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 | 1,626,867,293 | I_kwDOHa8MBc5g-AZd | 2,704 | https://github.com/huggingface/diffusers/issues/2704 | https://api.github.com/repos/huggingface/diffusers/issues/2704 | Batch Processing for Multi ControlNet. | ### Describe the bug
I am trying to use Batch Processing for Multi ControlNet, but it does not work as intended.
### Reproduction
```
pipe = StableDiffusionControlNetPipeline.from_pretrained(
"my_model", torch_dtype=torch.float16,
controlnet=[
controlnet_pose,
controlnet_hed
],
).... | closed | completed | false | 7 | [
"bug"
] | [
"williamberman"
] | 2023-03-16T07:22:36Z | 2023-03-28T02:54:18Z | 2023-03-28T02:54:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ahmadmustafaanis | 47,111,429 | MDQ6VXNlcjQ3MTExNDI5 | User | false |
huggingface/diffusers | 1,627,512,055 | I_kwDOHa8MBc5hAdz3 | 2,707 | https://github.com/huggingface/diffusers/issues/2707 | https://api.github.com/repos/huggingface/diffusers/issues/2707 | convert_diffusers_to_original_stable_diffusion.py syntax error on line 49 | ### Describe the bug
When I try to convert the weight of a trained model to a ckpt file, I get this error:
File "convert_diffusers_to_original_stable_diffusion.py", line 49
hf_down_res_prefix = f"down_blocks.{i}.resnets.{j}."
^
SyntaxError: invalid synta... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-03-16T13:46:17Z | 2023-04-24T15:03:33Z | 2023-04-24T15:03:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YubinKim16 | 117,627,980 | U_kgDOBwLcTA | User | false |
huggingface/diffusers | 1,627,855,392 | I_kwDOHa8MBc5hBxog | 2,709 | https://github.com/huggingface/diffusers/issues/2709 | https://api.github.com/repos/huggingface/diffusers/issues/2709 | Cannot save pytorch2.0-compiled models | For inference, it works fine. However, if I want to save the model (for example, after dreambooth-fine-tune), it gives error.
```
import torch
from diffusers import StableDiffusionPipeline
import os
pipe = StableDiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
p... | closed | completed | false | 1 | [
"stale"
] | [] | 2023-03-16T16:27:08Z | 2023-04-24T15:03:31Z | 2023-04-24T15:03:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xingzhehe | 28,001,407 | MDQ6VXNlcjI4MDAxNDA3 | User | false |
huggingface/diffusers | 1,354,848,964 | I_kwDOHa8MBc5QwVrE | 271 | https://github.com/huggingface/diffusers/issues/271 | https://api.github.com/repos/huggingface/diffusers/issues/271 | Samplers & Documentation | I am confused about how to properly implement the various samplers with the StableDiffusionPipeline. The default values output garbage. The only preset I could find is for LMS. It would be helpful to have documentation on-hand with how to properly implement them.
Additionally, more options would also be nice, speci... | closed | completed | false | 6 | [] | [] | 2022-08-29T20:41:00Z | 2022-08-30T21:07:19Z | 2022-08-30T21:07:19Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dblunk88 | 39,381,389 | MDQ6VXNlcjM5MzgxMzg5 | User | false |
huggingface/diffusers | 1,628,075,942 | I_kwDOHa8MBc5hCnem | 2,711 | https://github.com/huggingface/diffusers/issues/2711 | https://api.github.com/repos/huggingface/diffusers/issues/2711 | How to configure dim/rank for dreambooth lora training to get a larger lora model | I tried training_dreambooth_lora.py, and the fine-tuned lora model is about 3MB, quite small. How to get a larger Lora model (like 100MB)? How to change the dim rank for lora training, to make the model larger?
Thanks | closed | completed | false | 4 | [] | [] | 2023-03-16T18:42:34Z | 2023-03-23T04:47:11Z | 2023-03-23T04:47:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | THULiusj | 11,749,649 | MDQ6VXNlcjExNzQ5NjQ5 | User | false |
huggingface/diffusers | 1,628,331,024 | I_kwDOHa8MBc5hDlwQ | 2,712 | https://github.com/huggingface/diffusers/issues/2712 | https://api.github.com/repos/huggingface/diffusers/issues/2712 | {'norm_num_groups'} was not found in config. Values will be initialized to default values. | ### Describe the bug
I am trying to fine-tune a stable diffusion model on my custom dataset.
I am following Hugging face s documentation https://huggingface.co/docs/diffusers/training/text2image
I only changed the script file (you will find below my version)
for the train.py file you will find it here:https://githu... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-03-16T21:38:07Z | 2023-03-22T05:05:47Z | 2023-03-22T05:05:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | amine-boukriba | 43,059,827 | MDQ6VXNlcjQzMDU5ODI3 | User | false |
huggingface/diffusers | 1,628,519,173 | I_kwDOHa8MBc5hETsF | 2,714 | https://github.com/huggingface/diffusers/issues/2714 | https://api.github.com/repos/huggingface/diffusers/issues/2714 | How to add a lora model in controlnet pipeline? | **Describe the solution you'd like**
Just like diffusionpipeline, we can simplely add any lora model.
or Can I do it like this using 'StableDiffusionControlNetPipeline'?:
pipe = StableDiffusionControlNetPipeline.from_pretrained()
pipe.unet.load_attn_procs(lora_model_path) | closed | completed | false | 2 | [] | [] | 2023-03-17T00:58:41Z | 2023-03-21T03:53:44Z | 2023-03-21T03:53:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ymzlygw | 17,065,855 | MDQ6VXNlcjE3MDY1ODU1 | User | false |
huggingface/diffusers | 1,628,646,010 | I_kwDOHa8MBc5hEyp6 | 2,716 | https://github.com/huggingface/diffusers/issues/2716 | https://api.github.com/repos/huggingface/diffusers/issues/2716 | How to load checkpoint model? | ### Describe the bug
I want to load checkpoint 2000 model for space.
Here is my model:
https://huggingface.co/ethers/avril15s02-lora-model
### Reproduction
```
import gradio as gr
gr.Interface.load("models/ethers/avril15s02-lora-model/checkpoint-2500").launch()
```
This code cannot work.
```
from dif... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-03-17T03:55:02Z | 2023-08-18T19:53:53Z | 2023-04-24T15:03:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aaurix | 7,252,280 | MDQ6VXNlcjcyNTIyODA= | User | false |
huggingface/diffusers | 1,628,733,999 | I_kwDOHa8MBc5hFIIv | 2,719 | https://github.com/huggingface/diffusers/issues/2719 | https://api.github.com/repos/huggingface/diffusers/issues/2719 | [LoRA] allow fine-tuning of the text encoder with LoRA (using `peft`) | We have had many requests (rightfully so) for allowing to fine-tune the text encoder with LoRA (such as https://github.com/huggingface/diffusers/issues/2683). This is quite useful for improving the quality of the generated samples. This issue thread aims to discuss a solution candidate that combines our LoRA attention ... | closed | completed | false | 13 | [
"stale"
] | [
"sayakpaul"
] | 2023-03-17T05:48:32Z | 2023-05-01T15:03:23Z | 2023-05-01T15:03:23Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,355,169,685 | I_kwDOHa8MBc5Qxj-V | 272 | https://github.com/huggingface/diffusers/issues/272 | https://api.github.com/repos/huggingface/diffusers/issues/272 | Support K-LMS for inpainting | You have added K-LMS support to [image_to_image.py](https://github.com/huggingface/diffusers/blob/main/examples/inference/image_to_image.py), can you add it to [inpainting.py](https://github.com/huggingface/diffusers/blob/main/examples/inference/inpainting.py)?
| closed | completed | false | 3 | [] | [
"Inkorak"
] | 2022-08-30T05:01:13Z | 2022-09-20T17:10:46Z | 2022-09-20T17:10:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Inkorak | 52,286,717 | MDQ6VXNlcjUyMjg2NzE3 | User | false |
huggingface/diffusers | 1,628,735,145 | I_kwDOHa8MBc5hFIap | 2,720 | https://github.com/huggingface/diffusers/issues/2720 | https://api.github.com/repos/huggingface/diffusers/issues/2720 | what's the difference between 'train_text_to_image_lora.py' and 'train_dreambooth_lora.py' ? | what's the difference between examples/text_to_image/train_text_to_image_lora.py and examples/dreambooth/train_dreambooth_lora.py ? | closed | completed | false | 5 | [] | [] | 2023-03-17T05:50:02Z | 2024-08-12T09:48:37Z | 2023-03-21T03:47:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anonymoussss | 12,672,787 | MDQ6VXNlcjEyNjcyNzg3 | User | false |
huggingface/diffusers | 1,629,069,069 | I_kwDOHa8MBc5hGZ8N | 2,722 | https://github.com/huggingface/diffusers/issues/2722 | https://api.github.com/repos/huggingface/diffusers/issues/2722 | ControleNet unet model are not same for different models after conversion | ### Describe the bug
I try to convert different models from the [HuggingFace](https://huggingface.co/lllyasviel/ControlNet/tree/main/models) with this function:
https://github.com/huggingface/diffusers/blob/f20c8f5a1aba27f5972cad50516f18ba516e4d9e/src/diffusers/pipelines/stable_diffusion/convert_from_ckpt.py#L958
... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-03-17T10:28:03Z | 2023-03-17T13:15:51Z | 2023-03-17T13:15:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KernelA | 17,554,646 | MDQ6VXNlcjE3NTU0NjQ2 | User | false |
huggingface/diffusers | 1,630,018,082 | I_kwDOHa8MBc5hKBoi | 2,728 | https://github.com/huggingface/diffusers/issues/2728 | https://api.github.com/repos/huggingface/diffusers/issues/2728 | Cannot load safetensors files with UNet2DConditionModel.from_pretrained | ### Describe the bug
I am trying to load a UNet2DConditionModel model with the from_pretrained() function from a model on the hub: https://huggingface.co/xkkkk/realism_engine/tree/main/unet However, the code itself only looks for .bin files, not .safetensor files. I can't seem to find a parameter to pass it to look fo... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-03-17T23:15:38Z | 2023-03-17T23:56:00Z | 2023-03-17T23:56:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | enoreyes | 13,440,436 | MDQ6VXNlcjEzNDQwNDM2 | User | false |
huggingface/diffusers | 1,630,086,807 | I_kwDOHa8MBc5hKSaX | 2,729 | https://github.com/huggingface/diffusers/issues/2729 | https://api.github.com/repos/huggingface/diffusers/issues/2729 | Error when loading models with cache_dir set | I'm suddenly having a problem when I set cache_dir=path in .from_pretrained and can't get around it. It used to work fine a few days ago, and it works when cache_dir=None, but now when I have it pointing to my Google Drive on Colab, I'm getting this message:
```
File "<ipython-input-7-cc85556e36ff>", line 8965, in... | closed | completed | false | 9 | [] | [] | 2023-03-18T02:06:07Z | 2023-03-21T08:31:22Z | 2023-03-21T03:45:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Skquark | 3,192,935 | MDQ6VXNlcjMxOTI5MzU= | User | false |
huggingface/diffusers | 1,630,918,483 | I_kwDOHa8MBc5hNddT | 2,735 | https://github.com/huggingface/diffusers/issues/2735 | https://api.github.com/repos/huggingface/diffusers/issues/2735 | 无法在cmd中粘贴token/ Cannot paste token in cmd Body: | I am trying to convert the safetensors model to an onnx compatible model, following this tutorial (https://huggingface.co/transformers/serialization.html). When I log in to hugging face in cmd, I need to enter a token, but I cannot paste my token in the token text box. I have copied my token, but I cannot paste it into... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-03-19T12:22:08Z | 2023-04-26T15:03:57Z | 2023-04-26T15:03:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sisyphbaous-DT-Project | 96,930,391 | U_kgDOBccKVw | User | false |
huggingface/diffusers | 1,630,946,829 | I_kwDOHa8MBc5hNkYN | 2,736 | https://github.com/huggingface/diffusers/issues/2736 | https://api.github.com/repos/huggingface/diffusers/issues/2736 | Any plans to add ModelScope's 1.7B text2video synthesis diffusion model? | ### Model/Pipeline/Scheduler description
Hello!
There seems to be a new 1.7B-parameter Diffusion-based model by ModelScope allowing text2video synthesis as noted by AKHaliq https://twitter.com/_akhaliq/status/1637321077553606657?s=20. Both the [model implementation](https://github.com/modelscope/modelscope/tree/m... | closed | completed | false | 4 | [] | [] | 2023-03-19T13:40:02Z | 2023-03-22T21:32:30Z | 2023-03-22T21:32:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 1,631,028,342 | I_kwDOHa8MBc5hN4R2 | 2,737 | https://github.com/huggingface/diffusers/issues/2737 | https://api.github.com/repos/huggingface/diffusers/issues/2737 | CLIP Guidance support for DPMSolverMultistepScheduler | **Is your feature request related to a problem? Please describe.**
Hi, CLIP guidance can be very useful. I saw that there is a [community pipeline](https://github.com/huggingface/diffusers/blob/main/examples/community/clip_guided_stable_diffusion.py) but it only supports specific schedulers. Specifically, it does not ... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-19T17:03:04Z | 2023-04-28T15:04:00Z | 2023-04-28T15:04:00Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yuvalkirstain | 57,996,478 | MDQ6VXNlcjU3OTk2NDc4 | User | false |
huggingface/diffusers | 1,631,288,858 | I_kwDOHa8MBc5hO34a | 2,739 | https://github.com/huggingface/diffusers/issues/2739 | https://api.github.com/repos/huggingface/diffusers/issues/2739 | [Question] Can I load mutiple lora model? | I notice that diffusers can use `load_attn_procs` api to use lora model
But when I want to load more than one lora model, it seems not work.
And I want to dynamic load and unload lora model like `AUTOMATIC1111/stable-diffusion-webui`
How can I do like this | closed | completed | false | 14 | [
"stale"
] | [] | 2023-03-20T02:58:19Z | 2023-07-30T08:10:39Z | 2023-06-05T15:03:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | big-thousand | 114,633,619 | U_kgDOBtUrkw | User | false |
huggingface/diffusers | 1,631,455,494 | I_kwDOHa8MBc5hPgkG | 2,743 | https://github.com/huggingface/diffusers/issues/2743 | https://api.github.com/repos/huggingface/diffusers/issues/2743 | How to use "learned_embeds.bin" files? | I finetuend on textual inversion and got "learned_embeds.bin" files. I did not save checkpoints. Can I use these learned embedding files to test the model in specific training steps? | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-20T06:07:55Z | 2023-04-28T15:03:58Z | 2023-04-28T15:03:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yanzengatsmu | 43,662,186 | MDQ6VXNlcjQzNjYyMTg2 | User | false |
huggingface/diffusers | 1,631,483,201 | I_kwDOHa8MBc5hPnVB | 2,744 | https://github.com/huggingface/diffusers/issues/2744 | https://api.github.com/repos/huggingface/diffusers/issues/2744 | MultiControlNet failed to execute | Dear great author,
I've tried to apply multi-controlnet
While executing the code
"
pipe = StableDiffusionControlNetPipeline.from_pretrained(
"example/a-sd15-variant-model", torch_dtype=torch.float16,
controlnet=control_net_lists,
).to("cuda")
"
It will return the error as following ... | closed | completed | false | 3 | [] | [] | 2023-03-20T06:35:29Z | 2023-03-21T07:07:11Z | 2023-03-21T07:07:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | waquey | 31,580,848 | MDQ6VXNlcjMxNTgwODQ4 | User | false |
huggingface/diffusers | 1,631,629,031 | I_kwDOHa8MBc5hQK7n | 2,745 | https://github.com/huggingface/diffusers/issues/2745 | https://api.github.com/repos/huggingface/diffusers/issues/2745 | get core dump issue using diffusers in TPU | ### Describe the bug
I've been running into core dump errors using diffusers in flax/TPU
I got this issue when trying to run the text_to_image training script (followed these exact steps in the instruction https://github.com/huggingface/diffusers/tree/main/examples/text_to_image#training-with-flaxjax) - I was able ... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-03-20T08:32:35Z | 2023-03-24T15:35:46Z | 2023-03-21T15:10:05Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 1,631,804,105 | I_kwDOHa8MBc5hQ1rJ | 2,747 | https://github.com/huggingface/diffusers/issues/2747 | https://api.github.com/repos/huggingface/diffusers/issues/2747 | Update docs | On this page, "https://huggingface.co/docs/diffusers/main/en/optimization/torch2.0", it says "... as the stable version is yet to be released.". | closed | completed | false | 2 | [] | [] | 2023-03-20T10:20:50Z | 2023-03-21T17:42:39Z | 2023-03-21T17:42:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tolgacangoz | 46,008,593 | MDQ6VXNlcjQ2MDA4NTkz | User | false |
huggingface/diffusers | 1,355,861,143 | I_kwDOHa8MBc5Q0MyX | 275 | https://github.com/huggingface/diffusers/issues/275 | https://api.github.com/repos/huggingface/diffusers/issues/275 | About the performance of the safety checker? | This is an issue coming from https://github.com/huggingface/diffusers/pull/213#issuecomment-1228277919
My code referred to [stable diffusion](https://github.com/CompVis/stable-diffusion/blob/69ae4b35e0a0f... | closed | completed | false | 5 | [
"Safety Checker"
] | [
"patrickvonplaten"
] | 2022-08-30T14:28:39Z | 2022-10-17T11:26:07Z | 2022-09-13T15:58:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | BIGJUN777 | 35,102,882 | MDQ6VXNlcjM1MTAyODgy | User | false |
huggingface/diffusers | 1,632,022,765 | I_kwDOHa8MBc5hRrDt | 2,750 | https://github.com/huggingface/diffusers/issues/2750 | https://api.github.com/repos/huggingface/diffusers/issues/2750 | How about add face-landmarks at the control-net support? | https://huggingface.co/georgefen/Face-Landmark-ControlNet
I found new trial of controlnet though face-landmark.
It's useful for changing or modifying the face using diffusers.
Could you support this at the next release? | closed | completed | false | 8 | [
"stale"
] | [] | 2023-03-20T12:36:18Z | 2023-05-22T15:03:37Z | 2023-05-22T15:03:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | circuluspibo | 50,613,069 | MDQ6VXNlcjUwNjEzMDY5 | User | false |
huggingface/diffusers | 1,632,356,790 | I_kwDOHa8MBc5hS8m2 | 2,754 | https://github.com/huggingface/diffusers/issues/2754 | https://api.github.com/repos/huggingface/diffusers/issues/2754 | Virus dected in Latent Upscaler | Hi there, suddenly my latent upscaler stopped working in my gui, then I realyzed that the pytorch bin file was missing.
When I tried to download from here, chrome accused a virus in the file. Then I checked windows defender and saw that the file was missing because windows defender detected it as a virus and deleted i... | closed | completed | false | 6 | [
"stale"
] | [] | 2023-03-20T15:30:59Z | 2023-04-28T15:03:57Z | 2023-04-28T15:03:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | brurpo | 24,902,022 | MDQ6VXNlcjI0OTAyMDIy | User | false |
huggingface/diffusers | 1,632,638,654 | I_kwDOHa8MBc5hUBa- | 2,755 | https://github.com/huggingface/diffusers/issues/2755 | https://api.github.com/repos/huggingface/diffusers/issues/2755 | [Bug][StableDiffusion] Allow loading models with Float16 in `load_pipeline_from_original_stable_diffusion_ckpt` | ### Describe the bug
Models loaded from .ckpt / .safetensors file are using Float32 automatically, with no obvious way to change this behavior.
### Reproduction
- Import `load_pipeline_from_original_stable_diffusion_ckpt` from `diffusers.pipelines.stable_diffusion.convert_from_ckpt`
- Load model from .ckpt file
- ... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-03-20T18:16:51Z | 2023-06-28T13:27:45Z | 2023-03-28T14:06:50Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Stax124 | 60,222,162 | MDQ6VXNlcjYwMjIyMTYy | User | false |
huggingface/diffusers | 1,633,020,096 | I_kwDOHa8MBc5hVejA | 2,758 | https://github.com/huggingface/diffusers/issues/2758 | https://api.github.com/repos/huggingface/diffusers/issues/2758 | pytorch2.0 rocm torch.compile gives RuntimeWarning and generated black images | I was testing the https://huggingface.co/docs/diffusers/optimization/torch2.0 examples,
The normal 'Accelerated Transformers implementation' was successful.
But the torch.compile example:
pipe.unet = torch.compile(pipe.unet)
image = pipe(prompt).images[0]
gives warning:
pipelines/pipeline_utils.py:1023: Runtime... | closed | completed | false | 17 | [
"stale"
] | [] | 2023-03-20T23:39:49Z | 2024-05-14T10:26:16Z | 2023-06-17T15:03:03Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ttio2tech | 125,389,792 | U_kgDOB3lL4A | User | false |
huggingface/diffusers | 1,633,453,978 | I_kwDOHa8MBc5hXIea | 2,761 | https://github.com/huggingface/diffusers/issues/2761 | https://api.github.com/repos/huggingface/diffusers/issues/2761 | Documentation needed for UNet1DModel.extra_in_channels | ### Describe the bug
Documentation needed for UNet1DModel.extra_in_channels parameter, which is undocumented.
### Reproduction
https://huggingface.co/docs/diffusers/api/models#diffusers.UNet1DModel
### Logs
_No response_
### System Info
n/a | closed | completed | false | 6 | [
"bug",
"good first issue"
] | [] | 2023-03-21T08:48:49Z | 2023-03-28T16:04:49Z | 2023-03-28T16:04:48Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | turian | 65,918 | MDQ6VXNlcjY1OTE4 | User | false |
huggingface/diffusers | 1,633,524,388 | I_kwDOHa8MBc5hXZqk | 2,763 | https://github.com/huggingface/diffusers/issues/2763 | https://api.github.com/repos/huggingface/diffusers/issues/2763 | Increased VRAM usage with ControlNet when torch.backends.cudnn.benchmark=True | ### Describe the bug
Not entirely sure if this is a bug but setting `torch.backends.cudnn.benchmark=True` as per https://huggingface.co/docs/diffusers/optimization/fp16 significantly increases vram usage when used with `StableDiffusionControlNetPipeline`
With the example script I get the output
`2,746,436,608, 2... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-03-21T09:33:57Z | 2023-04-20T16:44:02Z | 2023-04-20T16:44:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | d1g1t | 661,828 | MDQ6VXNlcjY2MTgyOA== | User | false |
huggingface/diffusers | 1,633,587,482 | I_kwDOHa8MBc5hXpEa | 2,765 | https://github.com/huggingface/diffusers/issues/2765 | https://api.github.com/repos/huggingface/diffusers/issues/2765 | How to convert LoRA trained by diffusers to work on stable-diffusion-webui? | null | closed | completed | false | 5 | [
"stale"
] | [] | 2023-03-21T10:11:39Z | 2024-04-30T07:53:40Z | 2023-04-28T15:03:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zhaoyuzhi | 13,333,802 | MDQ6VXNlcjEzMzMzODAy | User | false |
huggingface/diffusers | 1,634,451,659 | I_kwDOHa8MBc5ha8DL | 2,769 | https://github.com/huggingface/diffusers/issues/2769 | https://api.github.com/repos/huggingface/diffusers/issues/2769 | Converting from Stable diffusion v2.1 CKPT or safetensors to diffusers, then back to CKPT/safetensors breaks the model | ### Describe the bug
I noticed when converting a v2.1 SD model to diffusers then back to CKPT again makes the model unusable.
### Reproduction
Try converting https://huggingface.co/stabilityai/stable-diffusion-2-1/resolve/main/v2-1_768-ema-pruned.ckpt here https://colab.research.google.com/github/TheLastBen/fa... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-03-21T17:59:43Z | 2023-09-28T09:18:06Z | 2023-05-02T15:03:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TheLastBen | 110,583,491 | U_kgDOBpdeww | User | false |
huggingface/diffusers | 1,356,076,883 | I_kwDOHa8MBc5Q1BdT | 277 | https://github.com/huggingface/diffusers/issues/277 | https://api.github.com/repos/huggingface/diffusers/issues/277 | More Samplers/Schedulers, e.g. Euler | Thank you guys for throwing in the documentation. Segregating this from the other Issue.
Is there any possibility to include more of the k-diffusion samplers? https://github.com/crowsonkb/k-diffusion/blob/master/k_diffusion/sampling.py
Heuler for example would be nice | closed | completed | false | 25 | [
"New scheduler",
"Good second issue"
] | [] | 2022-08-30T17:18:17Z | 2024-11-21T11:17:23Z | 2024-11-21T11:17:13Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dblunk88 | 39,381,389 | MDQ6VXNlcjM5MzgxMzg5 | User | false |
huggingface/diffusers | 1,634,620,816 | I_kwDOHa8MBc5hblWQ | 2,772 | https://github.com/huggingface/diffusers/issues/2772 | https://api.github.com/repos/huggingface/diffusers/issues/2772 | The `train_controlnet.py` script prevents valid local datasets from being used | ### Describe the bug
The `train_controlnet.py` script has a `--train_data_dir` flag for training with local datasets. However, using it seems impossible. This is because the code works by assuming the data is in the `imagefolder` format, which AFAICT tell from all the examples, expects a single image, but the controln... | closed | completed | false | 7 | [
"bug"
] | [
"pcuenca",
"williamberman",
"yiyixuxu"
] | 2023-03-21T20:07:44Z | 2023-04-07T18:12:42Z | 2023-04-04T17:37:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jfischoff | 168,507 | MDQ6VXNlcjE2ODUwNw== | User | false |
huggingface/diffusers | 1,634,935,998 | I_kwDOHa8MBc5hcyS- | 2,774 | https://github.com/huggingface/diffusers/issues/2774 | https://api.github.com/repos/huggingface/diffusers/issues/2774 | error converting controlnet safetensors file to diffusers | ### Describe the bug
Hi, I try to convert [safetensors file](https://huggingface.co/thibaud/controlnet-sd21/resolve/main/openpose-sd21-safe.safetensors) using `convert_original_controlnet_to_diffusers.py`, but still got errors like:
`new_checkpoint["time_embedding.linear_1.weight"] = unet_state_dict["time_embed.0.w... | closed | completed | false | 9 | [
"bug",
"stale"
] | [
"williamberman"
] | 2023-03-22T01:57:29Z | 2024-04-13T08:41:23Z | 2023-06-05T17:25:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | goodspb | 10,245,385 | MDQ6VXNlcjEwMjQ1Mzg1 | User | false |
huggingface/diffusers | 1,635,150,748 | I_kwDOHa8MBc5hdmuc | 2,775 | https://github.com/huggingface/diffusers/issues/2775 | https://api.github.com/repos/huggingface/diffusers/issues/2775 | RuntimeError: torch._dynamo.optimize is called on a non function object. | ### Describe the bug
When I am trying to run the example given by the authors, which is on the pokemon dataset. I just follows the instruction on huggingface and never revise the code. but I got this error:
RuntimeError:
torch._dynamo.optimize is called on a non function object.
If this is a callable class, pl... | closed | completed | false | 27 | [
"bug"
] | [
"pcuenca"
] | 2023-03-22T06:41:17Z | 2023-07-24T08:49:07Z | 2023-07-24T08:49:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xiaosaxuexige | 117,796,641 | U_kgDOBwVvIQ | User | false |
huggingface/diffusers | 1,635,198,796 | I_kwDOHa8MBc5hdydM | 2,776 | https://github.com/huggingface/diffusers/issues/2776 | https://api.github.com/repos/huggingface/diffusers/issues/2776 | Dreambooth train on 8GB should fix DeepSpeed==0.8.2 otherwise raise ZeRORuntimeException | ### Describe the bug
I followed the instructions to [train the model on a 8GB gpu](https://github.com/huggingface/diffusers/tree/main/examples/dreambooth#training-on-a-8-gb-gpu), installed DeepSpeed `pip3 install DeepSpeed`, run `accelerate launch`, and then got ZeRORuntimeException which suggest to set <"zero_force_... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-03-22T07:19:08Z | 2023-12-11T07:48:24Z | 2023-04-29T15:03:05Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | polym | 2,860,926 | MDQ6VXNlcjI4NjA5MjY= | User | false |
huggingface/diffusers | 1,635,199,012 | I_kwDOHa8MBc5hdygk | 2,777 | https://github.com/huggingface/diffusers/issues/2777 | https://api.github.com/repos/huggingface/diffusers/issues/2777 | Dreambooth for Alt Diffusion | Hello, I use DreamBooth's solution to fine-tune AltDiffusion. The data uses the toy dataset in ReadME, but I find that I can't get good results, but the same replacement with Stable Diffusion works very well. Can you provide an example of using DreamBooth to fine-tune Alt Diffusion? | closed | completed | false | 2 | [] | [] | 2023-03-22T07:19:23Z | 2023-03-28T02:36:08Z | 2023-03-28T02:36:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cjt222 | 17,508,662 | MDQ6VXNlcjE3NTA4NjYy | User | false |
huggingface/diffusers | 1,635,326,652 | I_kwDOHa8MBc5heRq8 | 2,780 | https://github.com/huggingface/diffusers/issues/2780 | https://api.github.com/repos/huggingface/diffusers/issues/2780 | ValueError: Attempting to unscale FP16 gradients. in diffusers/examples/research_projects/lora/train_text_to_image_lora.py | ### Describe the bug
Thanks for this excellent work！
When I run diffusers/examples/research_projects/lora/train_text_to_image_lora.py, I encountered this issue:
ValueError: Attempting to unscale FP16 gradients
Could you please tell me how can I fix it?
### Reproduction
accelerate launch train_text_to_image_lora... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-03-22T08:58:41Z | 2024-01-17T06:19:58Z | 2023-03-24T06:15:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | liushanyuan18 | 46,879,442 | MDQ6VXNlcjQ2ODc5NDQy | User | false |
huggingface/diffusers | 1,635,394,024 | I_kwDOHa8MBc5heiHo | 2,781 | https://github.com/huggingface/diffusers/issues/2781 | https://api.github.com/repos/huggingface/diffusers/issues/2781 | No attribute '_cuda_setDevice' when using gloo for data training | I tried using Dreambooth and Textual Inversion to train images and run into an error that haven't found the solution. I'm still able to use diffusers for text-to-image or img2img with CUDA
```ssh
module 'torch._C' has no attribute '_cuda_setDevice'
```
- Script used [examples/dreambooth](https://github.com/hugg... | closed | completed | false | 1 | [] | [] | 2023-03-22T09:38:04Z | 2023-03-23T07:08:42Z | 2023-03-23T07:08:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hieunc229 | 8,464,869 | MDQ6VXNlcjg0NjQ4Njk= | User | false |
huggingface/diffusers | 1,635,990,709 | I_kwDOHa8MBc5hgzy1 | 2,782 | https://github.com/huggingface/diffusers/issues/2782 | https://api.github.com/repos/huggingface/diffusers/issues/2782 | "sequential model offloading" Error | ### Describe the bug
Hi all!
**Example code:**
pipe = StableDiffusionPipeline.from_pretrained(
"runwayml/stable-diffusion-v1-5",
revision="fp16",
torch_dtype=torch.float16,
use_auth_token=access_token,
device_map="auto",
)
And we started getting an error
`ValueError: It seems like you... | closed | completed | false | 8 | [
"bug",
"stale"
] | [] | 2023-03-22T15:20:24Z | 2023-05-30T15:03:42Z | 2023-05-30T15:03:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Feniksss | 7,080,230 | MDQ6VXNlcjcwODAyMzA= | User | false |
huggingface/diffusers | 1,636,442,104 | I_kwDOHa8MBc5hih_4 | 2,783 | https://github.com/huggingface/diffusers/issues/2783 | https://api.github.com/repos/huggingface/diffusers/issues/2783 | Create a img2img (+ inpainting) controlnet pipeline | **Is your feature request related to a problem? Please describe.**
In auto1111 I can use img2img with inpainting in combination with controlnet canny edge detection to generate a new image.
With current tools in diffusers it appears I can't do that. I can only do text2img with controlnet
**Describe the solution ... | closed | completed | false | 7 | [
"stale"
] | [] | 2023-03-22T20:19:59Z | 2023-07-07T15:03:58Z | 2023-07-07T15:03:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tiimgreen | 4,130,574 | MDQ6VXNlcjQxMzA1NzQ= | User | false |
huggingface/diffusers | 1,636,532,684 | I_kwDOHa8MBc5hi4HM | 2,784 | https://github.com/huggingface/diffusers/issues/2784 | https://api.github.com/repos/huggingface/diffusers/issues/2784 | [Feature request]: Dreambooth support for text2video diffusion models | ### Model/Pipeline/Scheduler description
Now that [the first text2video pipeline is merged](https://github.com/huggingface/diffusers/pull/2738), the next frontier is to enable efficient fine-tuning of these models.
There's already Dreambooth method for the Diffusion-based text2img models, consisting of making a reg... | closed | completed | false | 11 | [
"stale"
] | [] | 2023-03-22T21:38:44Z | 2023-05-30T15:03:40Z | 2023-05-30T15:03:40Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kabachuha | 14,872,007 | MDQ6VXNlcjE0ODcyMDA3 | User | false |
huggingface/diffusers | 1,636,572,420 | I_kwDOHa8MBc5hjB0E | 2,785 | https://github.com/huggingface/diffusers/issues/2785 | https://api.github.com/repos/huggingface/diffusers/issues/2785 | Add M1/ M2 support to text-to-video pipeline | **Is your feature request related to a problem? Please describe.**
> Currently after installing Mac OS 13.3 to get pipeline_text_to_video_synth.py to execute I was disappointed to find out the script for Text-to-Video wasn't working on my Mac M2. It seems there's no fallback in the pipeline for non-cuda enabled devi... | closed | completed | false | 22 | [
"wip"
] | [] | 2023-03-22T22:19:22Z | 2026-01-29T16:09:34Z | 2024-11-25T13:05:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yearofthewhopper | 8,598,597 | MDQ6VXNlcjg1OTg1OTc= | User | false |
huggingface/diffusers | 1,636,605,157 | I_kwDOHa8MBc5hjJzl | 2,786 | https://github.com/huggingface/diffusers/issues/2786 | https://api.github.com/repos/huggingface/diffusers/issues/2786 | StableDiffusionDepth2ImgPipeline can't be used with numpy inputs and transformers-4.27 | ### Describe the bug
Like most other pipelines, `StableDiffusionDepth2ImgPipeline` can take either a numpy array or a PIL image as input image.
In diffusers, when using a numpy array, inputs are expected to be in the [-1, 1] range.
In `prepare_depth_map`, `StableDiffusionDepth2ImgPipeline` passes the input ima... | closed | completed | false | 3 | [
"bug"
] | [
"yiyixuxu"
] | 2023-03-22T22:52:28Z | 2023-03-23T22:04:01Z | 2023-03-23T22:04:01Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | antoche | 1,627,384 | MDQ6VXNlcjE2MjczODQ= | User | false |
huggingface/diffusers | 1,636,925,391 | I_kwDOHa8MBc5hkX_P | 2,789 | https://github.com/huggingface/diffusers/issues/2789 | https://api.github.com/repos/huggingface/diffusers/issues/2789 | [UNet3DConditionModel] implement rest of the test cases (LoRA, Slow tests) | For the LoRA and other attention-block related tests, we need to have `attn_processors` implemented for the `UNet3DConditionModel`.
`tests/models/test_models_unet_3d_condition.py` has sufficient comments about these. | closed | completed | false | 7 | [
"stale"
] | [
"sayakpaul"
] | 2023-03-23T06:22:30Z | 2023-05-09T02:07:43Z | 2023-05-09T02:07:42Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,638,266,448 | I_kwDOHa8MBc5hpfZQ | 2,797 | https://github.com/huggingface/diffusers/issues/2797 | https://api.github.com/repos/huggingface/diffusers/issues/2797 | Can't reproduce results from stability-sdk using StableDiffusionPipeline | I am trying to reproduce generated images between using the StableDiffusionPipeline and using the [stability-sdk](https://github.com/Stability-AI/stability-sdk). For the following example I used stable-diffusion-2-1-base but I tried all the models available in both and still saw a difference (though this one was the mo... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-03-23T20:32:38Z | 2023-05-01T15:03:19Z | 2023-05-01T15:03:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pmlakner | 15,306,382 | MDQ6VXNlcjE1MzA2Mzgy | User | false |
huggingface/diffusers | 1,638,273,487 | I_kwDOHa8MBc5hphHP | 2,798 | https://github.com/huggingface/diffusers/issues/2798 | https://api.github.com/repos/huggingface/diffusers/issues/2798 | Clarity embedding work | Looking at this colab for loading embedding it uses the `.bin`
https://colab.research.google.com/github/huggingface/notebooks/blob/main/diffusers/stable_conceptualizer_inference.ipynb#scrollTo=bJx50Ta37392
There is a conversion script for sd -> diffusers
I was wondering does this also apply to `.pt` files for e... | closed | completed | false | 4 | [] | [] | 2023-03-23T20:36:27Z | 2023-04-08T01:03:35Z | 2023-04-08T01:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NicholasKao1029 | 45,542,006 | MDQ6VXNlcjQ1NTQyMDA2 | User | false |
huggingface/diffusers | 1,638,498,689 | I_kwDOHa8MBc5hqYGB | 2,799 | https://github.com/huggingface/diffusers/issues/2799 | https://api.github.com/repos/huggingface/diffusers/issues/2799 | Raise Error on Mismatched Model Checkpoint and Pipeline | (Follow up to #2135)
If I load a model checkpoint using the `from_pretrained(...)` method which isn't correctly matched to the right `DiffusionPipeline` subclass, such as using the `runwayml/stable-diffusion-v1-5` text-to-image Stable Diffusion checkpoint with the `StableDiffusionInpaintPipeline` inpainting pipeline... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-03-23T23:48:59Z | 2023-05-01T15:03:17Z | 2023-05-01T15:03:17Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dg845 | 58,458,699 | MDQ6VXNlcjU4NDU4Njk5 | User | false |
huggingface/diffusers | 1,279,056,350 | I_kwDOHa8MBc5MPNne | 28 | https://github.com/huggingface/diffusers/issues/28 | https://api.github.com/repos/huggingface/diffusers/issues/28 | Support for imagen-pytorch? | Hi,
Curious if I could use this alongside the imagen-pytorch opensource model - https://github.com/lucidrains/imagen-pytorch
Any help with this would be great! | closed | completed | false | 3 | [] | [] | 2022-06-21T20:47:56Z | 2022-07-21T19:03:59Z | 2022-07-21T19:03:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 1,356,230,955 | I_kwDOHa8MBc5Q1nEr | 280 | https://github.com/huggingface/diffusers/issues/280 | https://api.github.com/repos/huggingface/diffusers/issues/280 | Add DEIS scheduler | **Is your feature request related to a problem? Please describe.**
State of the art scheduler is missing.
**Describe the solution you'd like**
DEIS scheduler should be integrated: https://github.com/qsh-zh/deis | closed | completed | false | 12 | [
"good first issue",
"New scheduler"
] | [] | 2022-08-30T19:53:38Z | 2023-02-13T13:09:30Z | 2023-02-13T13:09:30Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,638,556,336 | I_kwDOHa8MBc5hqmKw | 2,800 | https://github.com/huggingface/diffusers/issues/2800 | https://api.github.com/repos/huggingface/diffusers/issues/2800 | DiffEdit: Diffusion-based semantic image editing with mask guidance | ### Model/Pipeline/Scheduler description
DiffEdit is a zero-shot inpainting method given a text-to-image denoising diffusion model, requiring no additional fine-tuning or parameters.
The main innovations pioneered in DiffEdit are:
- Automated mask inference given the original and new conditioning text using the ... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-24T01:05:46Z | 2023-05-01T15:03:16Z | 2023-05-01T15:03:16Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clarencechen | 8,482,341 | MDQ6VXNlcjg0ODIzNDE= | User | false |
huggingface/diffusers | 1,638,655,509 | I_kwDOHa8MBc5hq-YV | 2,801 | https://github.com/huggingface/diffusers/issues/2801 | https://api.github.com/repos/huggingface/diffusers/issues/2801 | "DPMSolverMultistepScheduler "Is not" dpm++ 2M K ", "DPMSolverMultistepScheduler" = "dpm++ 2 m" Karras doesn't seem to be effective | Dear developers, first of all thank you for your pay, after I a lot of testing I found "DPMSolverMultistepScheduler
"Is not" dpm++ 2M K ",
"DPMSolverMultistepScheduler" = "dpm++ 2 m" Karras doesn't seem to be effective, it is very influence some model generated effect, can tell me the way to solve this problem?
This... | closed | completed | false | 12 | [
"stale"
] | [] | 2023-03-24T03:25:10Z | 2023-05-01T15:03:15Z | 2023-05-01T15:03:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | XXXKAY | 119,796,959 | U_kgDOByP03w | User | false |
huggingface/diffusers | 1,638,674,244 | I_kwDOHa8MBc5hrC9E | 2,802 | https://github.com/huggingface/diffusers/issues/2802 | https://api.github.com/repos/huggingface/diffusers/issues/2802 | How to prune diffusers model? | Pruning .ckpt is available, but is there script to prune diffusers? | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-24T03:53:58Z | 2023-05-22T15:03:28Z | 2023-05-22T15:03:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | adhikjoshi | 11,740,719 | MDQ6VXNlcjExNzQwNzE5 | User | false |
huggingface/diffusers | 1,638,682,624 | I_kwDOHa8MBc5hrFAA | 2,803 | https://github.com/huggingface/diffusers/issues/2803 | https://api.github.com/repos/huggingface/diffusers/issues/2803 | Cannot Build Docs Locally on WSL2 Ubuntu | ### Describe the bug
I tried generating the docs locally via
```bash
doc-builder preview diffusers docs/source/en
```
after following the instructions for [generating the documentation](https://github.com/huggingface/diffusers/tree/main/docs) and the [instructions for installing Node.js v14.x](https://github.c... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-03-24T04:04:16Z | 2023-03-28T05:50:54Z | 2023-03-28T02:18:14Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dg845 | 58,458,699 | MDQ6VXNlcjU4NDU4Njk5 | User | false |
huggingface/diffusers | 1,638,883,373 | I_kwDOHa8MBc5hr2At | 2,806 | https://github.com/huggingface/diffusers/issues/2806 | https://api.github.com/repos/huggingface/diffusers/issues/2806 | printing pipeline doesn't take customized ones into account | ### Describe the bug
I want to print current pipeline status as stated here [Diffusers doc](https://huggingface.co/docs/diffusers/using-diffusers/loading) `How does loading work?` section using `print(pipe)`.
It prints ok when default config is used, but if I change (customize) pipeline, it still prints same(defa... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-03-24T07:39:40Z | 2023-05-14T15:03:17Z | 2023-05-14T15:03:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | moono | 6,761,398 | MDQ6VXNlcjY3NjEzOTg= | User | false |
huggingface/diffusers | 1,638,888,947 | I_kwDOHa8MBc5hr3Xz | 2,807 | https://github.com/huggingface/diffusers/issues/2807 | https://api.github.com/repos/huggingface/diffusers/issues/2807 | Update "Controlled generation" doc to include the latest goodies | Doc: https://huggingface.co/docs/diffusers/main/en/using-diffusers/controlling_generation
Rather than updating the doc with the addition of each new model / pipeline, I'd prefer to have a batch 2/3 new models / pipelines, and then update this doc for better visibility. I think maintenance-wise it's also less burdern... | closed | completed | false | 2 | [
"stale"
] | [
"sayakpaul"
] | 2023-03-24T07:44:49Z | 2023-05-01T15:03:13Z | 2023-05-01T15:03:13Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,356,336,488 | I_kwDOHa8MBc5Q2A1o | 281 | https://github.com/huggingface/diffusers/issues/281 | https://api.github.com/repos/huggingface/diffusers/issues/281 | Optimize model loading by natively using Accelerate, please. | **Is your feature request related to a problem? Please describe.**
Loading pretrained models, for example `UNet2DConditionModel` peaks at 8GB when loading a pretrained one, while it takes a lot less memory when it is fully loaded to the GPU. That is happens likely because there are some redundancies between loaded wei... | closed | completed | false | 16 | [] | [] | 2022-08-30T21:32:22Z | 2022-10-04T13:21:41Z | 2022-10-04T13:21:41Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | piEsposito | 47,679,710 | MDQ6VXNlcjQ3Njc5NzEw | User | false |
huggingface/diffusers | 1,639,630,543 | I_kwDOHa8MBc5husbP | 2,816 | https://github.com/huggingface/diffusers/issues/2816 | https://api.github.com/repos/huggingface/diffusers/issues/2816 | Deepspeed or FSDP cannot be used for fine-tuning two models together such as text encoder and unet | Can you add some features when finetuning the whole diffusion model, not only unet?
I read the Dreambooth code and copy the some code such as ```train_text_encoder``` to ```diffusers/examples/text_to_image/train_text_to_image.py```, but it still doesn't work for Deepspeed or FSDP.
I am trying to fine-tune the text en... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-03-24T15:44:48Z | 2023-05-02T15:03:26Z | 2023-05-02T15:03:26Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | liuyugeng | 13,288,444 | MDQ6VXNlcjEzMjg4NDQ0 | User | false |
huggingface/diffusers | 1,640,009,308 | I_kwDOHa8MBc5hwI5c | 2,819 | https://github.com/huggingface/diffusers/issues/2819 | https://api.github.com/repos/huggingface/diffusers/issues/2819 | HTTPSConnectionPool(host='raw.githubusercontent.com', port=443): Max retries exceeded with url: /huggingface/diffusers/main/examples/community/lpw_stable_diffusion.py | ### Describe the bug
SSLError: HTTPSConnectionPool(host='raw.githubusercontent.com', port=443): Max
retries exceeded with url:
/huggingface/diffusers/main/examples/community/lpw_stable_diffusion.py (Ca... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-03-24T20:43:56Z | 2023-03-30T14:39:32Z | 2023-03-30T14:39:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Pirog17000 | 3,022,852 | MDQ6VXNlcjMwMjI4NTI= | User | false |
huggingface/diffusers | 1,356,430,087 | I_kwDOHa8MBc5Q2XsH | 282 | https://github.com/huggingface/diffusers/issues/282 | https://api.github.com/repos/huggingface/diffusers/issues/282 | inpainting doesn't work with CPU | ### Describe the bug
<img width="895" alt="Screen Shot 2022-08-30 at 3 59 50 PM" src="https://user-images.githubusercontent.com/17561003/187558163-fe0198c5-a09f-44b2-ba5c-182d5c66131e.png">
Using the inpainting example but trying to run it on the cpu instead of gpu (since i'm trying to push to hf space).
Hitting ... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2022-08-30T23:01:08Z | 2022-11-28T15:04:06Z | 2022-11-28T15:04:06Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 1,640,104,671 | I_kwDOHa8MBc5hwgLf | 2,820 | https://github.com/huggingface/diffusers/issues/2820 | https://api.github.com/repos/huggingface/diffusers/issues/2820 | Any chance someone is working on multi controlnet for img2img | Any chance someone is working on StableDiffusionControlNetImg2ImgPipeline to support multiple controlnet? | closed | completed | false | 5 | [] | [] | 2023-03-24T22:02:18Z | 2023-03-27T14:21:27Z | 2023-03-27T14:21:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | romanfurman6 | 22,365,403 | MDQ6VXNlcjIyMzY1NDAz | User | false |
huggingface/diffusers | 1,640,273,945 | I_kwDOHa8MBc5hxJgZ | 2,821 | https://github.com/huggingface/diffusers/issues/2821 | https://api.github.com/repos/huggingface/diffusers/issues/2821 | Dreambooth won't start: [00:51:51] WARNING The following values were not passed to | ### Describe the bug
Getting this whenever I run train_dreambooth.py on Colab but it won't get passed it, it seems to freeze up and just run and not do anything with this warning and the cursor keeps blinking, until you stop the cell, then it just stops immediately when you try to run it after that. It has always wo... | closed | completed | false | 22 | [
"bug",
"stale"
] | [] | 2023-03-25T00:56:31Z | 2023-06-28T15:03:35Z | 2023-06-28T15:03:35Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Leomn1234 | 127,356,389 | U_kgDOB5dN5Q | User | false |
huggingface/diffusers | 1,640,308,264 | I_kwDOHa8MBc5hxR4o | 2,822 | https://github.com/huggingface/diffusers/issues/2822 | https://api.github.com/repos/huggingface/diffusers/issues/2822 | Slow training converge of StableDiffusionControlNetPipeline than original repo | ### Describe the bug
Awsome works to train ControlNet with Diffuser from the [tutorial](https://github.com/huggingface/diffusers/tree/main/examples/controlnet)
I run the code and compare it with the original training code [here](https://github.com/lllyasviel/ControlNet/blob/main/docs/train.md).
It turns out to b... | closed | completed | false | 9 | [
"bug",
"stale"
] | [] | 2023-03-25T01:35:41Z | 2023-05-29T15:03:57Z | 2023-05-29T15:03:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jingyangcarl | 24,415,835 | MDQ6VXNlcjI0NDE1ODM1 | User | false |
huggingface/diffusers | 1,640,705,127 | I_kwDOHa8MBc5hyyxn | 2,829 | https://github.com/huggingface/diffusers/issues/2829 | https://api.github.com/repos/huggingface/diffusers/issues/2829 | convert_lora_safetensor_to_diffusers | this script `https://github.com/huggingface/diffusers/blob/main/scripts/convert_lora_safetensor_to_diffusers.py` integrates the lora into the Pipeline and then outputs said Pipeline.
Is there a method to turn the `.safetensors` Lora to a `.bin` file to be loaded in dynamically similar to the snippet below
```
from... | closed | completed | false | 17 | [] | [] | 2023-03-26T00:00:18Z | 2023-12-26T01:39:42Z | 2023-04-17T02:32:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NicholasKao1029 | 45,542,006 | MDQ6VXNlcjQ1NTQyMDA2 | User | false |
huggingface/diffusers | 1,356,666,812 | I_kwDOHa8MBc5Q3Re8 | 283 | https://github.com/huggingface/diffusers/issues/283 | https://api.github.com/repos/huggingface/diffusers/issues/283 | Missing easy way to set various renderers/samplers | ### Describe the bug
(https://www.reddit.com/r/StableDiffusion/comments/wwm2at/sampler_vs_steps_comparison_low_to_mid_step_counts/?utm_medium=android_app&utm_source=share)
Check here, now
```
Now I launch it like this:
import sys
import torch
from torch import autocast
from diffusers import StableDiffusio... | closed | completed | false | 3 | [
"stale"
] | [] | 2022-08-31T02:38:06Z | 2022-11-02T16:00:43Z | 2022-10-31T15:03:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | paciox | 6,546,735 | MDQ6VXNlcjY1NDY3MzU= | User | false |
huggingface/diffusers | 1,640,722,192 | I_kwDOHa8MBc5hy28Q | 2,830 | https://github.com/huggingface/diffusers/issues/2830 | https://api.github.com/repos/huggingface/diffusers/issues/2830 | Can i make image&text to video with modelscope video model? | ### Model/Pipeline/Scheduler description
### Make more video pipelines like img2video, video2video
### Open source status
- [ ] The model implementation is available
- [X] The model weights are available (Only relevant if addition is not a scheduler).
### Provide useful links for the implementation
_No response_ | closed | completed | false | 3 | [] | [] | 2023-03-26T01:17:35Z | 2023-03-29T06:00:08Z | 2023-03-29T05:59:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tonymacx86PRO | 42,518,347 | MDQ6VXNlcjQyNTE4MzQ3 | User | false |
huggingface/diffusers | 1,640,767,897 | I_kwDOHa8MBc5hzCGZ | 2,831 | https://github.com/huggingface/diffusers/issues/2831 | https://api.github.com/repos/huggingface/diffusers/issues/2831 | Stable Diffusion Unclip Broken | ### Describe the bug
StableUnCLIPPipeline doesn't work with default values. It's missing the prior models. I'm trying to update them here and even converted a checkpoint model, but can't see to get it working yet.
### Reproduction
!pip install git+https://github.com/huggingface/diffusers@main transformers accelerate... | closed | completed | false | 4 | [
"bug"
] | [] | 2023-03-26T04:36:54Z | 2023-03-30T02:45:52Z | 2023-03-30T02:45:52Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nbardy | 1,278,972 | MDQ6VXNlcjEyNzg5NzI= | User | false |
huggingface/diffusers | 1,641,250,070 | I_kwDOHa8MBc5h030W | 2,838 | https://github.com/huggingface/diffusers/issues/2838 | https://api.github.com/repos/huggingface/diffusers/issues/2838 | can you support controlnet_pipeline don't use control , and it looks like sample stable diffusion ? | **Is your feature request related to a problem? Please describe.**
I want to try a better way to assemble controlnet, when I try not to add control in the contronet pipeline, I will encounter an error
**Describe the solution you'd like**
I want the contronetpipeline to degenerate into a normal sd when the control is... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-03-27T02:22:13Z | 2023-05-04T15:03:43Z | 2023-05-04T15:03:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kelisiya | 23,169,766 | MDQ6VXNlcjIzMTY5NzY2 | User | false |
huggingface/diffusers | 1,356,713,383 | I_kwDOHa8MBc5Q3c2n | 284 | https://github.com/huggingface/diffusers/issues/284 | https://api.github.com/repos/huggingface/diffusers/issues/284 | ONNX compatibility issues | Thanks for the great library!
I'm looking into generating ONNX version of the stable diffusion models.
I had to make few changes to `unet_2d_condition.py` to get it to work. You can find some of the changes here.. https://github.com/harishanand95/diffusers/commit/8dd4e822f87e1b4259755a2181218797ceecc410
Let me ... | closed | completed | false | 6 | [] | [] | 2022-08-31T03:36:55Z | 2023-04-06T23:09:59Z | 2022-09-05T12:49:27Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | harishanand95 | 7,456,909 | MDQ6VXNlcjc0NTY5MDk= | User | false |
huggingface/diffusers | 1,641,759,298 | I_kwDOHa8MBc5h20JC | 2,840 | https://github.com/huggingface/diffusers/issues/2840 | https://api.github.com/repos/huggingface/diffusers/issues/2840 | [Examples] Discussions on using streaming datasets for ControlNet training in JAX | As we discussed how well a large dataset would play in conjunction with the ControlNet training script in JAX, I considered consolidating what I have been doing in this regard.
To mimic a mid-scale dataset for training, I created this dummy dataset: https://huggingface.co/datasets/sayakpaul/dummy-controlnet-100000-... | closed | completed | false | 2 | [] | [] | 2023-03-27T09:39:38Z | 2023-03-29T01:12:10Z | 2023-03-29T01:12:10Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,641,778,605 | I_kwDOHa8MBc5h242t | 2,841 | https://github.com/huggingface/diffusers/issues/2841 | https://api.github.com/repos/huggingface/diffusers/issues/2841 | Adding support for sampling temperature in schedulers similar to stable diffusion. | **Is your feature request related to a problem? Please describe.**
Scaling the gaussian noise depending on the temperature.
**Describe the solution you'd like**
As per the DDPM sampler implementation from https://nn.labml.ai/diffusion/stable_diffusion/sampler/ddpm.html#section-25. There can be support for adding a... | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-27T09:52:20Z | 2023-05-04T15:03:41Z | 2023-05-04T15:03:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shivammehta25 | 9,089,131 | MDQ6VXNlcjkwODkxMzE= | User | false |
huggingface/diffusers | 1,641,835,734 | I_kwDOHa8MBc5h3GzW | 2,843 | https://github.com/huggingface/diffusers/issues/2843 | https://api.github.com/repos/huggingface/diffusers/issues/2843 | Multi_GPU save checkpoint | ### Describe the bug
When saving checkpoint under multi_gpu mode in accelerate, ```type(model) == type(unet)``` is not correct because the type of model is the ```UNet2DConditionModel``` but the type of UNet is the ```torch.DistributedDataParallel```.
### Reproduction
It should be ```type(model) == type(unet.m... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-03-27T10:23:39Z | 2023-05-04T15:03:40Z | 2023-05-04T15:03:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | liuyugeng | 13,288,444 | MDQ6VXNlcjEzMjg4NDQ0 | User | false |
huggingface/diffusers | 1,642,114,943 | I_kwDOHa8MBc5h4K9_ | 2,844 | https://github.com/huggingface/diffusers/issues/2844 | https://api.github.com/repos/huggingface/diffusers/issues/2844 | Runtime error when trying to explicitly pass image embeddings to StableUnCLIPImg2ImgPipeline | ### Describe the bug
I'm trying to use StableUnCLIPImg2ImgPipeline with modified image embeddings. When trying to pass embeddings explicitly instead of image, the error is raised.
The fix is simple, just check `image_embeds is None` instead of casting `image_embeds` to bool. I'll create a pull request with the fi... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-03-27T13:50:17Z | 2023-03-27T18:04:02Z | 2023-03-27T18:04:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | unishift | 30,509,893 | MDQ6VXNlcjMwNTA5ODkz | User | false |
huggingface/diffusers | 1,642,631,388 | I_kwDOHa8MBc5h6JDc | 2,848 | https://github.com/huggingface/diffusers/issues/2848 | https://api.github.com/repos/huggingface/diffusers/issues/2848 | Can we modify the network architecture? | The inner code base seems to be quite complicated. I was wondering is there any easy way to modify part of the network architecture? For example, like in the Paint By Example, https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/paint_by_example/pipeline_paint_by_example.py, they add more channels... | closed | completed | false | 2 | [] | [] | 2023-03-27T18:48:33Z | 2023-03-29T05:06:47Z | 2023-03-29T05:06:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xingzhehe | 28,001,407 | MDQ6VXNlcjI4MDAxNDA3 | User | false |
huggingface/diffusers | 1,642,974,046 | I_kwDOHa8MBc5h7cte | 2,851 | https://github.com/huggingface/diffusers/issues/2851 | https://api.github.com/repos/huggingface/diffusers/issues/2851 | not able to use `push_to_hub` during tpu training | ### Describe the bug
Not able to use `---push_to_hub` option for TPU training
getting error
```
Waiting for the following commands to finish before shutting down: [[push command, status code: running, in progress. PID: 772274]].
03/27/2023 23:15:59 - ERROR - huggingface_hub.repository - Waiting for the followi... | closed | completed | false | 8 | [
"bug"
] | [] | 2023-03-27T23:33:42Z | 2023-03-28T16:19:45Z | 2023-03-28T16:19:45Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yiyixuxu | 12,631,849 | MDQ6VXNlcjEyNjMxODQ5 | User | false |
huggingface/diffusers | 1,642,997,012 | I_kwDOHa8MBc5h7iUU | 2,852 | https://github.com/huggingface/diffusers/issues/2852 | https://api.github.com/repos/huggingface/diffusers/issues/2852 | Extremely slow loading of model | Hi guys.
Cuda 12.0.
Memory: 16GB, Tesla T4. nproc output: 4
I am using local files only = True, I am pointing the from_pretrained to the /cache_folder/snapshot/234xxx88378
First request to initialize the models takes 40s for stable diffusion 1.5 I have seen people get around 3s, how is that even achievable?
... | closed | completed | false | 28 | [
"stale"
] | [] | 2023-03-28T00:04:06Z | 2024-04-24T17:24:36Z | 2023-06-19T15:03:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | laxuscullen | 49,696,073 | MDQ6VXNlcjQ5Njk2MDcz | User | false |
huggingface/diffusers | 1,643,037,047 | I_kwDOHa8MBc5h7sF3 | 2,854 | https://github.com/huggingface/diffusers/issues/2854 | https://api.github.com/repos/huggingface/diffusers/issues/2854 | Contributing Doc Missing PR Checklist | In the [`CONTRIBUTING.md` doc](https://github.com/huggingface/diffusers/blob/main/CONTRIBUTING.md), in the ["How to open a PR" section](https://github.com/huggingface/diffusers/blob/main/CONTRIBUTING.md#how-to-open-a-pr), number 6, it says:
> Once you are satisfied (**and the checklist below is happy too**), go to t... | closed | completed | false | 1 | [] | [] | 2023-03-28T00:59:31Z | 2023-03-28T16:32:21Z | 2023-03-28T16:32:21Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dg845 | 58,458,699 | MDQ6VXNlcjU4NDU4Njk5 | User | false |
huggingface/diffusers | 1,643,112,827 | I_kwDOHa8MBc5h7-l7 | 2,855 | https://github.com/huggingface/diffusers/issues/2855 | https://api.github.com/repos/huggingface/diffusers/issues/2855 | [ControlNet] Support multi-batch multi-controlnet | Currently, we don't support passing multiple batches for multiple ControlNets. Elaborate discussions on this are [here](https://github.com/huggingface/diffusers/issues/2704).
We recently merged https://github.com/huggingface/diffusers/pull/2804 in order to make out users better aware of what is currently supported ... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-03-28T02:53:45Z | 2024-05-30T23:26:28Z | 2023-05-14T15:03:15Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,643,217,282 | I_kwDOHa8MBc5h8YGC | 2,856 | https://github.com/huggingface/diffusers/issues/2856 | https://api.github.com/repos/huggingface/diffusers/issues/2856 | Can't convert a sd model to diffusers | When I use the script to convert a sd model to diffusers:
`python diffusers/scripts/convert_original_stable_diffusion_to_diffusers.py --checkpoint_path ./AOM3_orangemixs.safetensors --extract_ema --dump_path AbyssOrangeMix --from_safetensors`
I got this error:
```
A matching Triton is not available, some optimiza... | closed | completed | false | 1 | [] | [] | 2023-03-28T05:17:44Z | 2023-03-28T05:57:54Z | 2023-03-28T05:57:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Amazingldl | 54,729,177 | MDQ6VXNlcjU0NzI5MTc3 | User | false |
huggingface/diffusers | 1,643,324,432 | I_kwDOHa8MBc5h8yQQ | 2,857 | https://github.com/huggingface/diffusers/issues/2857 | https://api.github.com/repos/huggingface/diffusers/issues/2857 | Any plan to support UniDiffuser? | Any plan to support UniDiffuser? https://github.com/thu-ml/unidiffuser
versatile diffusion(VD) maybe won't satisfy for multimodal
| closed | completed | false | 6 | [
"New pipeline/model"
] | [] | 2023-03-28T07:00:04Z | 2024-11-21T15:38:17Z | 2024-11-21T15:38:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nemonameless | 13,104,100 | MDQ6VXNlcjEzMTA0MTAw | User | false |
huggingface/diffusers | 1,643,468,183 | I_kwDOHa8MBc5h9VWX | 2,858 | https://github.com/huggingface/diffusers/issues/2858 | https://api.github.com/repos/huggingface/diffusers/issues/2858 | Adding seeding image to stable diffusion unclip img2img pipeline | **Is your feature request related to a problem? Please describe.**
Currently, the stable diffusion unclip img2img pipeline does not provide option to seed the latents by encoding with VAE.
**Describe the solution you'd like**
Allow another argument which allows seeding the latents with images encoded with VAE:
ht... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-03-28T08:37:19Z | 2023-05-05T15:03:46Z | 2023-05-05T15:03:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aravind-h-v | 122,849,931 | U_kgDOB1KKiw | User | false |
huggingface/diffusers | 1,643,478,210 | I_kwDOHa8MBc5h9XzC | 2,860 | https://github.com/huggingface/diffusers/issues/2860 | https://api.github.com/repos/huggingface/diffusers/issues/2860 | Use HTTP-based `upload_folder` in training scripts | While investigating https://github.com/huggingface/diffusers/issues/2851 (getting error message when using `push_to_hub` in a training), I realized that most of the training scripts rely on `Repository` to push training data to the Hub at the end of the training ([`train_text_to_image_flax`](https://github.com/huggingf... | closed | completed | false | 11 | [
"stale"
] | [] | 2023-03-28T08:43:57Z | 2023-05-05T15:03:45Z | 2023-05-05T15:03:45Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Wauplin | 11,801,849 | MDQ6VXNlcjExODAxODQ5 | User | false |
huggingface/diffusers | 1,643,894,897 | I_kwDOHa8MBc5h-9hx | 2,866 | https://github.com/huggingface/diffusers/issues/2866 | https://api.github.com/repos/huggingface/diffusers/issues/2866 | [LoRA] Better conversion of LoRA weights from `.bin` to `.safetensors` | Context: https://github.com/huggingface/diffusers/issues/2829#issuecomment-1485334706.
Currently, we have a [script](https://github.com/huggingface/diffusers/blob/1d7b4b60b7b7e19a0347da5a04ec76c045d8dbf0/scripts/convert_lora_safetensor_to_diffusers.py#L125) that allows us to convert LoRA weights from PyTorch to Safe... | closed | completed | false | 6 | [
"stale"
] | [
"sayakpaul"
] | 2023-03-28T12:56:30Z | 2023-05-17T16:47:14Z | 2023-05-07T15:02:56Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,643,960,819 | I_kwDOHa8MBc5h_Nnz | 2,867 | https://github.com/huggingface/diffusers/issues/2867 | https://api.github.com/repos/huggingface/diffusers/issues/2867 | multiple training pairs of diffusion models? | I went through the documentation abaouot training diffusion models. There are now many type of diffusion models and different modalities. In a general cases, a training data would be consisted with image and prompt or prompt and image. Now, for model like intruction-pixel 2 pixel and controlnet, what would be the train... | closed | completed | false | 6 | [] | [] | 2023-03-28T13:31:39Z | 2023-03-31T18:56:25Z | 2023-03-31T18:56:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pure-rgb | 45,315,076 | MDQ6VXNlcjQ1MzE1MDc2 | User | false |
huggingface/diffusers | 1,643,965,411 | I_kwDOHa8MBc5h_Ovj | 2,868 | https://github.com/huggingface/diffusers/issues/2868 | https://api.github.com/repos/huggingface/diffusers/issues/2868 | ControlNet + OffsetNoise + LoRA | **Is your feature request related to a problem? Please describe.**
https://www.reddit.com/r/StableDiffusion/comments/11fsdtq/controlnet_offsetnoise_lora_stable_diffusion_30/
Diffusers doesn't support LoRA with control-net yet. And OffsetNoise anywhere.
[Illustration](https://preview.redd.it/bgxgf60ddhla1.png?wid... | closed | completed | false | 5 | [] | [] | 2023-03-28T13:34:09Z | 2023-03-31T18:56:47Z | 2023-03-31T18:56:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pure-rgb | 45,315,076 | MDQ6VXNlcjQ1MzE1MDc2 | User | false |
huggingface/diffusers | 1,357,017,577 | I_kwDOHa8MBc5Q4nHp | 287 | https://github.com/huggingface/diffusers/issues/287 | https://api.github.com/repos/huggingface/diffusers/issues/287 | [Community effort] Type Hints | **Is your feature request related to a problem? Please describe.**
Many important functions and classes are missing type hints
**Describe the solution you'd like**
It would be great if the following functions could have better type hints to understand what input is expected.
## Models
1. - [x] https://gi... | closed | completed | false | 21 | [
"good first issue"
] | [] | 2022-08-31T08:52:08Z | 2022-09-08T07:20:59Z | 2022-09-08T07:20:58Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,644,041,173 | I_kwDOHa8MBc5h_hPV | 2,871 | https://github.com/huggingface/diffusers/issues/2871 | https://api.github.com/repos/huggingface/diffusers/issues/2871 | Output type "latent" not correctly handled for the Stable Diffusion Inpainting pipeline | ### Describe the bug
The behavior of the option `output_type = "latent"` is not the same for the inpainting and the txt2image pipeline.
**Inpainting pipeline:**
If `output_type = "latent"`, the value returned is the decoded image (**`latents`** should be returned in this case):
[link to file](https://github.com... | closed | completed | false | 7 | [
"bug"
] | [
"yiyixuxu"
] | 2023-03-28T14:07:02Z | 2023-04-04T18:52:59Z | 2023-04-04T18:52:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | abdellah-lamrani-alaoui | 19,689,626 | MDQ6VXNlcjE5Njg5NjI2 | User | false |
huggingface/diffusers | 1,644,472,181 | I_kwDOHa8MBc5iBKd1 | 2,875 | https://github.com/huggingface/diffusers/issues/2875 | https://api.github.com/repos/huggingface/diffusers/issues/2875 | Text-to-image training loss becomes nan all of a sudden | ### Describe the bug
Hello,
I am trying to finetune the stable diffusion 2.1 model on a custom dataset. I am using the example script provided [here](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py). The training starts alright, the loss is decreasing, but then rando... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-03-28T18:33:11Z | 2023-03-31T21:45:18Z | 2023-03-31T21:45:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Hsn37 | 18,398,637 | MDQ6VXNlcjE4Mzk4NjM3 | User | false |
huggingface/diffusers | 1,644,499,013 | I_kwDOHa8MBc5iBRBF | 2,876 | https://github.com/huggingface/diffusers/issues/2876 | https://api.github.com/repos/huggingface/diffusers/issues/2876 | Support private hub | I would like to able to save to a private hub after making use of the `train_dreambooth` script | closed | not_planned | false | 1 | [] | [] | 2023-03-28T18:53:13Z | 2023-03-30T16:47:54Z | 2023-03-28T18:56:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chidindu-ogbonna | 25,098,427 | MDQ6VXNlcjI1MDk4NDI3 | User | false |
huggingface/diffusers | 1,644,552,541 | I_kwDOHa8MBc5iBeFd | 2,878 | https://github.com/huggingface/diffusers/issues/2878 | https://api.github.com/repos/huggingface/diffusers/issues/2878 | PyTorch 2.0 and ONNX converter | ```cmd
UnsupportedOperatorError: Exporting the operator 'aten::scaled_dot_product_attention' to ONNX opset version 14 is not supported.
```
More info: https://github.com/pytorch/pytorch/issues/97262 | closed | completed | false | 13 | [
"stale"
] | [] | 2023-03-28T19:37:41Z | 2023-06-05T14:01:42Z | 2023-05-31T15:03:47Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ForserX | 13,867,290 | MDQ6VXNlcjEzODY3Mjkw | User | false |
huggingface/diffusers | 1,644,642,557 | I_kwDOHa8MBc5iB0D9 | 2,880 | https://github.com/huggingface/diffusers/issues/2880 | https://api.github.com/repos/huggingface/diffusers/issues/2880 | stable diffusion jax colab doesn't run | ### Describe the bug
I'm using the colab stable_diffusion_jax_how_to.ipynb and trying to run stable diffusion in Jax on tpus.
I get an error on line import jax.tools.colab_tpu and the error is "AttributeError: partially initialized module 'jax' has no attribute 'version' (most likely due to a circular import)"
I... | closed | completed | false | 3 | [
"bug"
] | [
"yiyixuxu"
] | 2023-03-28T20:53:58Z | 2023-04-03T18:32:02Z | 2023-04-03T18:32:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shlokk | 5,174,718 | MDQ6VXNlcjUxNzQ3MTg= | User | false |
huggingface/diffusers | 1,644,709,087 | I_kwDOHa8MBc5iCETf | 2,881 | https://github.com/huggingface/diffusers/issues/2881 | https://api.github.com/repos/huggingface/diffusers/issues/2881 | Using accelerate to use multi gpu inference | Hi,
I would like to evaluate my model by generating quite a few images and comparing them. Therefore I want to use 2 or more GPUs to to generate lets say 1000 images.
For training i use accelerate which works really well but for inference I cant really get accelerate to work properly. Is there any way to use already ... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-03-28T21:57:34Z | 2023-05-07T15:02:53Z | 2023-05-07T15:02:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | flixmk | 66,331,106 | MDQ6VXNlcjY2MzMxMTA2 | User | false |
huggingface/diffusers | 1,645,243,985 | I_kwDOHa8MBc5iEG5R | 2,883 | https://github.com/huggingface/diffusers/issues/2883 | https://api.github.com/repos/huggingface/diffusers/issues/2883 | Image to Video Pipeline | ### Model/Pipeline/Scheduler description
How will this new pipeline work
You give a image and using the denoising strength adds noise to first frame and adding it to temporal memory
Output: Making from first image makes a sequence frame that represents a animation or video from the first frame
This very cool addi... | closed | completed | false | 2 | [
"New pipeline/model"
] | [] | 2023-03-29T08:03:11Z | 2024-11-21T07:25:02Z | 2024-11-21T07:25:02Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tonymacx86PRO | 42,518,347 | MDQ6VXNlcjQyNTE4MzQ3 | User | false |
huggingface/diffusers | 1,645,266,173 | I_kwDOHa8MBc5iEMT9 | 2,884 | https://github.com/huggingface/diffusers/issues/2884 | https://api.github.com/repos/huggingface/diffusers/issues/2884 | How to implement `cross_attention_kwargs` in `StableDiffusionImg2ImgPipeline`? | How to implement `cross_attention_kwargs` in `StableDiffusionImg2ImgPipeline`?
I want to use lora weight with parameter `cross_attention_kwargs` in Img2img pipeline, but the `cross_attention_kwargs` has not implemented in `StableDiffusionImg2ImgPipeline`. Is there another way to get the same effect?
The error m... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-03-29T08:17:32Z | 2023-05-07T15:02:51Z | 2023-05-07T15:02:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | youyuge34 | 15,784,256 | MDQ6VXNlcjE1Nzg0MjU2 | User | false |
huggingface/diffusers | 1,645,407,954 | I_kwDOHa8MBc5iEu7S | 2,885 | https://github.com/huggingface/diffusers/issues/2885 | https://api.github.com/repos/huggingface/diffusers/issues/2885 | Issues loading LoRA weights | Reported here by https://discuss.huggingface.co/u/dkackman: https://discuss.huggingface.co/t/does-load-attn-procs-work-with-safetensors/34827. See also proposed patch by user https://discuss.huggingface.co/u/slavakurilyak. | closed | completed | false | 4 | [
"stale"
] | [] | 2023-03-29T09:37:55Z | 2023-05-07T15:02:51Z | 2023-05-07T15:02:51Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,645,423,192 | I_kwDOHa8MBc5iEypY | 2,886 | https://github.com/huggingface/diffusers/issues/2886 | https://api.github.com/repos/huggingface/diffusers/issues/2886 | Download file to specific destination | Recently `huggingface_hub. snapshot_download ` added two new arguments: `local_dir` and `local_dir_use_symlinks`. Here's the merged PR: https://github.com/huggingface/huggingface_hub/pull/1360
This is very convenient if you want to use a local folder for storing models instead of the cache folder.
It would be gre... | closed | completed | false | 15 | [
"stale"
] | [] | 2023-03-29T09:46:04Z | 2023-05-07T15:02:50Z | 2023-05-07T15:02:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | peterschmidt85 | 54,148,038 | MDQ6VXNlcjU0MTQ4MDM4 | User | false |
huggingface/diffusers | 1,645,664,795 | I_kwDOHa8MBc5iFtob | 2,888 | https://github.com/huggingface/diffusers/issues/2888 | https://api.github.com/repos/huggingface/diffusers/issues/2888 | Community Sprint: Control Stable Diffusion with JAX & diffusers with (v4) TPUs ⚡️ | Hey folks! We are partnering with Google Cloud to organize a community event where we will be training models like ControlNet with JAX using free v4 TPUs provided by Google 🎁
You can attend with teams, brainstorm on ideas together or join another team to work on an idea. In the end, there will be super cool prizes t... | closed | completed | false | 1 | [
"stale"
] | [] | 2023-03-29T12:03:26Z | 2023-05-07T15:02:49Z | 2023-05-07T15:02:49Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | merveenoyan | 53,175,384 | MDQ6VXNlcjUzMTc1Mzg0 | User | false |
huggingface/diffusers | 1,645,738,423 | I_kwDOHa8MBc5iF_m3 | 2,890 | https://github.com/huggingface/diffusers/issues/2890 | https://api.github.com/repos/huggingface/diffusers/issues/2890 | Tweaking and training a diffusers architecture from scratch | Hi,
Training a diffusion model from scratch may be desirable especially if parts of the architecture are meant to be tweaked.
Although there is (training)[https://huggingface.co/docs/diffusers/training/text2image] it is not clear:
1. How one can do multi-gpu/multi-node training
2. Whether and how you can modif... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-03-29T12:45:46Z | 2023-05-07T15:02:48Z | 2023-05-07T15:02:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ysig | 28,439,529 | MDQ6VXNlcjI4NDM5NTI5 | User | false |
huggingface/diffusers | 1,645,945,894 | I_kwDOHa8MBc5iGyQm | 2,891 | https://github.com/huggingface/diffusers/issues/2891 | https://api.github.com/repos/huggingface/diffusers/issues/2891 | StableDiffusionUpscalePipeline does not support embeds, but docs say it does | ### Describe the bug
Running the example code with compel gives
```python
ValueError: `prompt` has to be of type `str` or `list` but is <class 'NoneType'>
```
at `diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_upscale.py:321 in check_inputs`
It appears `check_inputs` is ignoring the `negative... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-03-29T14:32:50Z | 2023-03-31T11:46:32Z | 2023-03-31T11:46:21Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | d1g1t | 661,828 | MDQ6VXNlcjY2MTgyOA== | User | false |
huggingface/diffusers | 1,646,639,454 | I_kwDOHa8MBc5iJble | 2,893 | https://github.com/huggingface/diffusers/issues/2893 | https://api.github.com/repos/huggingface/diffusers/issues/2893 | ControlNetPipeline mixes up width and height | ### Describe the bug
In the same vein as https://github.com/huggingface/diffusers/pull/2558
`StableDiffusionControlNetPipeline` gets width and height wrong in `_default_height_width` (https://github.com/huggingface/diffusers/blob/main/src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_controlnet.py#... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-03-29T22:44:52Z | 2023-03-31T13:59:52Z | 2023-03-31T13:59:52Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | antoche | 1,627,384 | MDQ6VXNlcjE2MjczODQ= | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.