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 | 2,167,264,220 | I_kwDOHa8MBc6BLdPc | 7,210 | https://github.com/huggingface/diffusers/issues/7210 | https://api.github.com/repos/huggingface/diffusers/issues/7210 | 'StableDiffusionXLPipeline' object has no attribute 'named_parameters' when using accelerate.load_checkpoint_and_dispatch | ### Describe the bug
How can I make the stable diffusion xl model to work with the `init_empty_weights` API? I've followed https://huggingface.co/docs/accelerate/v0.11.0/en/big_modeling but still got an unexpected error:
### Reproduction
This is what I got so far:
```python
import torch
from diffusers import Di... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-03-04T16:34:27Z | 2024-04-19T15:01:29Z | 2024-04-19T15:01:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thiagocrepaldi | 5,469,809 | MDQ6VXNlcjU0Njk4MDk= | User | false |
huggingface/diffusers | 2,167,944,490 | I_kwDOHa8MBc6BODUq | 7,214 | https://github.com/huggingface/diffusers/issues/7214 | https://api.github.com/repos/huggingface/diffusers/issues/7214 | DPMSolverMultistepScheduler using karras_sigmas doesn't work anymore | ### Describe the bug
I had modified the comfyui extension from Zuellni https://github.com/Zuellni/ComfyUI-Custom-Nodes
to enable karras_sigmas for the deepfloyd IF model.
This has stopped working a few months ago, i get an error when enabling karras_sigmas now:
ERROR:root:Traceback (most recent call last):
... | closed | completed | false | 13 | [
"bug"
] | [
"yiyixuxu"
] | 2024-03-04T23:03:32Z | 2024-04-05T17:29:57Z | 2024-04-04T22:20:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Bigfield77 | 108,305,548 | U_kgDOBnScjA | User | false |
huggingface/diffusers | 2,168,198,149 | I_kwDOHa8MBc6BPBQF | 7,217 | https://github.com/huggingface/diffusers/issues/7217 | https://api.github.com/repos/huggingface/diffusers/issues/7217 | stabilityai/stable-diffusion-2-base DDIM config is not compatible with current DDIM implementation | ### Describe the bug
Assume 30 inference steps.
There are two 925, and the number of inference steps is 31.
tensor([958, 925, 925, 892, 859, 826, 793, 760, 727, 694, 661, 628, 595, 562,
529, 496, 463, 430, 397, 364, 331, 298, 265, 232, 199, 166, 133, 100,
67, 34, 1], device='cuda:0')
```
... | open | null | false | 3 | [
"bug",
"stale",
"needs-code-example"
] | [] | 2024-03-05T03:08:21Z | 2024-04-04T15:02:37Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zeqiang-Lai | 26,198,430 | MDQ6VXNlcjI2MTk4NDMw | User | false |
huggingface/diffusers | 2,168,724,917 | I_kwDOHa8MBc6BRB21 | 7,219 | https://github.com/huggingface/diffusers/issues/7219 | https://api.github.com/repos/huggingface/diffusers/issues/7219 | Add SUPIR Upscaler | ### Model/Pipeline/Scheduler description
[SUPIR](https://supir.xpixel.group/) is a super-resolution model that looks like it produces excellent results
Github Repo: https://github.com/Fanghua-Yu/SUPIR
The model is quite memory intensive, so the optimisation features available in diffusers might be quite helpful ... | open | null | false | 36 | [
"help wanted",
"New pipeline/model",
"contributions-welcome"
] | [] | 2024-03-05T09:32:38Z | 2025-11-12T17:46:28Z | null | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DN6 | 7,529,846 | MDQ6VXNlcjc1Mjk4NDY= | User | false |
huggingface/diffusers | 1,396,669,532 | I_kwDOHa8MBc5TP3xc | 722 | https://github.com/huggingface/diffusers/issues/722 | https://api.github.com/repos/huggingface/diffusers/issues/722 | ¿Can i save all images of generation process? | I would like to save each image generated at each step of the image generation process.
¿Is possible? | closed | completed | false | 2 | [
"stale"
] | [] | 2022-10-04T18:20:08Z | 2022-11-13T15:02:57Z | 2022-11-13T15:02:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elcejo | 55,639,486 | MDQ6VXNlcjU1NjM5NDg2 | User | false |
huggingface/diffusers | 2,168,956,691 | I_kwDOHa8MBc6BR6cT | 7,220 | https://github.com/huggingface/diffusers/issues/7220 | https://api.github.com/repos/huggingface/diffusers/issues/7220 | .unload_ip_adapter() return error for StableDiffusionControlNetInpaintPipeline | ### Describe the bug
After using .unload_ip_adapter() for StableDiffusionControlNetInpaintPipeline, it returns an error.
### Reproduction
```
inpaint = StableDiffusionControlNetInpaintPipeline.from_pretrained(
"runwayml/stable-diffusion-inpainting",
controlnet = controlnet,
va... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-03-05T11:26:08Z | 2024-03-09T04:02:19Z | 2024-03-09T04:02:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wilfrediscoming | 137,744,070 | U_kgDOCDXOxg | User | false |
huggingface/diffusers | 2,169,070,752 | I_kwDOHa8MBc6BSWSg | 7,221 | https://github.com/huggingface/diffusers/issues/7221 | https://api.github.com/repos/huggingface/diffusers/issues/7221 | The picture is full of noise use playground-v2.5-1024px-aesthetic.fp16.safetensors | ### Describe the bug
with playground-v2.5-1024px-aesthetic.fp16.safetensors
### Reproduction
```
from diffusers import StableDiffusionXLPipeline
pipeline = StableDiffusionXLPipeline.from_single_file(
... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-03-05T12:23:13Z | 2024-03-07T10:01:05Z | 2024-03-07T10:01:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyun0071 | 35,762,050 | MDQ6VXNlcjM1NzYyMDUw | User | false |
huggingface/diffusers | 2,169,603,874 | I_kwDOHa8MBc6BUYci | 7,223 | https://github.com/huggingface/diffusers/issues/7223 | https://api.github.com/repos/huggingface/diffusers/issues/7223 | Add Latte: Latent Diffusion Transformer for Video Generation | ### Model/Pipeline/Scheduler description
Latte is a text2video diffusion transformer (similar to Sora), improving past the DiT and PixArt-alpha text2image models
The implementation is already based on diffusers (see [latte_t2v.py](https://github.com/Vchitect/Latte/blob/main/models/latte_t2v.py)), so adding it here ... | closed | completed | false | 5 | [
"stale"
] | [] | 2024-03-05T16:03:52Z | 2024-07-19T09:44:19Z | 2024-07-19T09:44:19Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,169,630,229 | I_kwDOHa8MBc6BUe4V | 7,224 | https://github.com/huggingface/diffusers/issues/7224 | https://api.github.com/repos/huggingface/diffusers/issues/7224 | Denoise latents after predicting `model pred` | Hi, I'm focusing on training controlnet:
https://github.com/huggingface/diffusers/blob/687bc2772721af584d649129f8d2a28ca56a9ad8/examples/controlnet/train_controlnet.py#L1063-L1072
And I want to know how can I get the denoised image using the `model_pred`, I prefer to get the denoised rgb image tensor in single denoi... | closed | completed | false | 1 | [] | [] | 2024-03-05T16:16:11Z | 2024-03-06T04:49:32Z | 2024-03-06T04:49:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | OrangeSodahub | 54,439,582 | MDQ6VXNlcjU0NDM5NTgy | User | false |
huggingface/diffusers | 2,170,287,867 | I_kwDOHa8MBc6BW_b7 | 7,228 | https://github.com/huggingface/diffusers/issues/7228 | https://api.github.com/repos/huggingface/diffusers/issues/7228 | when loading a traced model, get error AttributeError: 'StableDiffusionPipeline' object has no attribute 'dtype' | ### Describe the bug
when loading a pretraced model using torch.jit.load and setting it to pipeline.unet, I get the following error:
lib\site-packages\diffusers\configuration_utils.py", line 142, in __getattr__
raise AttributeError(f"'{type(self).__name__}' object has no attribute '{name}'")
AttributeError: '... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2024-03-05T23:02:24Z | 2024-06-30T05:38:06Z | 2024-06-30T05:37:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | peymoon | 61,159,755 | MDQ6VXNlcjYxMTU5NzU1 | User | false |
huggingface/diffusers | 2,170,299,587 | I_kwDOHa8MBc6BXCTD | 7,229 | https://github.com/huggingface/diffusers/issues/7229 | https://api.github.com/repos/huggingface/diffusers/issues/7229 | Advanced training has an issue when saving checkpoints | ### Describe the bug
I'm running both
https://github.com/huggingface/diffusers/blob/main/examples/advanced_diffusion_training/train_dreambooth_lora_sd15_advanced.py
and
https://github.com/huggingface/diffusers/blob/main/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py
and they ... | closed | completed | false | 7 | [
"bug"
] | [
"linoytsaban"
] | 2024-03-05T23:11:30Z | 2025-03-20T18:12:44Z | 2024-03-12T14:07:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | landmann | 12,756,081 | MDQ6VXNlcjEyNzU2MDgx | User | false |
huggingface/diffusers | 1,396,725,963 | I_kwDOHa8MBc5TQFjL | 723 | https://github.com/huggingface/diffusers/issues/723 | https://api.github.com/repos/huggingface/diffusers/issues/723 | Fast generation of smaller images with the option to generate the large 512x512 version later | Hello!
I am afraid that there could be two technical reasons that my request could not be solved. But may be I am wrong.
As everybody knows not every generated image is really good. And the generation time for a 512x512px image is rather long. (3 seconds on a A100, okay, but who have one..) So I think it would be s... | closed | completed | false | 5 | [
"stale"
] | [] | 2022-10-04T19:10:01Z | 2022-11-13T15:02:56Z | 2022-11-13T15:02:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Marcophono2 | 22,599,855 | MDQ6VXNlcjIyNTk5ODU1 | User | false |
huggingface/diffusers | 2,170,788,137 | I_kwDOHa8MBc6BY5kp | 7,231 | https://github.com/huggingface/diffusers/issues/7231 | https://api.github.com/repos/huggingface/diffusers/issues/7231 | Applying Blockwise Weights to LoRA | It would be wonderful if I could apply stepwise or blockwise weights to LoRA, just like the webui extension below.
https://github.com/hako-mikan/sd-webui-lora-block-weight
Does anyone know if it's also possible in diffusers??
| closed | completed | false | 3 | [
"contributions-welcome"
] | [] | 2024-03-06T07:00:22Z | 2024-03-29T15:45:58Z | 2024-03-29T15:45:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dain5832 | 62,231,006 | MDQ6VXNlcjYyMjMxMDA2 | User | false |
huggingface/diffusers | 2,170,898,567 | I_kwDOHa8MBc6BZUiH | 7,232 | https://github.com/huggingface/diffusers/issues/7232 | https://api.github.com/repos/huggingface/diffusers/issues/7232 | Support for Multimodal Diffusion Transformers (e.g. StableDiffusion3) | ### Model/Pipeline/Scheduler description
Yesterday StabilityAI published the details of their architecture MMDiT for the upcoming StableDiffusion3.
https://stability.ai/news/stable-diffusion-3-research-paper
Their approach differs quite much from the traditional Diffuser Transformers (like PixArt-alpha) in a way... | closed | completed | false | 15 | [
"stale"
] | [] | 2024-03-06T08:16:25Z | 2024-06-30T05:37:43Z | 2024-06-30T05:37:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,171,864,280 | I_kwDOHa8MBc6BdATY | 7,237 | https://github.com/huggingface/diffusers/issues/7237 | https://api.github.com/repos/huggingface/diffusers/issues/7237 | EDMEulerScheduler is not found by latest version 🤔 | ### Describe the bug
Running https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_sdxl.py doesn't work because `EDMEulerScheduler` can't be found!
### Reproduction
Run `pip install git+https://github.com/huggingface/diffusers.git`
Run `https://github.com/huggingface/dif... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-06T16:04:13Z | 2024-03-06T16:49:06Z | 2024-03-06T16:48:32Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | landmann | 12,756,081 | MDQ6VXNlcjEyNzU2MDgx | User | false |
huggingface/diffusers | 2,172,078,933 | I_kwDOHa8MBc6Bd0tV | 7,238 | https://github.com/huggingface/diffusers/issues/7238 | https://api.github.com/repos/huggingface/diffusers/issues/7238 | IP-Adapter masking | Continuing the issue from [here](https://github.com/huggingface/diffusers/pull/7226#discussion_r1513786107) about assigning a separate input image to each IP-Adapter without passing a mask. @sayakpaul suspects it's because the images need to have the exact same resolution. cc @yiyixuxu
Code to reproduce:
```py
... | closed | completed | false | 2 | [] | [] | 2024-03-06T17:38:12Z | 2024-03-06T21:24:18Z | 2024-03-06T21:24:18Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stevhliu | 59,462,357 | MDQ6VXNlcjU5NDYyMzU3 | User | false |
huggingface/diffusers | 2,172,199,888 | I_kwDOHa8MBc6BeSPQ | 7,239 | https://github.com/huggingface/diffusers/issues/7239 | https://api.github.com/repos/huggingface/diffusers/issues/7239 | Dreambooth SDXL LoRA - mat1 and mat2 shapes cannot be multiplied (2x2048 and 2816x1280) | ### Describe the bug
Trying to run an sdxl lora dreambooth training job with prior preservation. After class images are generated, it dies with the logged error, `mat1 and mat2 shapes cannot be multiplied (2x2048 and 2816x1280)`. Instance images are 37 photos of my dog in a variety of sizes
### Reproduction
``... | closed | completed | false | 5 | [
"bug",
"training"
] | [] | 2024-03-06T18:51:52Z | 2025-01-14T07:43:50Z | 2024-03-07T09:49:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shawnrushefsky | 6,685,047 | MDQ6VXNlcjY2ODUwNDc= | User | false |
huggingface/diffusers | 2,172,402,515 | I_kwDOHa8MBc6BfDtT | 7,240 | https://github.com/huggingface/diffusers/issues/7240 | https://api.github.com/repos/huggingface/diffusers/issues/7240 | Convert ImageNet conditional LDM to Diffusers get low-quality output | ### Describe the bug
I tried to convert the pre-trained ImageNet conditional LDM to diffusers, as in other conversion scripts.
I checked the converted results with the same latent, the schedulers and class embedders have the same behavior.
But the UNet model produces slightly different results.
![image](https... | closed | completed | false | 0 | [
"bug"
] | [] | 2024-03-06T20:50:30Z | 2024-03-06T23:35:12Z | 2024-03-06T23:35:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Hhhhhhao | 39,571,115 | MDQ6VXNlcjM5NTcxMTE1 | User | false |
huggingface/diffusers | 2,173,445,743 | I_kwDOHa8MBc6BjCZv | 7,243 | https://github.com/huggingface/diffusers/issues/7243 | https://api.github.com/repos/huggingface/diffusers/issues/7243 | ResAdapter: Domain Consistent Resolution Adapter for Diffusion Models | ### Model/Pipeline/Scheduler description
ResAdapter support image generation in arbitrary resolution for comunity models. The result is quite interesting
Project page: https://res-adapter.github.io/
Github: https://github.com/bytedance/res-adapter
### Open source status
- [X] The model implementation is avai... | closed | completed | false | 11 | [] | [] | 2024-03-07T09:51:31Z | 2024-11-20T00:29:30Z | 2024-11-20T00:29:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rootonchair | 23,548,268 | MDQ6VXNlcjIzNTQ4MjY4 | User | false |
huggingface/diffusers | 2,173,838,368 | I_kwDOHa8MBc6BkiQg | 7,249 | https://github.com/huggingface/diffusers/issues/7249 | https://api.github.com/repos/huggingface/diffusers/issues/7249 | [Diffusers] [SDv1.5 - Text-to-Image Finetuning] Lack of a validation loss mechanism in Stable Diffusion training. | ### Discussed in https://github.com/huggingface/diffusers/discussions/7248
<div type='discussions-op-text'>
<sup>Originally posted by **VietAnhNguyen20** March 7, 2024</sup>
# Search before asking:
- I have thoroughly reviewed the Diffusers and Huggingface documentation, the open or closed issues on github
, ... | closed | completed | false | 1 | [
"should-move-to-discussion"
] | [] | 2024-03-07T13:14:13Z | 2024-03-11T04:16:26Z | 2024-03-11T04:16:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | VietAnhNguyen20 | 75,252,999 | MDQ6VXNlcjc1MjUyOTk5 | User | false |
huggingface/diffusers | 1,396,846,782 | I_kwDOHa8MBc5TQjC- | 725 | https://github.com/huggingface/diffusers/issues/725 | https://api.github.com/repos/huggingface/diffusers/issues/725 | add device map and accelerate to DiffusionPipeline abstraction to reduce memory footprint when loading model | **Is your feature request related to a problem? Please describe.**
As a follow up for #281 we could add the device map and the possibility to load weights using `accelerate` to `DiffusionPipeline` abstraction for smaller memory footprint when loading models.
**Describe the solution you'd like**
```python
from d... | closed | completed | false | 9 | [] | [] | 2022-10-04T21:09:44Z | 2024-04-27T07:11:45Z | 2022-10-14T17:28:36Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | piEsposito | 47,679,710 | MDQ6VXNlcjQ3Njc5NzEw | User | false |
huggingface/diffusers | 2,174,355,406 | I_kwDOHa8MBc6BmgfO | 7,250 | https://github.com/huggingface/diffusers/issues/7250 | https://api.github.com/repos/huggingface/diffusers/issues/7250 | About SDXL text_encoder training | Mistaken issue sry.
| closed | completed | false | 0 | [] | [] | 2024-03-07T17:15:35Z | 2024-03-08T01:42:50Z | 2024-03-08T01:42:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rmac85 | 90,567,763 | MDQ6VXNlcjkwNTY3NzYz | User | false |
huggingface/diffusers | 2,175,086,921 | I_kwDOHa8MBc6BpTFJ | 7,253 | https://github.com/huggingface/diffusers/issues/7253 | https://api.github.com/repos/huggingface/diffusers/issues/7253 | Being able to input noisy init latents to StableDiffusionPipeline | At the moment, two seeds are completely orthogonal to another, i.e. seed=1 and seed=2 have nothing to do with another and are as different as seed=1 and seed=12891371. In Automatic1111, one can choose a subseed, which is basically the amount of morphing it does from the noisy pattern of one seed to another. It would be... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-03-08T00:59:58Z | 2024-11-28T05:40:14Z | 2024-11-28T05:40:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nicollegah | 55,440,280 | MDQ6VXNlcjU1NDQwMjgw | User | false |
huggingface/diffusers | 2,175,112,393 | I_kwDOHa8MBc6BpZTJ | 7,254 | https://github.com/huggingface/diffusers/issues/7254 | https://api.github.com/repos/huggingface/diffusers/issues/7254 | Request proper examples on how to training a diffusion models with diffusers on large scale dataset like LAION | Hi, I do not see any examples in diffusers/examples on how to training a diffusion models with diffusers on large scale dataset like LAION. However, it is important since many works and models is willing integrate their models into diffusers, so if they can train their models in diffusers, it would be more easy when t... | closed | completed | false | 7 | [
"stale"
] | [] | 2024-03-08T01:31:33Z | 2024-06-30T05:27:57Z | 2024-06-30T05:27:57Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Luciennnnnnn | 20,135,317 | MDQ6VXNlcjIwMTM1MzE3 | User | false |
huggingface/diffusers | 2,175,176,363 | I_kwDOHa8MBc6Bpo6r | 7,255 | https://github.com/huggingface/diffusers/issues/7255 | https://api.github.com/repos/huggingface/diffusers/issues/7255 | PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Accelerator | ### Model/Pipeline/Scheduler description
A competitor of LCM-Lora, with higher generation quality and consistency regardless number of steps
Project page: https://piecewise-rectified-flow.github.io/
Github: https://github.com/magic-research/piecewise-rectified-flow/tree/main
### Open source status
- [X] The mo... | open | null | false | 7 | [
"community-examples",
"Good second issue",
"contributions-welcome"
] | [] | 2024-03-08T02:50:45Z | 2024-11-20T00:28:24Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rootonchair | 23,548,268 | MDQ6VXNlcjIzNTQ4MjY4 | User | false |
huggingface/diffusers | 2,175,210,516 | I_kwDOHa8MBc6BpxQU | 7,256 | https://github.com/huggingface/diffusers/issues/7256 | https://api.github.com/repos/huggingface/diffusers/issues/7256 | StableCascadePriorPipeline has no attribute 'from_single_file' | Are there plans to support `from_single_file` in the StableCascade pipelines? I would greatly appreciate it it's more convenient for me.
Edit - Do the same for StableCascadePriorPipeline and StableCascadeDecoderPipeline
Thanks! | closed | completed | false | 8 | [
"stale",
"single_file"
] | [] | 2024-03-08T03:33:15Z | 2024-05-03T21:37:25Z | 2024-05-03T21:37:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ballenvironment | 111,252,136 | U_kgDOBqGSqA | User | false |
huggingface/diffusers | 1,396,952,405 | I_kwDOHa8MBc5TQ81V | 726 | https://github.com/huggingface/diffusers/issues/726 | https://api.github.com/repos/huggingface/diffusers/issues/726 | [Community] Move the number "0.18215" from the image2image process to VAE config | ## There is a magic number "0.18215" in the repository
In the file `src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_img2img.py`, there is a number "0.18215" in line 220 and line 342, which is strange since it does occur in the original [repo](https://github.com/CompVis/stable-diffusion/blob/69ae4b... | closed | completed | false | 12 | [
"good first issue",
"stale"
] | [
"patil-suraj"
] | 2022-10-04T23:19:49Z | 2023-01-26T13:38:37Z | 2023-01-26T13:38:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wangyu-ustc | 44,958,517 | MDQ6VXNlcjQ0OTU4NTE3 | User | false |
huggingface/diffusers | 2,177,209,435 | I_kwDOHa8MBc6BxZRb | 7,261 | https://github.com/huggingface/diffusers/issues/7261 | https://api.github.com/repos/huggingface/diffusers/issues/7261 | Unsuccessful cross-attention weight loading in Custom Diffusion | ### Describe the bug
If you have PEFT installed in your environment, then custom_diffusion will not successfully load the cross-attention parameter, leading to a poor generation result. Given the time cost of troubleshooting this issue, the documentation should state that the currently implemented code is incompatible... | closed | completed | false | 12 | [
"bug",
"stale"
] | [] | 2024-03-09T12:33:35Z | 2024-06-30T05:29:11Z | 2024-06-30T05:29:10Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Rbrq03 | 115,004,993 | U_kgDOBtrWQQ | User | false |
huggingface/diffusers | 2,177,329,986 | I_kwDOHa8MBc6Bx2tC | 7,263 | https://github.com/huggingface/diffusers/issues/7263 | https://api.github.com/repos/huggingface/diffusers/issues/7263 | Error using IP Adapter with enable_attention_slicing() | ### Describe the bug
The error occurs after loading an IP Adapter and then calling `pipeline.enable_attention_slicing()` afterword. It only works if you don't enable attention slicing
### Reproduction
```py
ip_adapter_path = os.path.join(os.path.dirname(__file__), "ipadapter")
pipeline.load_ip_adapter(ip_adapter_p... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2024-03-09T17:49:09Z | 2025-12-31T01:42:44Z | 2024-06-30T05:30:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ballenvironment | 111,252,136 | U_kgDOBqGSqA | User | false |
huggingface/diffusers | 2,177,447,356 | I_kwDOHa8MBc6ByTW8 | 7,264 | https://github.com/huggingface/diffusers/issues/7264 | https://api.github.com/repos/huggingface/diffusers/issues/7264 | `DiffusionPipeline` cannot serialize or load custom huggingface `transformers` model | **Is your feature request related to a problem? Please describe.**
Currently `DiffusionPipeline` class can only serialize official transformers models correctly and load them using `from_pretrained`. However, it does not work for custom huggingface models which could be serialized using `transformers` and loaded using... | closed | completed | false | 11 | [
"stale"
] | [] | 2024-03-09T23:43:01Z | 2024-06-30T05:36:28Z | 2024-06-30T05:36:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ChenchaoZhao | 35,147,961 | MDQ6VXNlcjM1MTQ3OTYx | User | false |
huggingface/diffusers | 2,177,534,151 | I_kwDOHa8MBc6ByojH | 7,265 | https://github.com/huggingface/diffusers/issues/7265 | https://api.github.com/repos/huggingface/diffusers/issues/7265 | Kohya Hires fix | is diffusers possible to support this hires fix?
it looks 1.5 work too
https://github.com/AUTOMATIC1111/stable-diffusion-webui/issues/13974
https://www.youtube.com/watch?v=SbgMwHDXthU
same seed at 1024x1024
with
![image](https://github.com/huggingface/diffusers/assets/1147704/7319ef5f-1581-407d-8ea7... | closed | completed | false | 25 | [
"good first issue",
"help wanted",
"community-examples",
"contributions-welcome"
] | [] | 2024-03-10T05:29:07Z | 2024-06-08T02:53:35Z | 2024-05-28T20:00:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | crapthings | 1,147,704 | MDQ6VXNlcjExNDc3MDQ= | User | false |
huggingface/diffusers | 2,177,633,651 | I_kwDOHa8MBc6BzA1z | 7,266 | https://github.com/huggingface/diffusers/issues/7266 | https://api.github.com/repos/huggingface/diffusers/issues/7266 | Cascade w/ new varnames produces funky images. | ### Describe the bug
Using Stable Cascade with the diffusers main branch and the HF PRs 2/44 as intended, it appears as though there's some misconfiguration that results in super funky images as shown.
Old Cascade branch before the variable renaming, using the current main Cascade revisions
![00021](https://github... | closed | completed | false | 13 | [
"bug"
] | [
"DN6"
] | 2024-03-10T09:59:01Z | 2024-03-28T07:15:40Z | 2024-03-28T07:15:40Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Beinsezii | 39,478,211 | MDQ6VXNlcjM5NDc4MjEx | User | false |
huggingface/diffusers | 2,177,637,102 | I_kwDOHa8MBc6BzBru | 7,267 | https://github.com/huggingface/diffusers/issues/7267 | https://api.github.com/repos/huggingface/diffusers/issues/7267 | train_controlnet.py broken local data loading | ### Describe the bug
The script `examples/controlnet/train_controlnet.py` is designed to either take a HF dataset with `--dataset_name` (as illustrated [here](https://huggingface.co/blog/train-your-controlnet)) or local data `--train_data_dir`, but the latter case fails out-of-the-box.
### Reproduction
Running scrip... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2024-03-10T10:08:10Z | 2024-11-23T15:27:46Z | 2024-11-23T15:27:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | morrisalp | 8,263,996 | MDQ6VXNlcjgyNjM5OTY= | User | false |
huggingface/diffusers | 2,178,702,354 | I_kwDOHa8MBc6B3FwS | 7,276 | https://github.com/huggingface/diffusers/issues/7276 | https://api.github.com/repos/huggingface/diffusers/issues/7276 | [Dreambooth example run error] When I use the downloaded pretrained_model I still get network errors | Hello,it's a great work!
I have downloaded the pre-trained model in advance. My question is why there are still network errors during the training process?
My windows system environment is as follows：
```bash
transformers version: 4.38.2
accelerate version: 0.27.2
peft version: 0.9.1.dev0
pytorch v... | closed | completed | false | 8 | [
"stale"
] | [] | 2024-03-10T12:05:44Z | 2024-06-30T05:29:41Z | 2024-06-30T05:29:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xdobetter | 26,523,963 | MDQ6VXNlcjI2NTIzOTYz | User | false |
huggingface/diffusers | 1,397,077,445 | I_kwDOHa8MBc5TRbXF | 728 | https://github.com/huggingface/diffusers/issues/728 | https://api.github.com/repos/huggingface/diffusers/issues/728 | IMG2IMG -> ValueError: At least one stride in the given numpy array is negative | ### Describe the bug
I am getting the following error with IMG2IMG
`ValueError: At least one stride in the given numpy array is negative, and tensors with negative strides are not currently supported. (You can probably work around this by making a copy of your array with array.copy().) `
The help topics onlin... | closed | completed | false | 4 | [
"bug"
] | [] | 2022-10-05T02:04:51Z | 2022-10-06T01:36:02Z | 2022-10-06T01:35:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WASasquatch | 1,151,589 | MDQ6VXNlcjExNTE1ODk= | User | false |
huggingface/diffusers | 2,180,548,406 | I_kwDOHa8MBc6B-Ic2 | 7,283 | https://github.com/huggingface/diffusers/issues/7283 | https://api.github.com/repos/huggingface/diffusers/issues/7283 | How to load lora trained with Stable Cascade? | I finished a lora traning based on Stable Cascade with onetrainer, but I cannot find a solution to load the load in diffusers pipeline. Anyone who can help me will be appreciated. | closed | completed | false | 5 | [
"stale"
] | [] | 2024-03-12T01:33:01Z | 2024-06-29T13:35:45Z | 2024-06-29T13:35:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zengjie617789 | 25,050,291 | MDQ6VXNlcjI1MDUwMjkx | User | false |
huggingface/diffusers | 2,180,943,122 | I_kwDOHa8MBc6B_o0S | 7,286 | https://github.com/huggingface/diffusers/issues/7286 | https://api.github.com/repos/huggingface/diffusers/issues/7286 | Unsuccessful cross-attention weight loading in Custom Diffusion with StableDiffusionAttendAndExcitePipeline | ### Describe the bug
I load the trained pytorch_custom_diffusion_weights.bin using StableDiffusionAttendAndExcitePipeline, however, loading is failed. I am wondering if is it the same cause with this issue (https://github.com/huggingface/diffusers/issues/7261#issuecomment-1990962268) .
### Reproduction
pipeline = ... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2024-03-12T07:45:45Z | 2024-06-30T05:28:56Z | 2024-06-30T05:28:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | daeunni | 62,705,839 | MDQ6VXNlcjYyNzA1ODM5 | User | false |
huggingface/diffusers | 2,183,019,567 | I_kwDOHa8MBc6CHjwv | 7,293 | https://github.com/huggingface/diffusers/issues/7293 | https://api.github.com/repos/huggingface/diffusers/issues/7293 | Feature request: AudioLDM2 Pipeline integrate with audioldm_48k model | **Is your feature request related to a problem? Please describe.**
For now it seems that both the AudioLDM and AudioLDM2 pipeline can only generate audio content in 16k sample rate. It will be very useful and great if [this model](https://huggingface.co/haoheliu/audioldm_48k/tree/main) is ported to AudioLDM2 pipeline ... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-03-13T02:58:36Z | 2024-06-29T13:35:13Z | 2024-06-29T13:35:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lingchuL | 51,895,279 | MDQ6VXNlcjUxODk1Mjc5 | User | false |
huggingface/diffusers | 2,183,025,755 | I_kwDOHa8MBc6CHlRb | 7,294 | https://github.com/huggingface/diffusers/issues/7294 | https://api.github.com/repos/huggingface/diffusers/issues/7294 | Wrong Text-encoder Grad Setting in Custom_Diffusion | ### Describe the bug
According to [Custom Diffusion](https://arxiv.org/pdf/2212.04488.pdf), It only trains modified tokens. In implementation, it set the grad of non-modified tokens to zero. But there is a bug in this implementation.
https://github.com/huggingface/diffusers/blob/531e719163d2d7cf0d725bb685c1e8fe3393... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-03-13T03:05:55Z | 2024-03-13T14:52:45Z | 2024-03-13T14:52:45Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Rbrq03 | 115,004,993 | U_kgDOBtrWQQ | User | false |
huggingface/diffusers | 2,183,052,626 | I_kwDOHa8MBc6CHr1S | 7,296 | https://github.com/huggingface/diffusers/issues/7296 | https://api.github.com/repos/huggingface/diffusers/issues/7296 | Update PEFT required version for advaced_diffusion_training to 0.9.0 to fix DoRA dependency | ### Describe the bug
DoRA, which was implemented quite recently, breaks runtime by calling itself into PEFT class LoraConfig, which incorporates DoRA only from 0.9.0 version.
I will submit a PR to fix it and fix documentation.
### Reproduction
Every time.
### Logs
_No response_
### System Info
PEFT 0.7.0 vers... | closed | not_planned | false | 1 | [
"bug"
] | [] | 2024-03-13T03:40:15Z | 2024-03-13T09:47:28Z | 2024-03-13T09:47:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NorilskMajor | 112,411,678 | U_kgDOBrNEHg | User | false |
huggingface/diffusers | 2,183,064,870 | I_kwDOHa8MBc6CHu0m | 7,298 | https://github.com/huggingface/diffusers/issues/7298 | https://api.github.com/repos/huggingface/diffusers/issues/7298 | Enable users to utilize tokens abstraction in the converted tokens to prevent multiple LoRAs collisions. | **Is your feature request related to a problem? Please describe.**
By I guess mostly historical reasons the <s0><s1> token is being used for the provided entity, which is fine, unless I load multiple LoRAs that use the same token. How do we distinguish which one entity will get called?
**Describe the solution you'd... | open | null | false | 1 | [
"stale"
] | [] | 2024-03-13T03:56:06Z | 2024-04-12T15:03:02Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NorilskMajor | 112,411,678 | U_kgDOBrNEHg | User | false |
huggingface/diffusers | 1,397,538,363 | I_kwDOHa8MBc5TTL47 | 730 | https://github.com/huggingface/diffusers/issues/730 | https://api.github.com/repos/huggingface/diffusers/issues/730 | DreamBooth doesn't learn locally, but learns in Colab | ### Describe the bug
I'm having issues with DreamBooth, but only locally — the model doesn't seem to learn the instance. However, it does work in Colab ([this one](https://colab.research.google.com/github/ShivamShrirao/diffusers/blob/main/examples/dreambooth/DreamBooth_Stable_Diffusion.ipynb#scrollTo=jjcSXTp-u-Eg)).
... | closed | completed | false | 6 | [
"bug"
] | [
"patil-suraj"
] | 2022-10-05T09:55:18Z | 2022-10-10T11:13:58Z | 2022-10-06T17:44:58Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pink-red | 94,079,189 | U_kgDOBZuI1Q | User | false |
huggingface/diffusers | 2,183,114,937 | I_kwDOHa8MBc6CH7C5 | 7,300 | https://github.com/huggingface/diffusers/issues/7300 | https://api.github.com/repos/huggingface/diffusers/issues/7300 | High GPU memory usage after `set_default_attn_processor()` in unet | ### Describe the bug
I was using IPAdapter and saw `unload_ip_adapter()` in `IPAdapterMixin` which helps to rollback to default unet status.
But found that after using that function (`unload_ip_adapter()`) the inference speed gets slower and GPU's reserved memory peaks too high.
### Reproduction
Posting minim... | closed | completed | false | 14 | [
"bug"
] | [] | 2024-03-13T04:54:20Z | 2024-06-27T12:29:02Z | 2024-04-22T23:59:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | moono | 6,761,398 | MDQ6VXNlcjY3NjEzOTg= | User | false |
huggingface/diffusers | 2,183,524,856 | I_kwDOHa8MBc6CJfH4 | 7,303 | https://github.com/huggingface/diffusers/issues/7303 | https://api.github.com/repos/huggingface/diffusers/issues/7303 | Can use zero_snr when I use "from_pretrained" to create the scheduler | **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... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-03-13T09:34:10Z | 2024-06-30T05:35:45Z | 2024-06-30T05:35:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | CuddleSabe | 61,224,076 | MDQ6VXNlcjYxMjI0MDc2 | User | false |
huggingface/diffusers | 2,183,822,507 | I_kwDOHa8MBc6CKnyr | 7,305 | https://github.com/huggingface/diffusers/issues/7305 | https://api.github.com/repos/huggingface/diffusers/issues/7305 | Do we need __init__ in ModelMixin class? | https://github.com/huggingface/diffusers/blame/46ab56a468cbc43c1ab48190e27fc42195f4c60b/src/diffusers/models/modeling_utils.py#L199
I've observed that `ModelMixin` class currently includes an `__init__` method, which does not accept any arguments (args) or keyword arguments (kwargs). This design choice restricts the... | closed | completed | false | 9 | [
"stale"
] | [] | 2024-03-13T11:58:33Z | 2024-04-22T21:56:51Z | 2024-04-22T21:56:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | eldercrow | 9,844,839 | MDQ6VXNlcjk4NDQ4Mzk= | User | false |
huggingface/diffusers | 2,183,873,324 | I_kwDOHa8MBc6CK0Ms | 7,306 | https://github.com/huggingface/diffusers/issues/7306 | https://api.github.com/repos/huggingface/diffusers/issues/7306 | I can't use float16 in LEditsPPPipelineStableDiffusionXL | ### Describe the bug
I can't use float16 in LEditsPPPipelineStableDiffusionXL.
### Reproduction
This script works fine.
```py
from diffusers import LEditsPPPipelineStableDiffusionXL, DDIMScheduler
from diffusers.utils import load_image
model_id = "stabilityai/stable-diffusion-xl-base-1.0"
scheduler = DDIMSche... | closed | completed | false | 5 | [
"bug"
] | [
"linoytsaban"
] | 2024-03-13T12:22:28Z | 2024-03-14T09:53:59Z | 2024-03-13T22:04:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dai-ichiro | 53,703,225 | MDQ6VXNlcjUzNzAzMjI1 | User | false |
huggingface/diffusers | 2,184,720,719 | I_kwDOHa8MBc6CODFP | 7,308 | https://github.com/huggingface/diffusers/issues/7308 | https://api.github.com/repos/huggingface/diffusers/issues/7308 | [<languageCode>] Translating docs to <languageName> | null | closed | completed | false | 0 | [] | [] | 2024-03-13T19:13:34Z | 2024-03-13T19:14:20Z | 2024-03-13T19:14:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rugerioExe | 84,404,887 | MDQ6VXNlcjg0NDA0ODg3 | User | false |
huggingface/diffusers | 2,184,897,198 | I_kwDOHa8MBc6COuKu | 7,310 | https://github.com/huggingface/diffusers/issues/7310 | https://api.github.com/repos/huggingface/diffusers/issues/7310 | train_text_to_image_lora_sdxl.py does not seem to work using defaults | ### Describe the bug
I'm trying to run the train_text_to_image_lora_sdxl.py script and the output in tensorboard is black images. Now I assume I should be able to resolve this by reducing learning rate etc (reducing to 1e-6 has not resolved the issue)... but it I would expect the default values would be generally pr... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-03-13T21:12:31Z | 2024-10-06T15:39:48Z | 2024-03-14T14:09:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jloveric | 5,200,721 | MDQ6VXNlcjUyMDA3MjE= | User | false |
huggingface/diffusers | 2,185,275,294 | I_kwDOHa8MBc6CQKee | 7,311 | https://github.com/huggingface/diffusers/issues/7311 | https://api.github.com/repos/huggingface/diffusers/issues/7311 | train_text_to_image_sdxl.py Can't save model at checkpoint | ### Describe the bug
I am trying to finetune SDXL but the training script crashes when saving the model at a checkpoint.
Training runs fine.
### Reproduction
Here is my accelerate config choices:
- This machine
- No distributed training
- No
- No
- yes (to use deepspeed)
- no (don't specify a json)
- 2... | closed | completed | false | 22 | [
"bug",
"stale"
] | [] | 2024-03-14T02:53:29Z | 2024-11-14T18:37:43Z | 2024-11-14T18:37:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clement-swk | 91,945,598 | U_kgDOBXr6fg | User | false |
huggingface/diffusers | 2,185,384,193 | I_kwDOHa8MBc6CQlEB | 7,312 | https://github.com/huggingface/diffusers/issues/7312 | https://api.github.com/repos/huggingface/diffusers/issues/7312 | AlignProp Support for direct reward finetuning | **Is your feature request related to a problem? Please describe.**
No. AlignProp makes reward finetuning very fast compared to DDPO (about 25x) becoz of backpropagating the gradients directly from the reward function.
**Describe the solution you'd like.**
A similar integration to DDPO.
https://github.com/hugging... | closed | completed | false | 3 | [] | [] | 2024-03-14T04:16:54Z | 2024-03-16T01:08:46Z | 2024-03-16T01:08:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | parthos86 | 163,376,672 | U_kgDOCbzuIA | User | false |
huggingface/diffusers | 2,185,851,991 | I_kwDOHa8MBc6CSXRX | 7,317 | https://github.com/huggingface/diffusers/issues/7317 | https://api.github.com/repos/huggingface/diffusers/issues/7317 | Is peft really necessary as a mandatory hardwired dependency? | Long story short - does `peft` really need to be installed for every user of diffusers? Even if they never really use any `peft` code paths?
I'm sorry for this nitpicky issue.
**What API design would you like to have changed or added to the library? Why?**
Easy Diffusion has a fairly large and stable installed b... | closed | completed | false | 3 | [] | [] | 2024-03-14T09:18:25Z | 2024-03-15T13:38:37Z | 2024-03-15T12:59:03Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cmdr2 | 844,287 | MDQ6VXNlcjg0NDI4Nw== | User | false |
huggingface/diffusers | 1,397,732,951 | I_kwDOHa8MBc5TT7ZX | 732 | https://github.com/huggingface/diffusers/issues/732 | https://api.github.com/repos/huggingface/diffusers/issues/732 | [Feature Request] Dreambooth - Save intermediate checkpoints | **Is your feature request related to a problem? Please describe.**
Dreambooth can drastically change its output quality between step counts, including to the worse if the chosen learning rate is too high for the step count or amount of training / regularization images. This implementation only saves the model after tr... | closed | completed | false | 22 | [
"stale"
] | [
"patil-suraj"
] | 2022-10-05T12:42:56Z | 2023-02-04T02:57:30Z | 2022-12-13T14:30:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DominikDoom | 34,448,969 | MDQ6VXNlcjM0NDQ4OTY5 | User | false |
huggingface/diffusers | 2,186,267,738 | I_kwDOHa8MBc6CT8xa | 7,322 | https://github.com/huggingface/diffusers/issues/7322 | https://api.github.com/repos/huggingface/diffusers/issues/7322 | train_dreambooth_lora_sdxl.py & diffusers0.27.0dev error during training Signals.SIGKILL: 9 | ### Describe the bug
The latest train_dreambooth_lora_sdxl.py script with diffusers 0.27.0dev produces an error Signals.SIGKILL: 9.
The train_dreambooth_lora_sdxl.py script (version from 28.02: 7db935a) works with diffusers 0.26.3, but the issue in this case is that resuming from checkpoint does not work. I have se... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-03-14T12:39:30Z | 2024-03-14T14:07:19Z | 2024-03-14T14:07:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | SylwiaNowakowska | 54,419,373 | MDQ6VXNlcjU0NDE5Mzcz | User | false |
huggingface/diffusers | 2,187,093,375 | I_kwDOHa8MBc6CXGV_ | 7,330 | https://github.com/huggingface/diffusers/issues/7330 | https://api.github.com/repos/huggingface/diffusers/issues/7330 | ValueError: Attempting to unscale FP16 gradients. when running examples/text_to_image/train_text_to_image_lora.py | ### Describe the bug
Tried following the LoRA training example from huggingface tutorials using the editable install. Failed.
### Reproduction
Default accelerate config:
```
compute_environment: LOCAL_MACHINE
debug: false
distributed_type: 'NO'
downcast_bf16: 'no'
gpu_ids: all
machine_rank: 0
main_trai... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2024-03-14T19:11:07Z | 2024-07-12T18:44:40Z | 2024-06-30T05:34:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | HemalPatil | 14,046,523 | MDQ6VXNlcjE0MDQ2NTIz | User | false |
huggingface/diffusers | 2,187,100,693 | I_kwDOHa8MBc6CXIIV | 7,331 | https://github.com/huggingface/diffusers/issues/7331 | https://api.github.com/repos/huggingface/diffusers/issues/7331 | Deteriorated Quality for DPM++ 2M Generations | ### Describe the bug
In Invoke's 3.6.3 version, we updated to the latest Diffusers available at the time (.26) for the benefit of the scheduler fixes.
A user has provided documentation/testing demonstrating that these updates seems to have resulted in some worsened quality for certain schedulers (tested evidence is... | closed | completed | false | 7 | [
"stale",
"scheduler"
] | [
"pcuenca",
"yiyixuxu"
] | 2024-03-14T19:15:51Z | 2024-11-04T19:33:42Z | 2024-11-04T19:33:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hipsterusername | 31,807,370 | MDQ6VXNlcjMxODA3Mzcw | User | false |
huggingface/diffusers | 2,187,464,061 | I_kwDOHa8MBc6CYg19 | 7,333 | https://github.com/huggingface/diffusers/issues/7333 | https://api.github.com/repos/huggingface/diffusers/issues/7333 | Tuning-Free Noise Rectification for High Fidelity Image-to-Video Generation | ### Model/Pipeline/Scheduler description
Applying pretrained Text-to-Video (T2V) Diffusion models to Image-to-video (I2V) generation tasks using SDEdit often results in low source image fidelity in open domains. This method achieves high source image fidelity in output videos through supplementing more precise sourc... | open | null | false | 15 | [
"community-examples",
"stale",
"contributions-welcome"
] | [] | 2024-03-14T23:51:19Z | 2024-12-21T15:06:48Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clarencechen | 8,482,341 | MDQ6VXNlcjg0ODIzNDE= | User | false |
huggingface/diffusers | 2,187,697,739 | I_kwDOHa8MBc6CZZ5L | 7,334 | https://github.com/huggingface/diffusers/issues/7334 | https://api.github.com/repos/huggingface/diffusers/issues/7334 | VAE config mismatch when using StableDiffusionInpainting | Currently, we have got a failing test:
```
tests/pipelines/stable_diffusion/test_stable_diffusion_inpaint.py::StableDiffusionInpaintPipelineSlowTests::test_single_file_component_configs
```
And it fails for the right reason! You will notice that the [assertion](https://github.com/huggingface/diffusers/blob/5d84... | closed | completed | false | 7 | [] | [] | 2024-03-15T04:27:20Z | 2024-12-05T09:41:54Z | 2024-12-05T09:41:54Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,187,773,389 | I_kwDOHa8MBc6CZsXN | 7,337 | https://github.com/huggingface/diffusers/issues/7337 | https://api.github.com/repos/huggingface/diffusers/issues/7337 | How to convert multiple piped files into a single SafeTensor file? | How to convert multiple piped files into a single SafeTensor file?
For example, from this address: https://huggingface.co/Vargol/sdxl-lightning-4-steps/tree/main
```python
import torch
from diffusers import StableDiffusionXLPipeline, EulerDiscreteScheduler
base = "Vargol/sdxl-lightning-4-steps"... | closed | completed | false | 2 | [] | [] | 2024-03-15T05:49:01Z | 2024-03-15T06:51:24Z | 2024-03-15T06:51:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xxddccaa | 25,652,821 | MDQ6VXNlcjI1NjUyODIx | User | false |
huggingface/diffusers | 2,188,487,415 | I_kwDOHa8MBc6Ccar3 | 7,341 | https://github.com/huggingface/diffusers/issues/7341 | https://api.github.com/repos/huggingface/diffusers/issues/7341 | TypeError: '<' not supported between instances of 'TorchVersion' and 'Version' | ### Describe the bug
My Google Colab code for training a stable diffusion works for many months until now, this is the error trace:
- `15 from diffusers import DDPMScheduler`
- `3 from .configuration_utils import ConfigMixin`
- `34 from .utils import DIFFUSERS_CACHE, HUGGINGFACE_CO_RESOLVE_ENDPOINT, DummyObject, ... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-03-15T13:12:18Z | 2024-03-16T02:11:27Z | 2024-03-16T02:11:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jamesnogra | 6,120,605 | MDQ6VXNlcjYxMjA2MDU= | User | false |
huggingface/diffusers | 2,188,597,001 | I_kwDOHa8MBc6Cc1cJ | 7,342 | https://github.com/huggingface/diffusers/issues/7342 | https://api.github.com/repos/huggingface/diffusers/issues/7342 | Any Specific Reason for redundant computations in `compute_snr()` ？ | Currently in `training_utils.py`, the `compute_snr()` is given as
```python
def compute_snr(noise_scheduler, timesteps):
"""
Computes SNR as per
https://github.com/TiankaiHang/Min-SNR-Diffusion-Training/blob/521b624bd70c67cee4bdf49225915f5945a872e3/guided_diffusion/gaussian_diffusion.py#L847-L849
... | closed | completed | false | 8 | [] | [] | 2024-03-15T14:03:36Z | 2024-03-18T10:51:11Z | 2024-03-18T10:51:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | e-yi | 20,715,359 | MDQ6VXNlcjIwNzE1MzU5 | User | false |
huggingface/diffusers | 2,189,309,595 | I_kwDOHa8MBc6Cfjab | 7,347 | https://github.com/huggingface/diffusers/issues/7347 | https://api.github.com/repos/huggingface/diffusers/issues/7347 | Training Lora Fails after latest update | ### Describe the bug
After updating to latest the script to train a lora fails, I'll attach the error below.
### Reproduction
accelerate launch train_dreambooth_lora_sdxl_advanced.py \
--pretrained_model_name_or_path=$MODEL_NAME \
--pretrained_vae_model_name_or_path=$VAE_PATH \
--dataset_name='boby-set' \
... | closed | completed | false | 6 | [
"bug"
] | [] | 2024-03-15T19:13:40Z | 2024-05-17T15:18:31Z | 2024-05-17T15:18:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pedropaf | 1,830,498 | MDQ6VXNlcjE4MzA0OTg= | User | false |
huggingface/diffusers | 2,189,787,882 | I_kwDOHa8MBc6ChYLq | 7,348 | https://github.com/huggingface/diffusers/issues/7348 | https://api.github.com/repos/huggingface/diffusers/issues/7348 | EulerAncestralDiscreteScheduler is generating weird results when used with StableDiffusionControlNetInpaintPipeline | ### Describe the bug
I was trying the demo from https://huggingface.co/docs/diffusers/en/using-diffusers/controlnet#inpainting
Only change I made was that I have used a different scheduler, i.e. instead of UniPCMultistepScheduler I have used EulerAncestralDiscreteScheduler
And the results are horrible, here is... | closed | completed | false | 11 | [
"bug"
] | [
"yiyixuxu"
] | 2024-03-16T05:24:39Z | 2024-03-22T03:55:40Z | 2024-03-19T10:51:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nayan-dhabarde | 24,761,808 | MDQ6VXNlcjI0NzYxODA4 | User | false |
huggingface/diffusers | 2,189,866,276 | I_kwDOHa8MBc6ChrUk | 7,350 | https://github.com/huggingface/diffusers/issues/7350 | https://api.github.com/repos/huggingface/diffusers/issues/7350 | Cannot initialize Controlnet from a Unet with lora | ### Describe the bug
I encounter errors when initialize controlnet's parameters from unet, since incompatible parameter keys, for example:
Missing key(s): 0.attentions.0.proj_in.weight ...
Unexpected key(s) in state_dict: "0.attentions.0.proj_in.base_layer.weight", "0.attentions.0.proj_in.base_layer.bias", "0.at... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-03-16T08:46:15Z | 2024-03-17T15:39:48Z | 2024-03-17T15:39:48Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Luciennnnnnn | 20,135,317 | MDQ6VXNlcjIwMTM1MzE3 | User | false |
huggingface/diffusers | 2,190,162,907 | I_kwDOHa8MBc6Cizvb | 7,354 | https://github.com/huggingface/diffusers/issues/7354 | https://api.github.com/repos/huggingface/diffusers/issues/7354 | PNDMScheduler Does Not Compatible with DDPMPipeline | ### Describe the bug
I was trying to test different schedulers under DDPMPipeline. And an error occurred if I use PNDMScheduler beforehand I have found that PNDMScheduler should be compatible with DDPMPipeline following the official tutorial.
```Python
pipeline = DDPMPipeline(unet=model, scheduler=noise_scheduler)
... | open | null | false | 9 | [
"bug",
"stale"
] | [] | 2024-03-16T18:02:51Z | 2024-12-14T15:07:17Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KeepNoob | 101,505,765 | U_kgDOBgza5Q | User | false |
huggingface/diffusers | 2,190,182,825 | I_kwDOHa8MBc6Ci4mp | 7,355 | https://github.com/huggingface/diffusers/issues/7355 | https://api.github.com/repos/huggingface/diffusers/issues/7355 | Stable Cascade, error when generating images with arbitrary resolutions and bfloat16. | ### Describe the bug
When I create an image of resolution 904x512 with a model using bfloat16 quantization I get an error stating that float32 input and bfloat16 bias are incompatible.
See the associated pull request for the line that seems to be problematic.
### Reproduction
```
import torch
from diffu... | closed | completed | false | 0 | [
"bug"
] | [
"DN6"
] | 2024-03-16T18:45:55Z | 2024-03-19T10:47:45Z | 2024-03-19T10:47:45Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lawfordp2017 | 24,662,465 | MDQ6VXNlcjI0NjYyNDY1 | User | false |
huggingface/diffusers | 2,190,221,780 | I_kwDOHa8MBc6CjCHU | 7,357 | https://github.com/huggingface/diffusers/issues/7357 | https://api.github.com/repos/huggingface/diffusers/issues/7357 | Slow inference(?) despite `torch.compile` and `fuse_loras` for SDXL | ### Describe the bug
I am attempting to fine-tune an SDXL model by fusing a few LoRA adapters, then utilizing `torch.compile` per this portion of the [Merge LoRAs documentation](https://huggingface.co/docs/diffusers/main/en/using-diffusers/merge_loras#torchcompile) to generate images of my desired style while improvin... | closed | completed | false | 3 | [] | [] | 2024-03-16T19:53:10Z | 2024-03-19T17:51:03Z | 2024-03-17T15:22:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | RizzwareEngineer | 67,522,964 | MDQ6VXNlcjY3NTIyOTY0 | User | false |
huggingface/diffusers | 1,397,957,965 | I_kwDOHa8MBc5TUyVN | 736 | https://github.com/huggingface/diffusers/issues/736 | https://api.github.com/repos/huggingface/diffusers/issues/736 | HFValidationError( '.ggingface_hub.utils._validators.HFValidationError: Repo id must use alphanumeric chars or '-', '_', '.', '--' and '..' are forbidden, '-' and '.' cannot start or end the name, max length is 96: 'CompVis/stable-diffusion-v1-4 | Tried to install Dreambooth according to Nerdy Rodent video (https://www.youtube.com/watch?v=w6PTviOCYQY)
I am absolutely new to Linux, and failed on last step running the training.
As I understood, something wrong with:
export MODEL_NAME="CompVis/stable-diffusion-v1-4" and --pretrained_model_name_or_path=$MO... | closed | completed | false | 17 | [
"stale"
] | [
"patil-suraj"
] | 2022-10-05T15:07:32Z | 2025-08-08T18:47:06Z | 2022-12-08T15:03:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Renaldas111 | 97,415,304 | U_kgDOBc5wiA | User | false |
huggingface/diffusers | 2,190,936,313 | I_kwDOHa8MBc6Clwj5 | 7,363 | https://github.com/huggingface/diffusers/issues/7363 | https://api.github.com/repos/huggingface/diffusers/issues/7363 | Safety Checker for stable_diffusion_xl | **Is your feature request related to a problem? Please describe.**
Yes, cannot use safety checker with stable diffusion xl directly inside pipeline
**Describe the solution you'd like.**
Ability to use safety checker directly inside pipeline
**Describe alternatives you've considered.**
Just running the image th... | closed | completed | false | 1 | [] | [] | 2024-03-17T23:37:56Z | 2024-03-18T10:00:24Z | 2024-03-18T10:00:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mantrakp04 | 87,142,457 | MDQ6VXNlcjg3MTQyNDU3 | User | false |
huggingface/diffusers | 2,191,129,124 | I_kwDOHa8MBc6Cmfok | 7,365 | https://github.com/huggingface/diffusers/issues/7365 | https://api.github.com/repos/huggingface/diffusers/issues/7365 | Provided pooled_prompt_embeds is overwritten via prompt_embeds[0] | https://github.com/huggingface/diffusers/blob/25caf24ef90fc44074f4fd3712f6ed5a1db4a5c3/src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py#L386
Simple fix:
```python
pooled_prompt_embeds = prompt_embeds[0] if pooled_prompt_embeds is not None else pooled_prompt_embeds
```
Sorry this is... | closed | completed | false | 13 | [
"stale"
] | [] | 2024-03-18T03:03:22Z | 2025-01-08T20:12:13Z | 2025-01-08T20:12:13Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cloneofsimo | 35,953,539 | MDQ6VXNlcjM1OTUzNTM5 | User | false |
huggingface/diffusers | 2,191,167,793 | I_kwDOHa8MBc6CmpEx | 7,366 | https://github.com/huggingface/diffusers/issues/7366 | https://api.github.com/repos/huggingface/diffusers/issues/7366 | error using load_lora_weights with dora | I got error from my trained xl dora file
pipe = StableDiffusionXLPipeline.from_pretrained(
"stabilityai/stable-diffusion-xl-base-1.0",
torch_dtype=torch.bfloat16
)
pipe.to("cuda")
pipe.load_lora_weights("e:/diffusion-models/Lora/test16/woman_young_64rank.safetensors")
pipe.fuse_lora(lora_scale=0.8... | closed | completed | false | 21 | [] | [] | 2024-03-14T04:47:06Z | 2024-03-26T04:07:34Z | 2024-03-26T04:07:34Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nullarch | 8,098,724 | MDQ6VXNlcjgwOTg3MjQ= | User | false |
huggingface/diffusers | 2,191,242,876 | I_kwDOHa8MBc6Cm7Z8 | 7,367 | https://github.com/huggingface/diffusers/issues/7367 | https://api.github.com/repos/huggingface/diffusers/issues/7367 | Open source upscaler: Clarity | **Is your feature request related to a problem? Please describe.**
Diffusers are missing a good image upscaler.
**Describe the solution you'd like.**
Implement the Clarity-upscaler:
Code: https://github.com/philz1337x/clarity-upscaler
Demo: https://replicate.com/philipp1337x/clarity-upscaler
https://github.co... | closed | not_planned | false | 7 | [
"stale",
"contributions-welcome"
] | [] | 2024-03-18T05:05:58Z | 2024-11-30T02:43:40Z | 2024-11-30T02:43:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 2,191,347,867 | I_kwDOHa8MBc6CnVCb | 7,368 | https://github.com/huggingface/diffusers/issues/7368 | https://api.github.com/repos/huggingface/diffusers/issues/7368 | Train GLIGEN in diffusers? | Thanks for your great job!!!
Now, I know how to infer GLIGEN with diffusers[https://github.com/gligen/diffusers/tree/gligen/examples/gligen](url). But how can I train GLIGEN with diffusers like ControlNet[https://github.com/huggingface/diffusers/blob/main/examples/controlnet/train_controlnet.py](url)?
Thanks agai... | closed | completed | false | 12 | [
"stale"
] | [] | 2024-03-18T06:34:27Z | 2024-11-20T00:23:43Z | 2024-11-20T00:23:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,191,646,965 | I_kwDOHa8MBc6CoeD1 | 7,370 | https://github.com/huggingface/diffusers/issues/7370 | https://api.github.com/repos/huggingface/diffusers/issues/7370 | Support for hpcaitech OpenSora's STDiT for text2video and text2image generation | ### Model/Pipeline/Scheduler description
STDiT builds on Latte and DiT and yields a trade-off between generation quality and speed
https://github-production-user-asset-6210df.s3.amazonaws.com/99191637/313485495-983a1965-a374-41a7-a76b-c07941a6c1e9.mp4?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=AKIAVCODYLSA... | open | null | false | 8 | [
"community-examples",
"Good second issue",
"contributions-welcome"
] | [] | 2024-03-18T09:18:47Z | 2024-11-20T00:22:55Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,192,063,172 | I_kwDOHa8MBc6CqDrE | 7,375 | https://github.com/huggingface/diffusers/issues/7375 | https://api.github.com/repos/huggingface/diffusers/issues/7375 | [Docs] Add outputs after FreeInit block | Would be nice to add an output after https://huggingface.co/docs/diffusers/main/en/api/pipelines/animatediff#using-freeinit.
| closed | completed | false | 0 | [] | [
"DN6"
] | 2024-03-18T12:24:23Z | 2024-03-19T16:14:22Z | 2024-03-19T16:14:22Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,192,422,801 | I_kwDOHa8MBc6CrbeR | 7,376 | https://github.com/huggingface/diffusers/issues/7376 | https://api.github.com/repos/huggingface/diffusers/issues/7376 | Diffusion Transformers quantization | **Is your feature request related to a problem? Please describe.**
Due to OpenAI's DALLE-3 and Sora's soaring success, many projects, such as StableDiffusion3, PixArt and Open-Sora, are trying to replicate its architecture, whose backbone is a Diffusion Transformer (DiT). Because it is a transformer, the same quanti... | closed | completed | false | 14 | [] | [] | 2024-03-18T14:45:52Z | 2024-11-19T09:32:59Z | 2024-11-19T09:32:54Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,192,459,014 | I_kwDOHa8MBc6CrkUG | 7,377 | https://github.com/huggingface/diffusers/issues/7377 | https://api.github.com/repos/huggingface/diffusers/issues/7377 | StableCascadeDecoderPipeline Error with num_images_per_prompt > 1 | ### Describe the bug
I'm encountering a RuntimeError when using the `StableCascadeDecoderPipeline` with `num_images_per_prompt` set to a value greater than 1. Since parallel processing functions correctly in the `StableCascadePriorPipeline`, I suspect an issue with how latents are being passed to the decoder pipeline.... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-03-18T14:59:08Z | 2024-03-19T17:35:56Z | 2024-03-19T17:35:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | IndigoDosSantos | 130,935,112 | U_kgDOB83pSA | User | false |
huggingface/diffusers | 2,193,812,158 | I_kwDOHa8MBc6Cwuq- | 7,378 | https://github.com/huggingface/diffusers/issues/7378 | https://api.github.com/repos/huggingface/diffusers/issues/7378 | Decoding latents independently using AnimateDiffVideoToVideoPipeline takes more memory than outputting images directly | ### Describe the bug
When I use the AnimateDiffVideoToVideo pipeline to output images, my server doesn't run out of memory. However, if I output the latents, and then manually run pipe.decode_latents, I somehow run out of memory.
### Reproduction
Code with manual latent decoding:
````
import imageio
from ... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-03-19T01:26:04Z | 2024-03-31T22:52:06Z | 2024-03-31T22:52:06Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AbhinavGopal | 43,016,805 | MDQ6VXNlcjQzMDE2ODA1 | User | false |
huggingface/diffusers | 2,193,982,446 | I_kwDOHa8MBc6CxYPu | 7,380 | https://github.com/huggingface/diffusers/issues/7380 | https://api.github.com/repos/huggingface/diffusers/issues/7380 | implementation of One-step Diffusion with Distribution Matching Distillation | ### Model/Pipeline/Scheduler description
https://github.com/Zeqiang-Lai/OpenDMD
<img width="780" alt="image" src="https://github.com/huggingface/diffusers/assets/26198430/bf5dc6df-447f-48f1-bf1a-c16040e02304">
### Open source status
- [X] The model implementation is available.
- [X] The model weights are availa... | open | null | false | 20 | [
"stale",
"training",
"contributions-welcome"
] | [] | 2024-03-19T04:23:11Z | 2026-02-05T08:08:56Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Zeqiang-Lai | 26,198,430 | MDQ6VXNlcjI2MTk4NDMw | User | false |
huggingface/diffusers | 2,194,044,172 | I_kwDOHa8MBc6CxnUM | 7,383 | https://github.com/huggingface/diffusers/issues/7383 | https://api.github.com/repos/huggingface/diffusers/issues/7383 | train controlnet sdxl with playground v2.5 | I try to train controlnet sdxl with playground v2.5. I saw the loss is smaller as training is on. But the validation image is ok at first few steps, but as the training is going on, the validation result images become full of noise like below. How to solve it?
I use the train_controlnet_sdxl.py to train and modify... | closed | completed | false | 2 | [] | [] | 2024-03-19T05:15:32Z | 2024-03-19T09:55:00Z | 2024-03-19T08:31:07Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | universewill | 6,790,730 | MDQ6VXNlcjY3OTA3MzA= | User | false |
huggingface/diffusers | 2,194,293,838 | I_kwDOHa8MBc6CykRO | 7,387 | https://github.com/huggingface/diffusers/issues/7387 | https://api.github.com/repos/huggingface/diffusers/issues/7387 | Is the code for Frame Interpolation in SVD open source?I can't find it. | null | closed | completed | false | 4 | [] | [] | 2024-03-19T08:06:31Z | 2024-03-19T08:53:04Z | 2024-03-19T08:29:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Dumeowmeow | 99,797,884 | U_kgDOBfLLfA | User | false |
huggingface/diffusers | 2,194,341,640 | I_kwDOHa8MBc6Cyv8I | 7,389 | https://github.com/huggingface/diffusers/issues/7389 | https://api.github.com/repos/huggingface/diffusers/issues/7389 | StableDiffusionXLPipeline.from_single_file Error | ### Describe the bug
version: 0.27.0, 0.27.1
when use StableDiffusionXLPipeline.from_single_file,error occur:
```
File "aaa/anaconda/envs/sd-webui/lib/python3.10/site-packages/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py", line 383, in encode_prompt
prompt_embeds = text_encoder... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2024-03-19T08:33:36Z | 2024-05-09T06:59:09Z | 2024-05-03T21:38:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cendelian | 50,360,389 | MDQ6VXNlcjUwMzYwMzg5 | User | false |
huggingface/diffusers | 2,194,378,830 | I_kwDOHa8MBc6Cy5BO | 7,390 | https://github.com/huggingface/diffusers/issues/7390 | https://api.github.com/repos/huggingface/diffusers/issues/7390 | StableDiffusionXLControlNetInpaintPipeline.from_single_file Error too!!! | ### Describe the bug
diffuers version: 0.27.1
when run StableDiffusionXLControlNetInpaintPipeline.from_single_file, error occur:
```
self.pipe = StableDiffusionXLControlNetInpaintPipeline.from_single_file(model_address['sdxl3D'],
File "aaaa/anaconda/envs/sd-webui/lib/python3.10/site-packages/hugging... | closed | completed | false | 16 | [
"bug"
] | [] | 2024-03-19T08:53:07Z | 2024-11-19T09:31:37Z | 2024-11-19T09:31:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cendelian | 50,360,389 | MDQ6VXNlcjUwMzYwMzg5 | User | false |
huggingface/diffusers | 2,194,655,923 | I_kwDOHa8MBc6Cz8qz | 7,392 | https://github.com/huggingface/diffusers/issues/7392 | https://api.github.com/repos/huggingface/diffusers/issues/7392 | How can I train SDXL with multiple aspect ratio images???? | **Is your feature request related to a problem? Please describe.**
When training SDXL, multiple aspect ratio images are needed, e.g 1024x1024, 832x1152... how can I collect the images with same size in the same batch?
**Describe the solution you'd like.**
Plz update the sdxl training script, maybe modifying colla... | closed | completed | false | 4 | [] | [] | 2024-03-19T11:01:53Z | 2024-03-20T02:31:54Z | 2024-03-20T02:31:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lawsonxwl | 52,605,055 | MDQ6VXNlcjUyNjA1MDU1 | User | false |
huggingface/diffusers | 2,196,551,688 | I_kwDOHa8MBc6C7LgI | 7,400 | https://github.com/huggingface/diffusers/issues/7400 | https://api.github.com/repos/huggingface/diffusers/issues/7400 | A problem regarding stable video diffusion, with inference results yielding black images | ### Describe the bug
I encountered an issue with inference in stable video diffusion. When I attempted to perform inference using the sample code, it indicated insufficient VRAM (video random access memory). Subsequently, I reduced the width and height of the output size, which resulted in the inference process runnin... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-03-20T02:24:42Z | 2024-04-02T03:39:10Z | 2024-04-02T03:38:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DoloresChong | 76,471,389 | MDQ6VXNlcjc2NDcxMzg5 | User | false |
huggingface/diffusers | 2,197,254,675 | I_kwDOHa8MBc6C93IT | 7,405 | https://github.com/huggingface/diffusers/issues/7405 | https://api.github.com/repos/huggingface/diffusers/issues/7405 | [new model] DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors | ### Model/Pipeline/Scheduler description
A very great model for image2video generation (also frame interpolation, storyboarding and looping videos). The architecture is based on the current text2video Unet paradigm, so it shouldn't be hard to adjust it.
The most important feature of the diffusers implementation wou... | closed | completed | false | 4 | [
"stale",
"contributions-welcome"
] | [] | 2024-03-20T11:03:02Z | 2024-06-29T13:33:01Z | 2024-06-29T13:33:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 2,197,261,323 | I_kwDOHa8MBc6C94wL | 7,406 | https://github.com/huggingface/diffusers/issues/7406 | https://api.github.com/repos/huggingface/diffusers/issues/7406 | A Possible Bug in EDMEulerScheduler | ### Describe the bug
I was trying to train a model in DDPM pipeline by using EDMEulerScheduler . I noticed that in EDMEulerScheduler the noise_scheduler.timesteps is not integer but is float. However, during training we had to sample the timesteps as an argument to passed to the Unet model, and here is the problem, wh... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-03-20T11:06:39Z | 2024-06-29T13:33:34Z | 2024-06-29T13:33:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KeepNoob | 101,505,765 | U_kgDOBgza5Q | User | false |
huggingface/diffusers | 2,197,313,268 | I_kwDOHa8MBc6C-Fb0 | 7,408 | https://github.com/huggingface/diffusers/issues/7408 | https://api.github.com/repos/huggingface/diffusers/issues/7408 | [Pipeline] FRESCO | ### Model/Pipeline/Scheduler description
https://github.com/huggingface/diffusers/assets/72266394/a0f2a4d9-ed6e-403e-8a44-f53a9831467d
### Open source status
- [X] The model implementation is available.
- [X] The model weights are available (Only relevant if addition is not a scheduler).
### Provide us... | closed | completed | false | 6 | [
"community-examples",
"stale",
"contributions-welcome"
] | [] | 2024-03-20T11:34:40Z | 2024-05-27T14:18:47Z | 2024-05-27T14:18:47Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | a-r-r-o-w | 72,266,394 | MDQ6VXNlcjcyMjY2Mzk0 | User | false |
huggingface/diffusers | 2,199,049,758 | I_kwDOHa8MBc6DEtYe | 7,415 | https://github.com/huggingface/diffusers/issues/7415 | https://api.github.com/repos/huggingface/diffusers/issues/7415 | AutoencoderKL will mess up with bfloat16 when running on TPU v4 devices | ### Describe the bug
AutoencoderKL will mess up with bfloat16 when running on TPU v4 devices.
Another issue I found is that, settings `mixed_precision: 'bf16'` in the yaml file has no effect at all, so that's why the `mixed_precision='bf16'` is still needed in the code when initializing Accelerator
### Reproductio... | closed | completed | false | 13 | [
"bug",
"stale"
] | [] | 2024-03-21T03:39:27Z | 2024-04-26T09:03:50Z | 2024-04-26T09:03:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | PikaCat-OuO | 73,384,062 | MDQ6VXNlcjczMzg0MDYy | User | false |
huggingface/diffusers | 2,199,694,240 | I_kwDOHa8MBc6DHKug | 7,417 | https://github.com/huggingface/diffusers/issues/7417 | https://api.github.com/repos/huggingface/diffusers/issues/7417 | `test_lora_fuse_nan()` fails on Mac | ### Describe the bug
The test `test_lora_fuse_nan()` (which is mixed into `StableDiffusionLoRATests` and `StableDiffusionXLLoRATests`) fails on Mac.
The test sets a lora weight to `inf` and expects the pipe to output `nans`. On Linux this works. On Mac, this fails, because the tensor is zeros, not nan.
___
Linu... | closed | completed | false | 8 | [
"bug"
] | [] | 2024-03-21T09:54:48Z | 2024-03-27T09:06:01Z | 2024-03-27T09:06:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | UmerHA | 40,663,591 | MDQ6VXNlcjQwNjYzNTkx | User | false |
huggingface/diffusers | 2,199,832,534 | I_kwDOHa8MBc6DHsfW | 7,418 | https://github.com/huggingface/diffusers/issues/7418 | https://api.github.com/repos/huggingface/diffusers/issues/7418 | Bug in https://github.com/huggingface/diffusers/blob/main/examples/community/pipeline_stable_diffusion_upscale_ldm3d.py#L29 | ### Describe the bug
In the following import is not correct anymore:
https://github.com/huggingface/diffusers/blob/main/examples/community/pipeline_stable_diffusion_upscale_ldm3d.py#L29
`from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion_ldm3d import LDM3DPipelineOutput
`
Should be
`from diffuse... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-03-21T10:49:22Z | 2024-04-24T08:34:51Z | 2024-04-24T08:34:41Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | estelleafl | 81,072,900 | MDQ6VXNlcjgxMDcyOTAw | User | false |
huggingface/diffusers | 2,199,896,740 | I_kwDOHa8MBc6DH8Kk | 7,420 | https://github.com/huggingface/diffusers/issues/7420 | https://api.github.com/repos/huggingface/diffusers/issues/7420 | unable to use torch.compile | ### Describe the bug
use SD XL will return an error
```
pipe.text_encoder_2 = torch.compile(pipe.text_encoder_2, mode="max-autotune", fullgraph=True)
```
### Reproduction
```
import torch
from diffusers import DiffusionPipeline
pipe = DiffusionPipeline.from_pretrained(
"cagliostrolab/animagine-xl-3.1", ... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-03-21T11:16:47Z | 2024-03-25T10:59:00Z | 2024-03-25T10:59:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yongjer | 54,315,206 | MDQ6VXNlcjU0MzE1MjA2 | User | false |
huggingface/diffusers | 2,199,971,191 | I_kwDOHa8MBc6DIOV3 | 7,421 | https://github.com/huggingface/diffusers/issues/7421 | https://api.github.com/repos/huggingface/diffusers/issues/7421 | Seed not working with StableDiffusionXLKDiffusionPipeline | ### Describe the bug
When I get an inference with the same seed, ı need to get the same result. It is important for reproducibility. But when I used StableDiffusionXLKDiffusionPipeline, each inference produced different results.
### Reproduction
StableDiffusionXLKDiffusionPipeline Example
```python
import to... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-03-21T11:50:11Z | 2024-03-21T13:15:27Z | 2024-03-21T12:47:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | m-pektas | 15,729,405 | MDQ6VXNlcjE1NzI5NDA1 | User | false |
huggingface/diffusers | 2,200,308,501 | I_kwDOHa8MBc6DJgsV | 7,422 | https://github.com/huggingface/diffusers/issues/7422 | https://api.github.com/repos/huggingface/diffusers/issues/7422 | inference DoRA of Advanced_diffusion_training issues | ### What does this PR do?
this is the command line for training LORA + DORA
`accelerate launch --num_cpu_threads_per_process=2 C:/Users/qkrwn/ML/diffusers/examples/advanced_diffusion_training/train_dreambooth_lora_sdxl_advanced.py --pretrained_model_name_or_path="streamize/xxmixgirl" --instance_prompt="a photo o... | closed | completed | false | 10 | [] | [] | 2024-03-21T14:12:49Z | 2024-03-26T04:07:34Z | 2024-03-26T04:07:34Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nullarch | 8,098,724 | MDQ6VXNlcjgwOTg3MjQ= | User | false |
huggingface/diffusers | 2,200,618,581 | I_kwDOHa8MBc6DKsZV | 7,426 | https://github.com/huggingface/diffusers/issues/7426 | https://api.github.com/repos/huggingface/diffusers/issues/7426 | [MPS] SDXL pipeline fails inference in fp16 mode | ### Describe the bug
When running a pipeline under 🤗 Accelerate, my M3 system crashes with a tensor dtype mismatch during validation inference.
This issue does not occur without 🤗 Accelerate.
### Reproduction
Unfortunately, the way the reproducer is executed within a training script makes it exceedingly d... | closed | completed | false | 13 | [
"bug"
] | [] | 2024-03-21T16:19:51Z | 2024-03-26T14:35:50Z | 2024-03-26T14:35:50Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,201,549,468 | I_kwDOHa8MBc6DOPqc | 7,428 | https://github.com/huggingface/diffusers/issues/7428 | https://api.github.com/repos/huggingface/diffusers/issues/7428 | Will support layer diffusion in diffusers? | https://github.com/huchenlei/ComfyUI-layerdiffuse | closed | completed | false | 1 | [] | [] | 2024-03-22T02:26:52Z | 2024-03-22T09:51:04Z | 2024-03-22T09:51:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kelisiya | 23,169,766 | MDQ6VXNlcjIzMTY5NzY2 | User | false |
huggingface/diffusers | 2,201,626,711 | I_kwDOHa8MBc6DOihX | 7,429 | https://github.com/huggingface/diffusers/issues/7429 | https://api.github.com/repos/huggingface/diffusers/issues/7429 | How to use k_diffusion with Controlnet (SDXL)? | Dear developer,
I try to modify the code of [k_diffusion](https://github.com/huggingface/diffusers/blob/9613576191d8613fc550a1ec286adc4f1fc208ec/src/diffusers/pipelines/stable_diffusion_k_diffusion/pipeline_stable_diffusion_xl_k_diffusion.py#L837) to be compatible with controlnet.
But I got incorrect results, t... | closed | completed | false | 1 | [] | [] | 2024-03-22T03:33:38Z | 2024-04-18T03:25:55Z | 2024-04-18T03:25:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YoucanBaby | 43,912,540 | MDQ6VXNlcjQzOTEyNTQw | User | false |
huggingface/diffusers | 1,398,358,975 | I_kwDOHa8MBc5TWUO_ | 743 | https://github.com/huggingface/diffusers/issues/743 | https://api.github.com/repos/huggingface/diffusers/issues/743 | Can't convert ckpt model to diffusers with convert_original_stable_diffusion_to_diffusers | ### Describe the bug
I have some ckpt files and try to convert it to diffusers in google colab.
with this command
`!python /content/diffusers/scripts/convert_original_stable_diffusion_to_diffusers.py --checkpoint_path /content/gdrive/MyDrive/Waifu_6e.ckpt --dump_path /content/stable-diffusion-alt`
But got ... | closed | completed | false | 5 | [
"bug"
] | [
"patrickvonplaten"
] | 2022-10-05T20:44:45Z | 2022-10-27T08:52:21Z | 2022-10-27T08:52:21Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Desm0nt | 11,135,734 | MDQ6VXNlcjExMTM1NzM0 | User | false |
huggingface/diffusers | 2,201,772,374 | I_kwDOHa8MBc6DPGFW | 7,431 | https://github.com/huggingface/diffusers/issues/7431 | https://api.github.com/repos/huggingface/diffusers/issues/7431 | Issues of training dreambooth using DoRA with text encoder | ### Describe the bug
I executed the following command to train a dreambooth model using the example script from the document. The only change I made is to change the `--train_text_encoder_ti` to `--train_text_encoder`. However, the loss is **always NaN**.
References:
+ https://github.com/huggingface/diffusers/b... | closed | completed | false | 11 | [
"bug",
"stale"
] | [] | 2024-03-22T06:10:26Z | 2024-10-21T15:14:20Z | 2024-10-21T15:14:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rfan-debug | 69,488,297 | MDQ6VXNlcjY5NDg4Mjk3 | User | false |
huggingface/diffusers | 2,201,845,834 | I_kwDOHa8MBc6DPYBK | 7,432 | https://github.com/huggingface/diffusers/issues/7432 | https://api.github.com/repos/huggingface/diffusers/issues/7432 | Image results of Diffusers are quite different, how can I get same image as WebUI with diffusers | ### Describe the bug
I found that the Image results of diffusers and Webui are quite different. Don't know what I did wrong @yiyixuxu @sayakpaul @DN6
WebUI
Diffusers:
![res_0](https://github.com/huggin... | closed | completed | false | 9 | [
"bug"
] | [] | 2024-03-22T07:11:10Z | 2024-03-22T14:18:03Z | 2024-03-22T14:18:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yibolu96 | 93,608,624 | U_kgDOBZRasA | User | false |
huggingface/diffusers | 2,202,459,938 | I_kwDOHa8MBc6DRt8i | 7,438 | https://github.com/huggingface/diffusers/issues/7438 | https://api.github.com/repos/huggingface/diffusers/issues/7438 | XPU loading completed, no progress in generation | ### Describe the bug
I have an Inter ARC GPU. After loading the model to the xpu, there is no progress during generation. Please help me.
### Reproduction
import torch
import intel_extension_for_pytorch as ipex
from diffusers import LCMScheduler, AutoPipelineForText2Image, ControlNetModel
... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2024-03-22T13:07:17Z | 2024-06-29T13:31:47Z | 2024-06-29T13:31:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 823863429 | 6,307,936 | MDQ6VXNlcjYzMDc5MzY= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.