diffusers/issues/3406 | The Attention Class is not working for only_cross_attention = True | https://github.com/huggingface/diffusers/blob/1a5797c6d4491a879ea5285c4efc377664e0332d/src/diffusers/models/attention_processor.py#L128
self.to_k and self.to_v will be none if the attention object is initialized with only_cross_attention = True | closed | completed | false | 6 | [] | [] | 2023-05-12T06:01:51Z | 2025-05-23T15:43:38Z | 2023-06-12T17:11:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | JayL0321 | 31,190,549 | MDQ6VXNlcjMxMTkwNTQ5 | User | false |
huggingface/diffusers | 1,707,016,200 | I_kwDOHa8MBc5lvwAI | 3,407 | https://github.com/huggingface/diffusers/issues/3407 | https://api.github.com/repos/huggingface/diffusers/issues/3407 | ControlNet v1.1.150 works worse than previous version makes blurry images | after update automatic1111 started to generate blurry images maybe someone knows what to do in this case, before images was sharper and more contrast, even if the input is low quality, but right now doesn't matter how much quality is your inout image the result is blurry and with low details | closed | completed | false | 3 | [
"stale"
] | [] | 2023-05-12T06:25:29Z | 2023-06-19T15:02:58Z | 2023-06-19T15:02:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | i4imaru | 114,496,149 | U_kgDOBtMSlQ | User | false |
huggingface/diffusers | 1,707,058,782 | I_kwDOHa8MBc5lv6Ze | 3,408 | https://github.com/huggingface/diffusers/issues/3408 | https://api.github.com/repos/huggingface/diffusers/issues/3408 | Replace deprecated `set-output` command with environment file | **Describe the issue**
In workflow, `set-output` command is deprecated and will be disabled soon. Please upgrade to using Environment Files. For more information, see: [https://github.blog/changelog/2022-10-11-github-actions-deprecating-save-state-and-set-output-commands/](https://github.blog/changelog/2022-10-11-gith... | closed | completed | false | 0 | [] | [] | 2023-05-12T07:03:35Z | 2023-05-16T11:51:11Z | 2023-05-16T11:51:11Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jongwooo | 44,025,432 | MDQ6VXNlcjQ0MDI1NDMy | User | false |
huggingface/diffusers | 1,360,817,924 | I_kwDOHa8MBc5RHG8E | 341 | https://github.com/huggingface/diffusers/issues/341 | https://api.github.com/repos/huggingface/diffusers/issues/341 | K-LMS scheduler for in-paint | **Is your feature request related to a problem? Please describe.**
The K-LMS scheduler currently doesn't work for Stable Diffusion In-painting.
It would be nice to make it work as also asked for in #329
cc @anton-l
**Additional context**
Add any other context or screenshots about the feature request here... | closed | completed | false | 0 | [] | [] | 2022-09-03T10:24:31Z | 2022-09-20T17:10:45Z | 2022-09-20T17:10:45Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,707,296,485 | I_kwDOHa8MBc5lw0bl | 3,411 | https://github.com/huggingface/diffusers/issues/3411 | https://api.github.com/repos/huggingface/diffusers/issues/3411 | [Tests] Training DreamBooth with LoRA with text encoder enabled | From [this](https://github.com/huggingface/diffusers/actions/runs/4955396756/jobs/8864770968?pr=3370) CI run:
```bash
E AttributeError: 'UNet2DConditionModel' object has no attribute 'text_model'
```
But this only seems to be an issue when `accelerate` is built from source. If we install the latest st... | closed | completed | false | 8 | [] | [] | 2023-05-12T09:35:31Z | 2023-05-15T03:03:10Z | 2023-05-15T03:03:10Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,707,357,442 | I_kwDOHa8MBc5lxDUC | 3,412 | https://github.com/huggingface/diffusers/issues/3412 | https://api.github.com/repos/huggingface/diffusers/issues/3412 | [Docs] Update https://huggingface.co/docs/diffusers/stable_diffusion#next-steps. Writing it down here so I don't forget. | Context: https://github.com/huggingface/diffusers/issues/3370#issuecomment-1545511069 | closed | completed | false | 0 | [] | [
"pcuenca"
] | 2023-05-12T10:14:45Z | 2023-05-16T18:35:49Z | 2023-05-16T18:35:49Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,708,038,784 | I_kwDOHa8MBc5lzpqA | 3,418 | https://github.com/huggingface/diffusers/issues/3418 | https://api.github.com/repos/huggingface/diffusers/issues/3418 | add text-encoder part in train_text_to_image_lora.py | **Is your feature request related to a problem? Please describe.**
The train_text_to_image_lora.py example is still using the old lora training format that only includes the UNET part.
**Describe the solution you'd like**
It would be nice to add the text-encoder part of training and to be inferenced with the load... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-05-12T18:29:08Z | 2023-07-04T15:03:15Z | 2023-07-04T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MikeHanKK | 123,093,985 | U_kgDOB1ZD4Q | User | false |
huggingface/diffusers | 1,708,108,185 | I_kwDOHa8MBc5lz6mZ | 3,420 | https://github.com/huggingface/diffusers/issues/3420 | https://api.github.com/repos/huggingface/diffusers/issues/3420 | StableDiffusionControlNetInpaintPipeline doesn't work with Diffusion pipeline | ### Describe the bug
DiffusionPipeline does not expect control_image argument
### Reproduction
```
DiffusionPipeline.from_pretrained(
pretrained_model_name_or_path=stable_model_path,
controlnet=controlnet,
custom_pipeline="stable_diffusion_controlnet_inpaint",
safety_checker=None,
torch_dty... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-05-12T19:31:15Z | 2023-05-12T19:41:44Z | 2023-05-12T19:41:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmanKishore | 10,477,380 | MDQ6VXNlcjEwNDc3Mzgw | User | false |
huggingface/diffusers | 1,708,153,138 | I_kwDOHa8MBc5l0Fky | 3,421 | https://github.com/huggingface/diffusers/issues/3421 | https://api.github.com/repos/huggingface/diffusers/issues/3421 | Model not loading during second attempt | ### Describe the bug
SD model doesn't/takes too long to load the second time. The model is stored in GCS bucket and accessed using gcsfuse in GKE. It works fine for the first load but not after that.
### Reproduction
```
In [3]: import torch
...: from diffusers import StableDiffusionInpaintPipeline
In [4]:... | closed | completed | false | 8 | [
"bug"
] | [] | 2023-05-12T20:12:45Z | 2024-07-29T04:00:28Z | 2023-06-01T13:41:41Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hari10599 | 37,787,894 | MDQ6VXNlcjM3Nzg3ODk0 | User | false |
huggingface/diffusers | 1,708,208,076 | I_kwDOHa8MBc5l0S_M | 3,422 | https://github.com/huggingface/diffusers/issues/3422 | https://api.github.com/repos/huggingface/diffusers/issues/3422 | How to disable safety checker | **Is your feature request related to a problem? Please describe.**
Yes, sometimes when i run inference, it generate an all black photos. I suspected this is due to the safety checker
**Describe the solution you'd like**
An option to disable safety checker
**Describe alternatives you've considered**
I dont know... | closed | completed | false | 6 | [] | [] | 2023-05-12T21:01:41Z | 2023-07-30T00:33:32Z | 2023-06-01T12:36:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MaxTran96 | 15,274,946 | MDQ6VXNlcjE1Mjc0OTQ2 | User | false |
huggingface/diffusers | 1,708,344,264 | I_kwDOHa8MBc5l00PI | 3,423 | https://github.com/huggingface/diffusers/issues/3423 | https://api.github.com/repos/huggingface/diffusers/issues/3423 | Stuck at launch of code for creation of model |
Hi all,
I'm stuck when trying to run the code to create the model. it renders the below:
2023-05-11 10:07:31.040186: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
/usr/local/lib/python3.10/dist-packages/accelerate/accelerator.py:258: FutureWarning: logging_di... | closed | completed | false | 3 | [] | [] | 2023-05-12T23:46:58Z | 2023-06-01T12:39:10Z | 2023-06-01T12:39:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anthonysom | 16,055,144 | MDQ6VXNlcjE2MDU1MTQ0 | User | false |
huggingface/diffusers | 1,708,641,398 | I_kwDOHa8MBc5l18x2 | 3,427 | https://github.com/huggingface/diffusers/issues/3427 | https://api.github.com/repos/huggingface/diffusers/issues/3427 | makeavid-sd-jax | ### Model/Pipeline/Scheduler description
New text2video project:
https://huggingface.co/spaces/TempoFunk/makeavid-sd-jax
https://github.com/lopho/makeavid-sd-tpu
### Open source status
- [x] The model implementation is available
- [x] The model weights are available (Only relevant if addition is not a scheduler).
... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-05-13T14:59:22Z | 2023-06-20T15:02:49Z | 2023-06-20T15:02:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 1,708,882,865 | I_kwDOHa8MBc5l23ux | 3,428 | https://github.com/huggingface/diffusers/issues/3428 | https://api.github.com/repos/huggingface/diffusers/issues/3428 | An unexpected result when I try to change the StableDiffusionImg2ImgPipeline | ### Describe the bug
I'm not sure I could call it a bug.
The difference is a modification based on the example using StableDiffusionImg2ImgPipeline
`pipe.unet = UNet2DConditionModel.from_config(pipe.unet.config).to(device)`
But the two pieces of code give very different results, especially if I insert the ... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-05-14T09:48:00Z | 2023-06-21T15:03:00Z | 2023-06-21T15:03:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | LeoooooMM | 88,709,183 | MDQ6VXNlcjg4NzA5MTgz | User | false |
huggingface/diffusers | 1,708,907,218 | I_kwDOHa8MBc5l29rS | 3,429 | https://github.com/huggingface/diffusers/issues/3429 | https://api.github.com/repos/huggingface/diffusers/issues/3429 | does diffusers have the equivalent to hires fix from A1111? | hi, does diffusers have this: https://github.com/Kahsolt/stable-diffusion-webui-hires-fix-progressive
I realized that there's a lot of other features that I was not aware of when seeing this comment about how to get high quality results: https://www.reddit.com/r/StableDiffusion/comments/13gr5rg/comment/jk2g3vd/?utm_... | closed | completed | false | 57 | [
"stale"
] | [] | 2023-05-14T11:17:21Z | 2024-03-17T12:04:59Z | 2023-11-08T15:10:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,708,933,189 | I_kwDOHa8MBc5l3EBF | 3,430 | https://github.com/huggingface/diffusers/issues/3430 | https://api.github.com/repos/huggingface/diffusers/issues/3430 | Multi-GPU training OOM | I'm trying to execute the training for multi-gpu case with the following config:
```
compute_environment: LOCAL_MACHINE
distributed_type: MULTI_GPU
downcast_bf16: 'no'
gpu_ids: 1,2
machine_rank: 0
main_training_function: main
mixed_precision: fp16
num_machines: 1
num_processes: 2
rdzv_backend: static
same_n... | closed | completed | false | 1 | [] | [] | 2023-05-14T12:48:49Z | 2023-05-30T11:35:10Z | 2023-05-30T11:35:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xenia-kra | 81,739,495 | MDQ6VXNlcjgxNzM5NDk1 | User | false |
huggingface/diffusers | 1,709,080,593 | I_kwDOHa8MBc5l3oAR | 3,431 | https://github.com/huggingface/diffusers/issues/3431 | https://api.github.com/repos/huggingface/diffusers/issues/3431 | Inpainting+ControlNet poor results with canny or hed | ### Describe the bug
I'm using the stable_diffusion_controlnet_inpaint pipeline - I know this is a community pipeline, but I'm filing here as it seems likely this will be added to the main src at some point.
Pretty much all the controlnet methods work as expected, except canny and hed produce weird results that I... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-05-14T21:20:35Z | 2023-05-17T10:03:43Z | 2023-05-16T20:44:40Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | reimager | 20,918,094 | MDQ6VXNlcjIwOTE4MDk0 | User | false |
huggingface/diffusers | 1,709,109,178 | I_kwDOHa8MBc5l3u-6 | 3,432 | https://github.com/huggingface/diffusers/issues/3432 | https://api.github.com/repos/huggingface/diffusers/issues/3432 | examples/text_to_image RuntimeError: "slow_conv2d_cpu" not implemented for 'Half' | ### Describe the bug
Following the ``examples/text_to_image`` README leads to a reproducible RuntimeError. RuntimeError: "slow_conv2d_cpu" not implemented for 'Half'.
https://github.com/huggingface/diffusers/tree/main/examples/text_to_image
### Reproduction
```
git clone https://github.com/huggingface/diff... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2023-05-14T23:22:00Z | 2024-02-15T06:25:10Z | 2023-07-04T15:03:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zdwolfe | 719,730 | MDQ6VXNlcjcxOTczMA== | User | false |
huggingface/diffusers | 1,709,712,658 | I_kwDOHa8MBc5l6CUS | 3,434 | https://github.com/huggingface/diffusers/issues/3434 | https://api.github.com/repos/huggingface/diffusers/issues/3434 | [Text-to-video Zero] improvements in the attention processor with PT 2.0 | @19and99 is it possible to use the new efficient `scaled_dot_product_attention` to perform the attention computation [here](https://github.com/huggingface/diffusers/blob/7a32b6beeb0cfdefed645253dce23d9b0a78597f/src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_zero.py#L39)? This essentially reduces... | closed | completed | false | 0 | [] | [] | 2023-05-15T09:42:28Z | 2023-06-12T16:03:20Z | 2023-06-12T16:03:20Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,711,152,890 | I_kwDOHa8MBc5l_h76 | 3,441 | https://github.com/huggingface/diffusers/issues/3441 | https://api.github.com/repos/huggingface/diffusers/issues/3441 | Higher VRAM usage with PyTorch2 without xformers under certain situations | ### Describe the bug
_Edited to reflect the actual issue_
In the latest development version, PR #3365 introduced a confusion which makes users believe that PT2 variant of attention processors are fully supported and `xformers` is no longer needed. This results in higher VRAM usage under certain situations (using Lo... | closed | completed | false | 3 | [
"bug"
] | [
"sayakpaul"
] | 2023-05-16T03:15:08Z | 2023-05-17T06:34:45Z | 2023-05-17T06:34:08Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wfng92 | 43,742,196 | MDQ6VXNlcjQzNzQyMTk2 | User | false |
huggingface/diffusers | 1,711,202,772 | I_kwDOHa8MBc5l_uHU | 3,442 | https://github.com/huggingface/diffusers/issues/3442 | https://api.github.com/repos/huggingface/diffusers/issues/3442 | Textual Inversion training example produces embeddings that don't work: incompatible token dimension: 768 vs 1. | ### Describe the bug
I am new and trying to use the textual inversion [tutorial](https://huggingface.co/docs/diffusers/training/text_inversion). I followed all steps as close as possible. After a few hours training finishes, but when trying to load the result in InvokeAI I get the error:
**Notice: cattu/learned_emb... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-05-16T04:10:28Z | 2023-06-24T15:02:55Z | 2023-06-24T15:02:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Barafu | 6,612,364 | MDQ6VXNlcjY2MTIzNjQ= | User | false |
huggingface/diffusers | 1,711,708,316 | I_kwDOHa8MBc5mBpic | 3,444 | https://github.com/huggingface/diffusers/issues/3444 | https://api.github.com/repos/huggingface/diffusers/issues/3444 | Integration of ImageBind and StableUnCLIPImg2ImgPipeline for audio2image generation | ### Model/Pipeline/Scheduler description
For anyone who need, here is a simple demo to illustrate how to integrate ImageBind and StableUnCLIPImg2ImgPipeline for audio2image generation.
### Open source status
- [X] The model implementation is available
- [X] The model weights are available (Only relevant if ad... | open | null | false | 4 | [
"New pipeline/model",
"community-examples"
] | [] | 2023-05-16T10:15:38Z | 2024-11-22T01:31:51Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zeqiang-Lai | 26,198,430 | MDQ6VXNlcjI2MTk4NDMw | User | false |
huggingface/diffusers | 1,711,818,432 | I_kwDOHa8MBc5mCEbA | 3,445 | https://github.com/huggingface/diffusers/issues/3445 | https://api.github.com/repos/huggingface/diffusers/issues/3445 | dreambooth lora: text encoder monkey-patching not working | ### Describe the bug
In the dreambooth lora script, the lora layers are monkey patched into the text encoder as the text encoder comes from transformers. However, this has some undesirable results:
This is b... | closed | completed | false | 6 | [
"bug",
"stale"
] | [
"sayakpaul"
] | 2023-05-16T11:23:08Z | 2023-06-16T17:59:51Z | 2023-06-16T17:59:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rvorias | 21,060,408 | MDQ6VXNlcjIxMDYwNDA4 | User | false |
huggingface/diffusers | 1,712,488,743 | I_kwDOHa8MBc5mEoEn | 3,450 | https://github.com/huggingface/diffusers/issues/3450 | https://api.github.com/repos/huggingface/diffusers/issues/3450 | StableDiffusionPipeline.from_ckpt is not working on dev version | ### Describe the bug
It seems that from_ckpt is not working properly in the latest main. It seems to be fine in release version v0.16.1. I tried to follow some of the changes, but couldn't track them all.
diff: https://github.com/huggingface/diffusers/compare/9b14ce397e53fc5f5b909b07b6e992a2afe8e3af...main
### Rep... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-05-16T18:02:47Z | 2023-06-28T12:49:44Z | 2023-05-17T14:42:36Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | takuma104 | 10,776 | MDQ6VXNlcjEwNzc2 | User | false |
huggingface/diffusers | 1,712,607,590 | I_kwDOHa8MBc5mFFFm | 3,451 | https://github.com/huggingface/diffusers/issues/3451 | https://api.github.com/repos/huggingface/diffusers/issues/3451 | train_text_to_image.py | fp16 uses more memory than fp32 | ### Describe the bug
Setting --mixed_precision="fp16" takes more VRAM memory (26235MiB) than --mixed_precision="no" (25361MiB).
### Reproduction
export MODEL_NAME="CompVis/stable-diffusion-v1-4"
export dataset_name="lambdalabs/pokemon-blip-captions"
accelerate launch --mixed_precision="fp16" train_text_t... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-05-16T19:21:56Z | 2023-06-24T15:02:54Z | 2023-06-24T15:02:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | morgankohler | 53,620,460 | MDQ6VXNlcjUzNjIwNDYw | User | false |
huggingface/diffusers | 1,712,611,464 | I_kwDOHa8MBc5mFGCI | 3,452 | https://github.com/huggingface/diffusers/issues/3452 | https://api.github.com/repos/huggingface/diffusers/issues/3452 | training dreambooth with 4 gb vram | ### Describe the bug
I tried to run dreambooth with my 4 gb gpu and I couldn't make it run. I used all the optimization techniques but still it couldn't run. I tired both dreambooth and lora
### Reproduction
Code
https://github.com/huggingface/diffusers/tree/main/examples/dreambooth
### Logs
_No response_
###... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-05-16T19:25:07Z | 2023-06-24T15:02:53Z | 2023-06-24T15:02:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pure-rgb | 45,315,076 | MDQ6VXNlcjQ1MzE1MDc2 | User | false |
huggingface/diffusers | 1,712,617,054 | I_kwDOHa8MBc5mFHZe | 3,453 | https://github.com/huggingface/diffusers/issues/3453 | https://api.github.com/repos/huggingface/diffusers/issues/3453 | train_text_to_image.py | RuntimeError: "triu_tril_cuda_template" not implemented for 'BFloat16' | ### Describe the bug
When running train_text_to_image.py, setting --mixed_precision="bf16" causes an error in the transformers clip model. I am opening this here as I am not sure how to reproduce this from the transformers repo.
### Reproduction
#!/bin/bash
export MODEL_NAME="CompVis/stable-diffusion-v1-4"
export... | closed | completed | false | 7 | [
"bug"
] | [] | 2023-05-16T19:29:13Z | 2023-06-01T20:10:26Z | 2023-06-01T20:10:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | morgankohler | 53,620,460 | MDQ6VXNlcjUzNjIwNDYw | User | false |
huggingface/diffusers | 1,712,664,676 | I_kwDOHa8MBc5mFTBk | 3,454 | https://github.com/huggingface/diffusers/issues/3454 | https://api.github.com/repos/huggingface/diffusers/issues/3454 | train_dreambooth_lora.py Does not run on latest | ### Describe the bug
train_dreambooth_lora.py example no longer runs when installing diffusers from source.
If I run the script using the provided snippet, I get the following error.
Command run:
accelerate launch train_dreambooth_lora.py --pretrained_model_name_or_path="runwayml/stable-diffusion-v1-5" --instanc... | closed | completed | false | 11 | [
"bug"
] | [] | 2023-05-16T20:03:09Z | 2024-04-25T14:44:02Z | 2023-05-22T17:55:23Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Ambrosiussen | 23,486,196 | MDQ6VXNlcjIzNDg2MTk2 | User | false |
huggingface/diffusers | 1,712,702,443 | I_kwDOHa8MBc5mFcPr | 3,455 | https://github.com/huggingface/diffusers/issues/3455 | https://api.github.com/repos/huggingface/diffusers/issues/3455 | Use pretrained instruct pix2pix in training of instruct pix2pix | Can I use "timbrooks/instruct-pix2pix" model as the pre-trained weight in [train_instruct_pix2pix.py](https://github.com/huggingface/diffusers/blob/main/examples/instruct_pix2pix/train_instruct_pix2pix.py) code? | closed | completed | false | 8 | [] | [] | 2023-05-16T20:33:19Z | 2023-12-21T05:05:36Z | 2023-06-01T12:27:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mahdip72 | 42,680,708 | MDQ6VXNlcjQyNjgwNzA4 | User | false |
huggingface/diffusers | 1,712,862,156 | I_kwDOHa8MBc5mGDPM | 3,456 | https://github.com/huggingface/diffusers/issues/3456 | https://api.github.com/repos/huggingface/diffusers/issues/3456 | ClipGuidedDiffusion lacks prompt latents and negative latents, making it impossible to use with lpw | As follows from the title: ClipGuidedDiffusion pipeline lacks latents inputs, so it is not possible at the moment to get the long prompt, convert it with LPW pipeline and put latents into clip-guided pipeline.
**Describe the solution you'd like**
• add support of latents for this pipeline
• make prompt an optiona... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-05-16T23:21:29Z | 2023-06-24T15:02:52Z | 2023-06-24T15:02:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Pirog17000 | 3,022,852 | MDQ6VXNlcjMwMjI4NTI= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.