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 | 1,862,342,145 | I_kwDOHa8MBc5vARYB | 4,726 | https://github.com/huggingface/diffusers/issues/4726 | https://api.github.com/repos/huggingface/diffusers/issues/4726 | StableDiffusionPipeline has incorrect type for `callback` in __call__ | ### Describe the bug
Documentation (code) for `__call__` in StableDiffusionPipeline says:
```
callback: Optional[Callable[[int, int, torch.FloatTensor], None]] = None,
callback (`Callable`, *optional*):
A function that calls every `callback_steps` steps during inference. The function is called wi... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-08-22T23:43:56Z | 2023-11-09T15:06:11Z | 2023-11-09T15:06:11Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bartekleon | 28,489,774 | MDQ6VXNlcjI4NDg5Nzc0 | User | false |
huggingface/diffusers | 1,862,378,356 | I_kwDOHa8MBc5vAaN0 | 4,727 | https://github.com/huggingface/diffusers/issues/4727 | https://api.github.com/repos/huggingface/diffusers/issues/4727 | Auto Pipelines May Break When Using local_files_only | ### Describe the bug
Auto pipelines fetch a model's config file to determine what class to use. They only pass `pretrained_model_or_path` to `load_config()`, leaving meaningful arguments like `local_files_only`, `cache_dir`, `use_auth_token`, and others to only be passed to the later `from_pretrained()` call.
The... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-08-23T00:39:42Z | 2023-08-28T17:42:17Z | 2023-08-28T17:42:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | NullSenseStudio | 47,096,043 | MDQ6VXNlcjQ3MDk2MDQz | User | false |
huggingface/diffusers | 1,862,430,660 | I_kwDOHa8MBc5vAm_E | 4,728 | https://github.com/huggingface/diffusers/issues/4728 | https://api.github.com/repos/huggingface/diffusers/issues/4728 | I am trying to fine-tune the unet model using text_to_image.py. However, when I try to resume training using --resume_from_checkpoint, I encounter an "out of memory" error with the GPU. What could be causing this issue? | ### Describe the bug
I am trying to fine-tune the unet model using text_to_image.py. However, when I try to resume training using --resume_from_checkpoint in 4 V100, I encounter an "out of memory" error with the GPU. What could be causing this issue?
### Reproduction
accelerate launch --mixed_precision="fp16" ... | closed | completed | false | 3 | [
"bug",
"stale"
] | [] | 2023-08-23T02:02:48Z | 2023-11-08T15:09:24Z | 2023-11-08T15:09:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | cjt222 | 17,508,662 | MDQ6VXNlcjE3NTA4NjYy | User | false |
huggingface/diffusers | 1,862,511,539 | I_kwDOHa8MBc5vA6uz | 4,730 | https://github.com/huggingface/diffusers/issues/4730 | https://api.github.com/repos/huggingface/diffusers/issues/4730 | data_collator can't work in Trainer with LoRA Model | ### Describe the bug
I defined a data_collator and used it in the Transformers Trainer and it worked. Due to de "CUDA out of memory", I choose LoRA to adapt my model. But then `features` in data_collator has nothing. I don't know why.
### Reproduction
# this defines data_collator
@dataclass
class DataCollatorForM... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-08-23T03:39:33Z | 2023-08-23T04:14:28Z | 2023-08-23T04:14:17Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | z379035389 | 48,674,444 | MDQ6VXNlcjQ4Njc0NDQ0 | User | false |
huggingface/diffusers | 1,862,584,322 | I_kwDOHa8MBc5vBMgC | 4,731 | https://github.com/huggingface/diffusers/issues/4731 | https://api.github.com/repos/huggingface/diffusers/issues/4731 | torch.compile failed for multi-controlnet of sdxl | ### Describe the bug
torch.compile failed for multi-controlnet of sdxl
### Reproduction
'''
print("Run torch compile")
torch._dynamo.config.verbose = True
pipe.unet = torch.compile(pipe.unet, mode="reduce-overhead", fullgraph=True)
pipe.controlnet = torch.compile(pipe.controlnet, mode="reduce-overhead",... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-08-23T05:17:46Z | 2023-08-27T01:09:01Z | 2023-08-27T01:09:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YongtaoGE | 22,744,013 | MDQ6VXNlcjIyNzQ0MDEz | User | false |
huggingface/diffusers | 1,862,689,119 | I_kwDOHa8MBc5vBmFf | 4,732 | https://github.com/huggingface/diffusers/issues/4732 | https://api.github.com/repos/huggingface/diffusers/issues/4732 | My prompt is more important in sd-webui-controlnet and diffusers | in [sd-webui-controlnet](https://github.com/Mikubill/sd-webui-controlnet) we can choose My prompt is more important.
in diffusers how can we get the same operation？
I checked the official website documentation, it... | closed | completed | false | 5 | [
"stale"
] | [] | 2023-08-23T06:56:00Z | 2023-11-08T15:09:22Z | 2023-11-08T15:09:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Laidawang | 85,244,566 | MDQ6VXNlcjg1MjQ0NTY2 | User | false |
huggingface/diffusers | 1,862,730,186 | I_kwDOHa8MBc5vBwHK | 4,733 | https://github.com/huggingface/diffusers/issues/4733 | https://api.github.com/repos/huggingface/diffusers/issues/4733 | train_dreambooth_lora.py can not resume from checkpoint," No inf checks were recorded for this optimizer." | ### Describe the bug
i use diffusers==0.21.0 ,accelerate==0.22.0
when i running "accelerate launch train_dreambooth_lora.py --pretrained_model_name_or_path="/home/gxh/.cache/huggingface/hub/models--runwayml--stable-diffusion-v1-5/snapshots/aa9ba505e1973ae5cd05f5aedd345178f52f8e6a" --instance_data_dir="/home/gx... | closed | completed | false | 10 | [
"bug"
] | [] | 2023-08-23T07:24:56Z | 2023-09-01T06:59:12Z | 2023-08-24T03:02:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hollow1123 | 142,968,865 | U_kgDOCIWIIQ | User | false |
huggingface/diffusers | 1,862,961,516 | I_kwDOHa8MBc5vCols | 4,734 | https://github.com/huggingface/diffusers/issues/4734 | https://api.github.com/repos/huggingface/diffusers/issues/4734 | [Core] shareable generation configs for pipelines | `transformers` has got a component called "generation config": https://huggingface.co/docs/transformers/main_classes/text_generation#transformers.GenerationConfig. Here's how it's used:
```python
from transformers import AutoModelForSeq2SeqLM, AutoTokenizer, GenerationConfig
tokenizer = AutoTokenizer.from_pretr... | closed | completed | false | 14 | [
"stale"
] | [] | 2023-08-23T09:41:39Z | 2023-11-08T15:09:20Z | 2023-11-08T15:09:20Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 1,863,018,605 | I_kwDOHa8MBc5vC2ht | 4,735 | https://github.com/huggingface/diffusers/issues/4735 | https://api.github.com/repos/huggingface/diffusers/issues/4735 | Loading LoRA with 12 hidden layer TextEncoder and use clip_skip=2 | While attempting to load the LoRA model, I've encountered an issue when using the parameter clip_skip=2.
The problem arises due to the textencoder component within the LoRA model containing parameters for 12 layers. If I load
pipline with config[num_hidden_layers=11], an error occurs when using load_lora_weights()... | closed | completed | false | 2 | [] | [] | 2023-08-23T10:11:40Z | 2023-08-24T17:36:13Z | 2023-08-23T14:50:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | EnzoWuu | 45,420,316 | MDQ6VXNlcjQ1NDIwMzE2 | User | false |
huggingface/diffusers | 1,863,050,505 | I_kwDOHa8MBc5vC-UJ | 4,736 | https://github.com/huggingface/diffusers/issues/4736 | https://api.github.com/repos/huggingface/diffusers/issues/4736 | CUDA out of memory and invalid value encountered in cast with train_text_to_image_lora_sdxl.py | ### Describe the bug
I encountered two distinct issues while attempting to run the `lambdalabs/pokemon-blip-captions` example of train_text_to_image_lora_sdxl.py on an RTX 4090, utilizing bf16.
### Problem 1: RuntimeWarning and Image Processing:
```
RuntimeWarning: invalid value encountered in cast
images = (i... | closed | completed | false | 20 | [
"bug"
] | [] | 2023-08-23T10:31:44Z | 2023-08-28T08:27:42Z | 2023-08-28T08:27:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mnslarcher | 24,874,992 | MDQ6VXNlcjI0ODc0OTky | User | false |
huggingface/diffusers | 1,369,512,032 | I_kwDOHa8MBc5RoRhg | 474 | https://github.com/huggingface/diffusers/issues/474 | https://api.github.com/repos/huggingface/diffusers/issues/474 | Can I adapt existing UNet Blocks? | **What API design would you like to have changed or added to the library? Why?**
In diffusers, we add Downsample2D (for the non-final block), [see here](https://github.com/huggingface/diffusers/blob/8eaaa546d89f836b716e92348786d878f883ee86/src/diffusers/models/unet_2d.py#L114). This could be a conv layer or avg. po... | closed | completed | false | 7 | [
"stale"
] | [] | 2022-09-12T08:52:07Z | 2022-10-12T17:50:48Z | 2022-10-12T17:50:48Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,863,591,418 | I_kwDOHa8MBc5vFCX6 | 4,742 | https://github.com/huggingface/diffusers/issues/4742 | https://api.github.com/repos/huggingface/diffusers/issues/4742 | Implementing inference from Custom Diffusion Checkpoints | The section detailing inference steps from Custom Diffusion Checkpoints has been left as TODO in the official documentation. Is there an official timeline for this? Would it be possible to get a simple checkpoint inference script soon? | closed | completed | false | 6 | [
"stale"
] | [] | 2023-08-23T15:45:20Z | 2024-05-17T13:11:02Z | 2023-11-08T15:09:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aashish2000 | 30,340,846 | MDQ6VXNlcjMwMzQwODQ2 | User | false |
huggingface/diffusers | 1,864,240,174 | I_kwDOHa8MBc5vHgwu | 4,747 | https://github.com/huggingface/diffusers/issues/4747 | https://api.github.com/repos/huggingface/diffusers/issues/4747 | OSError: Error no file named model_index.json found in directory after load fine tuned SDXL LoRA | ### Describe the bug
I follow the SDXL fine tuning LoRA from [here](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/README_sdxl.md)
And python raises error
```bash
Traceback (most recent call last):
File "/Users/sarit/anaconda3/envs/try_openai/lib/python3.11/site-packages/gradio/route... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-08-24T01:49:04Z | 2023-08-24T02:10:58Z | 2023-08-24T02:05:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elcolie | 18,206,728 | MDQ6VXNlcjE4MjA2NzI4 | User | false |
huggingface/diffusers | 1,369,638,154 | I_kwDOHa8MBc5RowUK | 475 | https://github.com/huggingface/diffusers/issues/475 | https://api.github.com/repos/huggingface/diffusers/issues/475 | JAX Integration | ## JAX Integration
This issue will be used as a tracker to integrate Stable Diffusion in JAX natively to `diffusers`. This will enable many cool use cases noteably running stable diffusion on a google colab.
### General design:
We will make loosen the forced PyTorch dependency and instead force the user to eit... | closed | completed | false | 10 | [] | [] | 2022-09-12T10:22:55Z | 2022-10-27T09:12:24Z | 2022-10-27T09:12:23Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,864,351,827 | I_kwDOHa8MBc5vH8BT | 4,750 | https://github.com/huggingface/diffusers/issues/4750 | https://api.github.com/repos/huggingface/diffusers/issues/4750 | Inpaint + Lama Preprocessor | ### Model/Pipeline/Scheduler description
Automatic1111 supports preprocessor of Inpaint+Lama for the Inpaint ControlNet model over web ui.
Is there a way to add the same functionality for Diffusers pipeline?
I wish to perform outpainting like how it's mentioned here (https://github.com/Mikubill/sd-webui-controln... | closed | completed | false | 11 | [
"stale"
] | [] | 2023-08-24T03:48:19Z | 2023-11-20T15:06:44Z | 2023-11-20T15:06:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 0sparsh2 | 50,074,241 | MDQ6VXNlcjUwMDc0MjQx | User | false |
huggingface/diffusers | 1,864,443,312 | I_kwDOHa8MBc5vISWw | 4,751 | https://github.com/huggingface/diffusers/issues/4751 | https://api.github.com/repos/huggingface/diffusers/issues/4751 | LoRA Still Influencing Output Despite Setting "scale" to 0 | Hi, thanks and great job for implementing Kohya LoRA format into diffusers. However, I'm running into an issue where the LoRA settings don't seem to be working as described in the documentation.
According to the [official documentation](https://huggingface.co/docs/diffusers/training/lora#dreambooth-inference), setti... | closed | completed | false | 8 | [] | [
"sayakpaul"
] | 2023-08-24T05:41:35Z | 2023-09-05T05:30:34Z | 2023-09-04T21:52:32Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Linaqruf | 50,163,983 | MDQ6VXNlcjUwMTYzOTgz | User | false |
huggingface/diffusers | 1,864,469,483 | I_kwDOHa8MBc5vIYvr | 4,752 | https://github.com/huggingface/diffusers/issues/4752 | https://api.github.com/repos/huggingface/diffusers/issues/4752 | Input type (c10::Half) and bias type (float) mismatch in pipeline_stable_diffusion_xl.py | ### Describe the bug
I'm using the official example and trying to call StableDiffusionXLPipeline (fp16) twice. But in the second call, an error message appeared:
> RuntimeError: Input type (c10::Half) and bias type (float) should be the same.
There are similar issues like #4619 and #4478 but don't seem to apply in... | closed | completed | false | 12 | [
"bug"
] | [] | 2023-08-24T06:08:25Z | 2023-08-25T13:58:49Z | 2023-08-25T03:46:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Yukun-Huang | 16,999,639 | MDQ6VXNlcjE2OTk5NjM5 | User | false |
huggingface/diffusers | 1,864,481,055 | I_kwDOHa8MBc5vIbkf | 4,753 | https://github.com/huggingface/diffusers/issues/4753 | https://api.github.com/repos/huggingface/diffusers/issues/4753 | Something wrong with train_dreambooth_lora.py | ### Describe the bug
I think it should be
```python
if args.class_prompt is not None:
pre_computed_class_prompt_encoder_hidden_states = compute_text_embeddings(args.class_prompt)
else:
pre_computed_class_prompt_encoder_hidden_states = None
```
rather than
https://gith... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-08-24T06:19:22Z | 2023-08-25T18:53:59Z | 2023-08-25T18:53:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hollow1123 | 142,968,865 | U_kgDOCIWIIQ | User | false |
huggingface/diffusers | 1,864,629,523 | I_kwDOHa8MBc5vI_0T | 4,754 | https://github.com/huggingface/diffusers/issues/4754 | https://api.github.com/repos/huggingface/diffusers/issues/4754 | OSError: Error no file named model_index.json found in directory after load fine tuned textual inversion | I train the textual inversion fine tuning cat toy example from [here](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion)
And python raises error when inference
`Traceback (most recent call last):
File "infer.py", line 6, in <module>
pipe = StableDiffusionPipeline.from_pretrain... | closed | completed | false | 8 | [] | [] | 2023-08-24T08:04:34Z | 2024-12-04T04:28:24Z | 2023-08-29T10:08:28Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Adorablepet | 29,625,016 | MDQ6VXNlcjI5NjI1MDE2 | User | false |
huggingface/diffusers | 1,864,678,311 | I_kwDOHa8MBc5vJLun | 4,756 | https://github.com/huggingface/diffusers/issues/4756 | https://api.github.com/repos/huggingface/diffusers/issues/4756 | Support for training style with examples/dreambooth/train_dreambooth_lora_sdxl.py | **Is your feature request related to a problem? Please describe.**
I'm trying to use _examples/dreambooth/train_dreambooth_lora_sdxl.py_ to train an original style dreambooth SDXL model. If I'm right, this script only support train object for now.
**Describe the solution you'd like**
Support for training style wit... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-08-24T08:35:53Z | 2023-11-08T15:09:12Z | 2023-11-08T15:09:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | OranHe | 22,882,062 | MDQ6VXNlcjIyODgyMDYy | User | false |
huggingface/diffusers | 1,864,978,068 | I_kwDOHa8MBc5vKU6U | 4,758 | https://github.com/huggingface/diffusers/issues/4758 | https://api.github.com/repos/huggingface/diffusers/issues/4758 | [Docs] Extensive / Improved guides about "Text2Image", "Image2Image", "Inpainting" | We very often run into issues like the following https://github.com/huggingface/diffusers/issues/4392 where people don't know how to correctly use `diffusers` or are unaware of all the existing features.
I propose to write three very extensive guides or possible even write whole subsections about the main important ... | closed | completed | false | 12 | [
"stale"
] | [] | 2023-08-24T11:41:54Z | 2023-10-18T16:09:18Z | 2023-10-18T16:09:18Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,369,678,346 | I_kwDOHa8MBc5Ro6IK | 476 | https://github.com/huggingface/diffusers/issues/476 | https://api.github.com/repos/huggingface/diffusers/issues/476 | What are the original NSFW concepts used in the safety checker? | It looks like the safety checker uses some NSFW concept embeddings generated from CLIP to filter out unsafe content.
I wonder if we can get the original concepts in text instead of in CLIP embeddings?
Because I want to know if those concepts already cover what I want to filter, if not, I can generate my extra NSFW ... | closed | completed | false | 9 | [
"stale"
] | [] | 2022-09-12T10:54:10Z | 2022-11-13T15:03:10Z | 2022-11-13T15:03:10Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Nash2325138 | 17,662,095 | MDQ6VXNlcjE3NjYyMDk1 | User | false |
huggingface/diffusers | 1,865,345,373 | I_kwDOHa8MBc5vLuld | 4,764 | https://github.com/huggingface/diffusers/issues/4764 | https://api.github.com/repos/huggingface/diffusers/issues/4764 | can't load lora weight generated from: https://github.com/Zeju1997/oft | I am trying to load lora weights to try inference after finetuning with train_dreambooth_oft.py in https://github.com/Zeju1997/oft
I have tried both .load_lora_weights() and .unet_load_attn_procs(). both giving the same error:
Traceback (most recent call last):
File "/home/depecikbora/oft/oft-db/fullimg.py", lin... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-08-24T14:55:59Z | 2023-11-08T15:09:09Z | 2023-11-08T15:09:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | boradepecik | 72,090,381 | MDQ6VXNlcjcyMDkwMzgx | User | false |
huggingface/diffusers | 1,865,660,880 | I_kwDOHa8MBc5vM7nQ | 4,767 | https://github.com/huggingface/diffusers/issues/4767 | https://api.github.com/repos/huggingface/diffusers/issues/4767 | EulerAncestralDiscreteScheduler prediction_type `sample` support? | ### Describe the bug
Around line 246 in `scheduling_euler_ancestral_discrete`:
```python
if self.config.prediction_type == "epsilon":
pred_original_sample = sample - sigma * model_output
elif self.config.prediction_type == "v_prediction":
# * c_out + input * c_skip
... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-08-24T18:18:03Z | 2023-11-08T15:09:05Z | 2023-11-08T15:09:05Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | eliphatfs | 23,738,781 | MDQ6VXNlcjIzNzM4Nzgx | User | false |
huggingface/diffusers | 1,865,676,044 | I_kwDOHa8MBc5vM_UM | 4,768 | https://github.com/huggingface/diffusers/issues/4768 | https://api.github.com/repos/huggingface/diffusers/issues/4768 | Support saving multiple t2i adapter models under one checkpoint | Similarly to how we support serializing multiple controlnet checkpoints under one combined controlnet checkpoint. We should be able to support multiple t2i adapter checkpoints under one combined t2i adapter checkpoint
see some discussion here: https://github.com/huggingface/diffusers/pull/4621/files#r1295338783 | closed | completed | false | 1 | [
"good first issue",
"contributions-welcome"
] | [] | 2023-08-24T18:29:23Z | 2023-08-29T07:24:42Z | 2023-08-29T07:24:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | williamberman | 6,505,431 | MDQ6VXNlcjY1MDU0MzE= | User | false |
huggingface/diffusers | 1,369,726,430 | I_kwDOHa8MBc5RpF3e | 477 | https://github.com/huggingface/diffusers/issues/477 | https://api.github.com/repos/huggingface/diffusers/issues/477 | off-by-one error in pndm scheduler | There appears to be an off-by-one error on this line:
https://github.com/huggingface/diffusers/blob/25a51b63ca75e1351069bee87a0fb3df5abb89c3/src/diffusers/schedulers/scheduling_pndm.py#L338
I got `IndexError: index 1000 is out of bounds for dimension 0 with size 1000`
Repro script:
```import torch
from diffuse... | closed | completed | false | 2 | [] | [] | 2022-09-12T11:29:22Z | 2022-09-12T19:41:58Z | 2022-09-12T19:41:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rokbok | 113,351,946 | U_kgDOBsGdCg | User | false |
huggingface/diffusers | 1,866,122,582 | I_kwDOHa8MBc5vOsVW | 4,770 | https://github.com/huggingface/diffusers/issues/4770 | https://api.github.com/repos/huggingface/diffusers/issues/4770 | checkpoint_merger not support sdxl | examples/community/checkpoint_merger.py
It's a great tool, When Will SDXL be supported | closed | completed | false | 2 | [
"community-examples",
"stale",
"contributions-welcome"
] | [] | 2023-08-25T01:48:21Z | 2023-11-08T15:09:03Z | 2023-11-08T15:09:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yanchaoguo | 11,827,802 | MDQ6VXNlcjExODI3ODAy | User | false |
huggingface/diffusers | 1,866,198,739 | I_kwDOHa8MBc5vO-7T | 4,771 | https://github.com/huggingface/diffusers/issues/4771 | https://api.github.com/repos/huggingface/diffusers/issues/4771 | Loading a fine-tuned diffusion model | ### Describe the bug
Hello team,
Thank you for this amazing open-source work.
Recently, I fine-tuned a diffusion model using the [blog](https://huggingface.co/docs/diffusers/v0.11.0/en/training/text2image)
On obtaining the trained model in the `OUTPUT_DIR`, I am unable to load it using the code snippet pr... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-08-25T03:12:46Z | 2023-11-08T15:09:01Z | 2023-11-08T15:09:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sushantmenon1 | 74,258,021 | MDQ6VXNlcjc0MjU4MDIx | User | false |
huggingface/diffusers | 1,866,271,654 | I_kwDOHa8MBc5vPQum | 4,772 | https://github.com/huggingface/diffusers/issues/4772 | https://api.github.com/repos/huggingface/diffusers/issues/4772 | Memory issue on colab free for SDXL + Clip-vit-G + Xformers + accelerate | Reproduction code demo https://colab.research.google.com/drive/15CsxWNpALN-gnQJO4ad2pQy0CMotf58h
error message:
Your session crashed after using all available RAM. If you are interested in access to high-RAM runtimes, you may want to check out Colab Pro.
Is there a solution for this that would work within the ... | closed | completed | false | 2 | [] | [] | 2023-08-25T04:42:49Z | 2023-08-25T23:07:39Z | 2023-08-25T23:07:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | putuoka | 20,880,839 | MDQ6VXNlcjIwODgwODM5 | User | false |
huggingface/diffusers | 1,866,344,738 | I_kwDOHa8MBc5vPiki | 4,773 | https://github.com/huggingface/diffusers/issues/4773 | https://api.github.com/repos/huggingface/diffusers/issues/4773 | Loading CivitAI Lora | ### Describe the bug
According to https://github.com/huggingface/diffusers/releases/tag/v0.19.2, i can load Lora by
```
pipe = DiffusionPipeline.from_pretrained("stabilityai/stable-diffusion-xl-base-1.0", torch_dtype=torch.float16)
pipe.load_lora_weights("stabilityai/stable-diffusion-xl-base-1.0", weight_name="sd_x... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-08-25T05:48:25Z | 2023-09-19T06:04:39Z | 2023-09-19T06:04:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MaxTran96 | 15,274,946 | MDQ6VXNlcjE1Mjc0OTQ2 | User | false |
huggingface/diffusers | 1,866,753,600 | I_kwDOHa8MBc5vRGZA | 4,778 | https://github.com/huggingface/diffusers/issues/4778 | https://api.github.com/repos/huggingface/diffusers/issues/4778 | [Discussion] How to allow for more dynamic prompt_embed scaling/weighting/fusion? | We have a couple of issues and requests for the community that ask for the possibility to **dynamically** change certain knobs of Stable Diffusion that are applied at **every denoising step**.
- 1. **Prompt Fusion**. as stated [here](https://github.com/huggingface/diffusers/issues/4496). To implement prompt fusion ... | closed | completed | false | 9 | [
"stale"
] | [] | 2023-08-25T10:03:17Z | 2023-11-09T21:42:39Z | 2023-11-09T21:42:39Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,866,829,573 | I_kwDOHa8MBc5vRY8F | 4,779 | https://github.com/huggingface/diffusers/issues/4779 | https://api.github.com/repos/huggingface/diffusers/issues/4779 | ModelScope Image to VIdeo Pipeline | ### Model/Pipeline/Scheduler description
ModelScope Image to VIdeo Pipeline.
Open sourced and checkpoints are available on huggingface. It is similar to modelscope text to video pipeline.
### Open source status
- [X] The model implementation is available
- [X] The model weights are available (Only relevant if ad... | closed | completed | false | 4 | [
"New pipeline/model",
"wip"
] | [] | 2023-08-25T10:56:34Z | 2024-11-21T15:41:03Z | 2024-11-21T15:41:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | junhaozhang98 | 37,370,309 | MDQ6VXNlcjM3MzcwMzA5 | User | false |
huggingface/diffusers | 1,369,836,481 | I_kwDOHa8MBc5RpgvB | 478 | https://github.com/huggingface/diffusers/issues/478 | https://api.github.com/repos/huggingface/diffusers/issues/478 | 4. Add automatic conversion PT <=> Flax script. | null | closed | completed | false | 1 | [] | [] | 2022-09-12T12:35:58Z | 2022-10-27T09:10:47Z | 2022-10-27T09:10:46Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 1,867,199,774 | I_kwDOHa8MBc5vSzUe | 4,782 | https://github.com/huggingface/diffusers/issues/4782 | https://api.github.com/repos/huggingface/diffusers/issues/4782 | Inpainting modifies areas where mask == 0 | ### Describe the bug
With a inpainting (also controlnet inpainting), areas outside of the mask are slightly modified.
In the example below, we inspect the first pixel, which despite being outside of the mask `(0,0,0)`, the pixel value has changed.
### Reproduction
```python
from diffusers import AutoPipelineFor... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-08-25T14:51:54Z | 2023-11-08T15:08:49Z | 2023-11-08T15:08:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | rb-synth | 135,021,519 | U_kgDOCAxDzw | User | false |
huggingface/diffusers | 1,867,615,207 | I_kwDOHa8MBc5vUYvn | 4,785 | https://github.com/huggingface/diffusers/issues/4785 | https://api.github.com/repos/huggingface/diffusers/issues/4785 | Innequality typo in euler discrete scheduler | ### Describe the bug
I think this is a typo for multiple innequalities. I'd just propose a code change but I'm not familiar with the math for if it is really supposed to be just a < b and b < c
https://github.com/huggingface/diffusers/blob/2764db3194fc1b5069df7292fd938657d8568995/src/diffusers/schedulers/scheduli... | closed | completed | false | 3 | [
"bug",
"stale"
] | [
"williamberman"
] | 2023-08-25T19:58:19Z | 2023-11-08T15:08:48Z | 2023-11-08T15:08:47Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | williamberman | 6,505,431 | MDQ6VXNlcjY1MDU0MzE= | User | false |
huggingface/diffusers | 1,867,686,464 | I_kwDOHa8MBc5vUqJA | 4,786 | https://github.com/huggingface/diffusers/issues/4786 | https://api.github.com/repos/huggingface/diffusers/issues/4786 | [docs] Optimization/special hardware | From https://github.com/huggingface/diffusers/pull/4428#discussion_r1304062172, the `Optimization/Special Hardware` section has grown and it may be time to reorganize it. I want to explore how we can do this, and my initial thoughts are:
To me, there are three subtopics in this section that we can reorganize on:
... | closed | completed | false | 2 | [] | [] | 2023-08-25T21:07:10Z | 2023-09-22T15:38:57Z | 2023-09-22T15:38:57Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stevhliu | 59,462,357 | MDQ6VXNlcjU5NDYyMzU3 | User | false |
huggingface/diffusers | 1,867,914,831 | I_kwDOHa8MBc5vVh5P | 4,788 | https://github.com/huggingface/diffusers/issues/4788 | https://api.github.com/repos/huggingface/diffusers/issues/4788 | SD XL Inpaint: Out of memory on 6 GB VRAM | ### Describe the bug
I'm trying to implement SD XL inpaint in my [add-on](https://github.com/tin2tin/Generative_AI) for Blender. I'm on 6 GB VRAM, and so far I've been able to make ex. txt2vid and img2vid and vid2vid working. I've tried model_cpu_offload and vae_slicing without success. As it is properly not a bug m... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-08-26T05:27:00Z | 2023-10-30T15:07:57Z | 2023-10-30T15:07:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 1,867,969,666 | I_kwDOHa8MBc5vVvSC | 4,789 | https://github.com/huggingface/diffusers/issues/4789 | https://api.github.com/repos/huggingface/diffusers/issues/4789 | Cannot download DeepFloyd IF | ### Describe the bug
I, as well as some of my friends, cannot download the DeepFloyd IF models anymore
### Reproduction
from diffusers import DiffusionPipeline
pipeline = DiffusionPipeline.from_pretrained("DeepFloyd/IF-I-XL-v1.0")
### Logs
_No response_
### System Info
- `diffusers` version: 0.21.0.dev0
- Pl... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2023-08-26T08:11:52Z | 2023-10-30T15:07:55Z | 2023-10-30T15:07:55Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thuanz123 | 32,274,287 | MDQ6VXNlcjMyMjc0Mjg3 | User | false |
huggingface/diffusers | 1,868,006,922 | I_kwDOHa8MBc5vV4YK | 4,790 | https://github.com/huggingface/diffusers/issues/4790 | https://api.github.com/repos/huggingface/diffusers/issues/4790 | `convert_diffusers_to_original_stable_diffusion.py` produces files that `convert_original_stable_diffusion_to_diffusers.py` cannot open | ### Describe the bug
Original stable diffusion checkpoints that have been extracted from diffusers folders via `convert_diffusers_to_original_stable_diffusion.py` cause `convert_original_stable_diffusion_to_diffusers.py` to fail with `AttributeError: 'Attention' object has no attribute 'to_to_k'`:
```
File "/wo... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-08-26T09:56:23Z | 2024-01-08T09:57:38Z | 2023-11-20T15:06:40Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | damian0815 | 144,366 | MDQ6VXNlcjE0NDM2Ng== | User | false |
huggingface/diffusers | 1,868,112,992 | I_kwDOHa8MBc5vWSRg | 4,792 | https://github.com/huggingface/diffusers/issues/4792 | https://api.github.com/repos/huggingface/diffusers/issues/4792 | Docs has a spelling errors in 0.19.3 | 
| closed | completed | false | 4 | [] | [] | 2023-08-26T14:17:24Z | 2023-08-28T06:45:58Z | 2023-08-27T06:51:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | icarried | 54,706,854 | MDQ6VXNlcjU0NzA2ODU0 | User | false |
huggingface/diffusers | 1,868,423,962 | I_kwDOHa8MBc5vXeMa | 4,801 | https://github.com/huggingface/diffusers/issues/4801 | https://api.github.com/repos/huggingface/diffusers/issues/4801 | mix precision bug about examples, train_unconditional.py | ### Describe the bug
Traceback (most recent call last):
File "/home/wayne/project/auto_flow/diffusers/examples/unconditional_image_generation/train_unconditional.py", line 715, in <module>
main(args)
File "/home/wayne/project/auto_flow/diffusers/examples/unconditional_image_generation/train_unconditional.py... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2023-08-27T09:20:27Z | 2023-11-22T15:06:50Z | 2023-11-22T15:06:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WeianMao | 49,094,689 | MDQ6VXNlcjQ5MDk0Njg5 | User | false |
huggingface/diffusers | 1,868,474,119 | I_kwDOHa8MBc5vXqcH | 4,802 | https://github.com/huggingface/diffusers/issues/4802 | https://api.github.com/repos/huggingface/diffusers/issues/4802 | Translate the document into Chinese | #[Docs] zh translation is too slow
| closed | completed | false | 3 | [
"stale"
] | [] | 2023-08-27T12:13:40Z | 2023-10-30T15:07:51Z | 2023-10-30T15:07:51Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | WADreaming | 30,436,033 | MDQ6VXNlcjMwNDM2MDMz | User | false |
huggingface/diffusers | 1,868,484,551 | I_kwDOHa8MBc5vXs_H | 4,803 | https://github.com/huggingface/diffusers/issues/4803 | https://api.github.com/repos/huggingface/diffusers/issues/4803 | Error : could not open requirements file | 
| closed | completed | false | 2 | [] | [] | 2023-08-27T12:48:51Z | 2023-08-28T11:52:32Z | 2023-08-28T11:52:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hsnaf | 83,532,496 | MDQ6VXNlcjgzNTMyNDk2 | User | false |
huggingface/diffusers | 1,868,587,100 | I_kwDOHa8MBc5vYGBc | 4,804 | https://github.com/huggingface/diffusers/issues/4804 | https://api.github.com/repos/huggingface/diffusers/issues/4804 | LoraLoaderMixin | RuntimeError: dictionary changed size during iteration | ### Describe the bug
Related to the fix for #4797
```
File "/workspace/sandbox/test.py", line 18, in <module>
pipe.load_lora_weights(
File "/home/user/mambaforge/envs/tensorml/lib/python3.10/site-packages/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py", line 904, in load_lora_weig... | closed | completed | false | 0 | [
"bug"
] | [] | 2023-08-27T18:23:26Z | 2023-08-28T05:53:27Z | 2023-08-28T05:53:26Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chillpixelfun | 124,188,297 | U_kgDOB2b2iQ | User | false |
huggingface/diffusers | 1,868,859,186 | I_kwDOHa8MBc5vZIcy | 4,806 | https://github.com/huggingface/diffusers/issues/4806 | https://api.github.com/repos/huggingface/diffusers/issues/4806 | Incorrect Parameter Type on UNet2DModel | ### Describe the bug
The parameter type for `UNet2DModel::__init__` seems to be incorrect, at least on these three parameters:
https://github.com/huggingface/diffusers/blob/c4d282360184686f7d4a66787a4be9898cf1b8b0/src/diffusers/models/unet_2d.py#L97-L99
I believe the correct type is `Tuple[str, ...]` and `Tupl... | closed | completed | false | 4 | [
"bug",
"stale"
] | [] | 2023-08-28T02:47:12Z | 2023-10-30T15:07:49Z | 2023-10-30T15:07:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fpgaminer | 1,585,817 | MDQ6VXNlcjE1ODU4MTc= | User | false |
huggingface/diffusers | 1,869,031,658 | I_kwDOHa8MBc5vZyjq | 4,808 | https://github.com/huggingface/diffusers/issues/4808 | https://api.github.com/repos/huggingface/diffusers/issues/4808 | Failing Integration Test on SHAP-E Img2Img Pipeline | ### Describe the bug
The following test:
https://github.com/huggingface/diffusers/blob/766aa50f70120ab8c08c221fb95c68661422fef0/tests/pipelines/shap_e/test_shap_e_img2img.py#L256
Fails to generate the appropriate image from the SHAP-E Pipeline. Looking at the generated and expected images, it looks like the gene... | closed | completed | false | 3 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2023-08-28T06:01:34Z | 2023-10-30T15:07:46Z | 2023-10-30T15:07:46Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DN6 | 7,529,846 | MDQ6VXNlcjc1Mjk4NDY= | User | false |
huggingface/diffusers | 1,869,529,688 | I_kwDOHa8MBc5vbsJY | 4,813 | https://github.com/huggingface/diffusers/issues/4813 | https://api.github.com/repos/huggingface/diffusers/issues/4813 | StableDiffusionControlNetImg2ImgPipeline | how to train a StableDiffusionControlNetImg2ImgPipeline? The official code works for text2img controlnet, but there is no img2img controlnet training code, I don't know where to find the relevant training code. | closed | completed | false | 1 | [] | [] | 2023-08-28T11:01:05Z | 2023-08-28T15:04:24Z | 2023-08-28T15:04:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wanglaofei | 26,437,644 | MDQ6VXNlcjI2NDM3NjQ0 | User | false |
huggingface/diffusers | 1,869,728,854 | I_kwDOHa8MBc5vccxW | 4,814 | https://github.com/huggingface/diffusers/issues/4814 | https://api.github.com/repos/huggingface/diffusers/issues/4814 | How to add more weight to the text prompt in ControlNet? | Hi,
I want to know if there is a quick way of adding more weight to the text prompt in ControlNet during inference.
If so, which parameter needs to be changed?
Thanks, | closed | completed | false | 3 | [
"stale"
] | [] | 2023-08-28T13:05:16Z | 2023-10-30T15:07:45Z | 2023-10-30T15:07:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | miquel-espinosa | 34,089,118 | MDQ6VXNlcjM0MDg5MTE4 | User | false |
huggingface/diffusers | 1,870,047,568 | I_kwDOHa8MBc5vdqlQ | 4,815 | https://github.com/huggingface/diffusers/issues/4815 | https://api.github.com/repos/huggingface/diffusers/issues/4815 | Allow inpainting pipelines to accept latents only inputs | `diffusers` allows to keep text-to-image outputs in latent format to be passed directly to image-2-image pipelines which can save a lot of time & memory, especially when pipelines are chained together (such as is the case for sdxl-base + refiner).
Now we should allow the same functionality for inpainting. We have re... | closed | completed | false | 3 | [] | [
"yiyixuxu"
] | 2023-08-28T16:05:52Z | 2024-12-25T15:32:02Z | 2023-10-02T17:20:16Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,870,055,443 | I_kwDOHa8MBc5vdsgT | 4,816 | https://github.com/huggingface/diffusers/issues/4816 | https://api.github.com/repos/huggingface/diffusers/issues/4816 | Allow image resizing also for torch and numpy inputs | At the moment we only allow to resize images if they are in the PIL format, see here: https://github.com/huggingface/diffusers/blob/e3f3672f46877c18aa4088d852988b35fc4d1e1b/src/diffusers/image_processor.py#L277
I think it's time to allow also such tensors to be resized. In PyTorch this is quite simple by using: http... | closed | completed | false | 7 | [] | [] | 2023-08-28T16:11:12Z | 2023-08-30T07:43:02Z | 2023-08-29T20:45:07Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,870,559,032 | I_kwDOHa8MBc5vfnc4 | 4,818 | https://github.com/huggingface/diffusers/issues/4818 | https://api.github.com/repos/huggingface/diffusers/issues/4818 | Kandinsky2_2 call ignores latents | The latest Kandinsky2_2 version in diffusers `0.20.1` completely ignores the input noise latents. Here are one of the offending lines:
https://github.com/huggingface/diffusers/blob/5eeedd9e3336882d598091e191559f67433b6427/src/diffusers/pipelines/kandinsky2_2/pipeline_kandinsky2_2_prior_emb2emb.py#L532C4-L532C4
Wh... | closed | completed | false | 6 | [] | [
"yiyixuxu"
] | 2023-08-28T21:51:12Z | 2023-09-05T08:19:50Z | 2023-09-05T08:19:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | thoppe | 2,707,106 | MDQ6VXNlcjI3MDcxMDY= | User | false |
huggingface/diffusers | 1,370,054,349 | I_kwDOHa8MBc5RqV7N | 482 | https://github.com/huggingface/diffusers/issues/482 | https://api.github.com/repos/huggingface/diffusers/issues/482 | Some methods mix the usage of numpy and torch | ### Describe the bug
Here are what I find in `resnet.py` (not checking other files at this moment).
Goal: replace this with pure torch methods
- [FirUpsample2D._upsample_2d](https://github.com/huggingface/diffusers/blob/f4781a0b27ffb3ea61ecd25b0b87305e0960304e/src/diffusers/models/resnet.py#L137)
- [FirDownsa... | closed | completed | false | 1 | [
"bug"
] | [] | 2022-09-12T14:58:32Z | 2022-09-16T13:15:06Z | 2022-09-16T13:15:06Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ydshieh | 2,521,628 | MDQ6VXNlcjI1MjE2Mjg= | User | false |
huggingface/diffusers | 1,870,838,299 | I_kwDOHa8MBc5vgrob | 4,820 | https://github.com/huggingface/diffusers/issues/4820 | https://api.github.com/repos/huggingface/diffusers/issues/4820 | SDXL trainig loss nan,validate generage black image | tranning met loss is nan, and the pred_noise cntain nan, which will case the tranning fail， the traing target is predict noise with given noise, which mse alwas nearby 1 (~= 1),
possible reason
1、 lr too large
2、 data had some nan value
i try to continue after find the loss and noise_pred be nan
is there ant... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-08-29T04:01:27Z | 2023-10-30T15:07:43Z | 2023-10-30T15:07:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zdxpan | 16,832,891 | MDQ6VXNlcjE2ODMyODkx | User | false |
huggingface/diffusers | 1,871,113,808 | I_kwDOHa8MBc5vhu5Q | 4,823 | https://github.com/huggingface/diffusers/issues/4823 | https://api.github.com/repos/huggingface/diffusers/issues/4823 | RubberDiffusers. Diffusers but far more flexible! | ### Model/Pipeline/Scheduler description
Every month a new feature comes out and it only works for txt2img or img2img. Why? Because diffusers requires a new pipeline for every single variation. Also, those variations are not very different from one another. So, I made RubberDiffusers! It works just like diffusers, but... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-08-29T08:00:34Z | 2023-10-30T15:07:41Z | 2023-10-30T15:07:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,871,363,261 | I_kwDOHa8MBc5viry9 | 4,826 | https://github.com/huggingface/diffusers/issues/4826 | https://api.github.com/repos/huggingface/diffusers/issues/4826 | ControlNet Reference with MultiControlNet gives errors | ### Describe the bug
I want to use Stable Diffusion ControlNet reference pipeline but I am coming across 2 issues
1) Bug - When executing the below code, I receive an error
```ValueError: `control_guidance_start`: [0.0] has 1 elements but there are 2 controlnets available. Make sure to provide 2.```
I believe the... | closed | completed | false | 13 | [
"bug",
"stale"
] | [] | 2023-08-29T10:29:45Z | 2023-11-20T15:06:36Z | 2023-11-20T15:06:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 0sparsh2 | 50,074,241 | MDQ6VXNlcjUwMDc0MjQx | User | false |
huggingface/diffusers | 1,871,374,054 | I_kwDOHa8MBc5viubm | 4,827 | https://github.com/huggingface/diffusers/issues/4827 | https://api.github.com/repos/huggingface/diffusers/issues/4827 | The finetuned SDXL shows bad performance on Pokeman | Hello, I just git clone the repo and conduct the training along the guidance here https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/README_sdxl.md where I finetuned the SDXL on Pokeman dataset with script train_text_to_image_sdxl.py. However, the generated images are bad-looking. For example,
... | closed | completed | false | 15 | [
"stale"
] | [] | 2023-08-29T10:34:08Z | 2023-11-26T03:51:30Z | 2023-10-30T15:07:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | fiona-lxd | 58,323,411 | MDQ6VXNlcjU4MzIzNDEx | User | false |
huggingface/diffusers | 1,871,567,101 | I_kwDOHa8MBc5vjdj9 | 4,828 | https://github.com/huggingface/diffusers/issues/4828 | https://api.github.com/repos/huggingface/diffusers/issues/4828 | ConnectionError when use StableDiffusionPipeline.from_single_file | ### Describe the bug
ConnectionError: HTTPSConnectionPool(host='raw.githubusercontent.com', port=443): Max retries exceeded with url: /CompVis/stable-diffusion/main/configs/stable-diffusion/v1-inference.yaml (Caused by NewConnectionError('<urllib3.connection.HTTPSConnection object at 0x7f7272dbb550>: Failed to establi... | closed | completed | false | 10 | [
"bug",
"stale"
] | [] | 2023-08-29T12:28:12Z | 2024-02-11T09:29:41Z | 2023-11-20T15:06:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Lanrzip | 48,851,384 | MDQ6VXNlcjQ4ODUxMzg0 | User | false |
huggingface/diffusers | 1,871,675,694 | I_kwDOHa8MBc5vj4Eu | 4,830 | https://github.com/huggingface/diffusers/issues/4830 | https://api.github.com/repos/huggingface/diffusers/issues/4830 | MultiControlNet, wrong tensor size | ### Describe the bug
File "C:\Users\Yakon\AppData\Local\Programs\Python\Python310\lib\site-packages\diffusers\models\controlnet.py", line 754, in forward
sample = sample + controlnet_cond
RuntimeError: The size of tensor a (96) must match the size of tensor b (128) at non-singleton dimension 3
### Reprodu... | closed | completed | false | 8 | [
"bug"
] | [] | 2023-08-29T13:27:06Z | 2023-09-02T22:34:46Z | 2023-09-02T22:34:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Yakonrus | 4,978,662 | MDQ6VXNlcjQ5Nzg2NjI= | User | false |
huggingface/diffusers | 1,871,686,777 | I_kwDOHa8MBc5vj6x5 | 4,831 | https://github.com/huggingface/diffusers/issues/4831 | https://api.github.com/repos/huggingface/diffusers/issues/4831 | How to preview the image during generation,any demo for gradio? | How to preview the image during generation,any demo for gradio? | closed | completed | false | 2 | [] | [] | 2023-08-29T13:32:07Z | 2023-08-30T15:31:31Z | 2023-08-30T15:31:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wodsoe | 40,934,627 | MDQ6VXNlcjQwOTM0NjI3 | User | false |
huggingface/diffusers | 1,872,557,768 | I_kwDOHa8MBc5vnPbI | 4,834 | https://github.com/huggingface/diffusers/issues/4834 | https://api.github.com/repos/huggingface/diffusers/issues/4834 | [Feature Request] Add clip skip | I think it's time to add CLIP skip as described in: https://github.com/huggingface/diffusers/issues/3212#issuecomment-1691714105
Too many issues have been popping up: https://github.com/huggingface/diffusers/issues/3870 e.g.
Anybody interested in taking this one?
@sayakpaul maybe since it's closely related to t... | closed | completed | false | 6 | [] | [] | 2023-08-29T22:01:40Z | 2023-09-19T09:30:55Z | 2023-09-19T09:30:54Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 1,872,768,725 | I_kwDOHa8MBc5voC7V | 4,835 | https://github.com/huggingface/diffusers/issues/4835 | https://api.github.com/repos/huggingface/diffusers/issues/4835 | Bug in StableDiffusionXLControlNetImg2ImgPipeline when using it with MultiControlNetModel | ### Describe the bug
There is an error when using `StableDiffusionXLControlNetImg2ImgPipeline` with `MultiControlNetModel`.
I think in [pipeline_controlnet_sd_xl_img2img.py](https://github.com/huggingface/diffusers/blob/8ccb619416a36f5951dac654a92e869d76db4bbc/src/diffusers/pipelines/controlnet/pipeline_controlnet_... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-08-30T02:15:23Z | 2023-09-01T16:52:00Z | 2023-09-01T16:52:00Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | richardSHkim | 45,237,598 | MDQ6VXNlcjQ1MjM3NTk4 | User | false |
huggingface/diffusers | 1,872,824,190 | I_kwDOHa8MBc5voQd- | 4,836 | https://github.com/huggingface/diffusers/issues/4836 | https://api.github.com/repos/huggingface/diffusers/issues/4836 | A new pipeline for inpainting(object removal) is needed!! | **Is your feature request related to a problem? Please describe.**
Thanks for your great work!!
Right now I'm using StableDiffusionInpaintPipeline with fixed prompt(such as "the empty background") to do object removal task, but one thing i've found is that **when the inpaint mask doesn't cover the shadow or hairline,... | open | null | false | 8 | [
"New pipeline/model",
"community-examples"
] | [] | 2023-08-30T03:31:30Z | 2024-11-30T16:08:32Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lawsonxwl | 52,605,055 | MDQ6VXNlcjUyNjA1MDU1 | User | false |
huggingface/diffusers | 1,872,833,312 | I_kwDOHa8MBc5voSsg | 4,837 | https://github.com/huggingface/diffusers/issues/4837 | https://api.github.com/repos/huggingface/diffusers/issues/4837 | convert_original_stable_diffusion_to_diffusers.py failed SDXL model | ### Describe the bug
Use this script to Convert SDXL based model,
failed in diffusers==0.20.0 or 0.20.1, error:
```bash
download_from_original_stable_diffusion_ckpt
pipe = pipeline_class(
TypeError: __init__() got an unexpected keyword argument 'text_encoder_2'
```
But success in diffusers==0.19.3
### Re... | closed | completed | false | 6 | [
"bug"
] | [] | 2023-08-30T03:42:48Z | 2023-11-12T01:17:03Z | 2023-09-01T03:17:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TimYao18 | 15,173,100 | MDQ6VXNlcjE1MTczMTAw | User | false |
huggingface/diffusers | 1,872,865,939 | I_kwDOHa8MBc5voaqT | 4,839 | https://github.com/huggingface/diffusers/issues/4839 | https://api.github.com/repos/huggingface/diffusers/issues/4839 | Type error: lora_scale | ### Describe the bug
Getting the following error TypeError: Linear.forward() got an unexpected keyword argument 'lora_scale' with this command
accelerate launch train_dreambooth.py --pretrained_model_name_or_path=$MODEL_NAME --instance_data_dir=$INSTANCE_DIR --class_data_dir=$CLASS_DIR --output_dir=$OUTPUT_DIR... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-08-30T04:24:12Z | 2023-08-30T08:40:58Z | 2023-08-30T08:40:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 1,370,107,046 | I_kwDOHa8MBc5Rqiym | 484 | https://github.com/huggingface/diffusers/issues/484 | https://api.github.com/repos/huggingface/diffusers/issues/484 | PNDMSchedulerTest::test_full_loop_no_noise uses time index rather than time_step in `step` | ### Describe the bug
One can see the usage of the `PNDMScheduler` in the `PNDMPipeline` (Or stable diffusion for that matter), and it looks like this, with timesteps passed into the scheduler
https://github.com/huggingface/diffusers/blob/f4781a0b27ffb3ea61ecd25b0b87305e0960304e/src/diffusers/pipelines/pndm/pipeline... | closed | completed | false | 5 | [
"bug",
"good first issue"
] | [] | 2022-09-12T15:32:54Z | 2022-12-20T00:51:38Z | 2022-12-20T00:51:38Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | natolambert | 10,695,622 | MDQ6VXNlcjEwNjk1NjIy | User | false |
huggingface/diffusers | 1,872,880,979 | I_kwDOHa8MBc5voeVT | 4,840 | https://github.com/huggingface/diffusers/issues/4840 | https://api.github.com/repos/huggingface/diffusers/issues/4840 | FlaxAutoencoderKL does not load torch checkpoints correctly | ### Describe the bug
loading sdxl vae weights in flax is missing many required keys
### Reproduction
```py
from diffusers import FlaxAutoencoderKL
flax_model, flax_params = FlaxAutoencoderKL.from_pretrained('stabilityai/sdxl-vae', from_pt=True)
```
### Logs
```
The checkpoint stabilityai/sdxl-vae i... | closed | completed | false | 2 | [
"bug",
"stale"
] | [
"pcuenca"
] | 2023-08-30T04:34:44Z | 2023-11-22T15:06:45Z | 2023-11-22T15:06:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | GallagherCommaJack | 5,674,469 | MDQ6VXNlcjU2NzQ0Njk= | User | false |
huggingface/diffusers | 1,872,933,218 | I_kwDOHa8MBc5vorFi | 4,841 | https://github.com/huggingface/diffusers/issues/4841 | https://api.github.com/repos/huggingface/diffusers/issues/4841 | Controldet image size issue | ### Describe the bug
sholdn't the size of input image and output image same as in original ? the image i got is /8 images.
### Reproduction
using diffusers-0.21.0.dev0
### Logs
_No response_
### System Info
4 2080 ti gpu
### Who can help?
_No response_ | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-08-30T05:30:55Z | 2023-10-30T15:07:33Z | 2023-10-30T15:07:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AparnaAgrawal02 | 52,852,194 | MDQ6VXNlcjUyODUyMTk0 | User | false |
huggingface/diffusers | 1,873,130,595 | I_kwDOHa8MBc5vpbRj | 4,843 | https://github.com/huggingface/diffusers/issues/4843 | https://api.github.com/repos/huggingface/diffusers/issues/4843 | Does sdxl support loading multiple lora? | Does sdxl support loading multiple lora?
| closed | completed | false | 3 | [
"stale"
] | [] | 2023-08-30T07:53:45Z | 2023-10-30T15:07:31Z | 2023-10-30T15:07:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wen020 | 54,690,997 | MDQ6VXNlcjU0NjkwOTk3 | User | false |
huggingface/diffusers | 1,873,351,080 | I_kwDOHa8MBc5vqRGo | 4,846 | https://github.com/huggingface/diffusers/issues/4846 | https://api.github.com/repos/huggingface/diffusers/issues/4846 | train dreambooth ,have strange result | ### Model/Pipeline/Scheduler description
Hello, I trained the dreambooth according to the document, and trained the text_encoder unet at the same time, but how could the predicted result be like this, and I urgently need help. If you don't use unet text encoder, it can be normal a little bit, looking forward to reply
... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-08-30T10:01:52Z | 2023-10-30T15:07:29Z | 2023-10-30T15:07:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 1235555555shan | 85,985,529 | MDQ6VXNlcjg1OTg1NTI5 | User | false |
huggingface/diffusers | 1,874,108,665 | I_kwDOHa8MBc5vtKD5 | 4,848 | https://github.com/huggingface/diffusers/issues/4848 | https://api.github.com/repos/huggingface/diffusers/issues/4848 | lpw_stable_diffusion_xl doesn't work | ### Describe the bug
Prompt is truncated to 77 tokens
prompt="a huge exhibition hall, similar to a hangar, it has no windows, the walls are filled with projection with scenes of winter nature and ice of Baikal, the floor of the hall is made of fabric, under which there are fans, the fabric floats above the floor ... | closed | completed | false | 5 | [
"bug"
] | [] | 2023-08-30T17:39:01Z | 2024-05-08T07:08:06Z | 2023-09-01T04:11:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sergeykorablin | 37,698,016 | MDQ6VXNlcjM3Njk4MDE2 | User | false |
huggingface/diffusers | 1,874,704,269 | I_kwDOHa8MBc5vvbeN | 4,849 | https://github.com/huggingface/diffusers/issues/4849 | https://api.github.com/repos/huggingface/diffusers/issues/4849 | how to use multiple GPUs to train textual inversion? |
I train the textual inversion fine tuning cat toy example from [here](https://github.com/huggingface/diffusers/tree/main/examples/textual_inversion)
my env:
diffusers: 0.20.0
torch: 1.12.1+cu113
accelerate: 0.22.0
train script, as follow:
```
CUDA_VISIBLE_DEVICES="0,1,2,3" python -u textual_inversion.py ... | closed | completed | false | 1 | [] | [] | 2023-08-31T02:56:39Z | 2023-09-11T01:07:49Z | 2023-09-11T01:07:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Adorablepet | 29,625,016 | MDQ6VXNlcjI5NjI1MDE2 | User | false |
huggingface/diffusers | 1,875,134,979 | I_kwDOHa8MBc5vxEoD | 4,851 | https://github.com/huggingface/diffusers/issues/4851 | https://api.github.com/repos/huggingface/diffusers/issues/4851 | Tensor Parallel demo | Hi, I have run SDXL with `Tensor Parallel` as well as `sequence parallel`. Below is my PR, and may it help those who need it.
The Motivation:
Just trying to avoid using `grad checkpointing` to get higher throughput when inputs have higher resolution like 720p.
However, tensor parallel comes at a cost, and I have... | closed | completed | false | 6 | [] | [] | 2023-08-31T09:25:15Z | 2023-09-27T02:57:31Z | 2023-09-07T07:27:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | KimmiShi | 13,433,751 | MDQ6VXNlcjEzNDMzNzUx | User | false |
huggingface/diffusers | 1,875,166,204 | I_kwDOHa8MBc5vxMP8 | 4,852 | https://github.com/huggingface/diffusers/issues/4852 | https://api.github.com/repos/huggingface/diffusers/issues/4852 | Error while loading lora | ### Describe the bug
I am trying to load a lora to StableDiffusionPipeline , it is giving the error :
"""
[/usr/local/lib/python3.10/dist-packages/diffusers/loaders.py] in _modify_text_encoder(cls, text_encoder, lora_scale, network_alphas, rank, dtype, patch_mlp)
1388 if patch_mlp:
1389 f... | closed | not_planned | false | 3 | [
"bug"
] | [] | 2023-08-31T09:43:12Z | 2023-09-04T08:27:55Z | 2023-09-04T08:27:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Garry435 | 143,331,750 | U_kgDOCIsRpg | User | false |
huggingface/diffusers | 1,875,375,900 | I_kwDOHa8MBc5vx_cc | 4,854 | https://github.com/huggingface/diffusers/issues/4854 | https://api.github.com/repos/huggingface/diffusers/issues/4854 | Inpainting only does the first image | ### Describe the bug
I'm trying to inpaint 4 different images, but I only get the first one repeated 4 times.
I believe the error is in this file src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint.py
I did a bit of digging and it seems `prepare_mask_and_maked_image` works fine, masking ... | closed | completed | false | 9 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2023-08-31T11:53:13Z | 2023-11-08T15:08:29Z | 2023-11-08T15:08:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | landmann | 12,756,081 | MDQ6VXNlcjEyNzU2MDgx | User | false |
huggingface/diffusers | 1,875,739,993 | I_kwDOHa8MBc5vzYVZ | 4,857 | https://github.com/huggingface/diffusers/issues/4857 | https://api.github.com/repos/huggingface/diffusers/issues/4857 | Function to Convert PIL->Latents | **Is your feature request related to a problem? Please describe.**
I'm trying to use a StableDiffusionXLControlNetPipeline which takes "latents" as an argument to set the initial noise for the diffusion process. In other image-to-image diffuser pipelines, we can pass an image to serve this purpose, which then gets con... | closed | completed | false | 5 | [] | [] | 2023-08-31T15:14:59Z | 2023-11-15T23:38:58Z | 2023-09-07T04:06:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ksteinfe | 2,659,599 | MDQ6VXNlcjI2NTk1OTk= | User | false |
huggingface/diffusers | 1,370,292,446 | I_kwDOHa8MBc5RrQDe | 486 | https://github.com/huggingface/diffusers/issues/486 | https://api.github.com/repos/huggingface/diffusers/issues/486 | Invisible Watermarking and Inclusion as Default | Hi all,
Love the work y’all are doing!
The original [CompVis Stable Diffusion repo](https://github.com/CompVis/stable-diffusion) includes invisible watermarking so that images generated don’t end-up polluting newly scraped training data with automatically generated results, [as is already happening for text data.... | closed | completed | false | 3 | [
"stale"
] | [] | 2022-09-12T17:42:34Z | 2022-11-16T15:04:14Z | 2022-11-16T15:04:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ari-holtzman | 20,871,523 | MDQ6VXNlcjIwODcxNTIz | User | false |
huggingface/diffusers | 1,876,911,969 | I_kwDOHa8MBc5v32dh | 4,864 | https://github.com/huggingface/diffusers/issues/4864 | https://api.github.com/repos/huggingface/diffusers/issues/4864 | fix guidance_rescale docstring to reflect true default value | guidance_rescale is a parameter available in many pipelines that was implemented in this [pr](https://github.com/huggingface/diffusers/pull/3664).
To maintain backwards compatibility, and due to concerns about the results with guidance_scale applied not being objectively better all the time, the default value was de... | closed | completed | false | 2 | [] | [] | 2023-09-01T07:37:06Z | 2023-09-18T11:39:14Z | 2023-09-18T11:39:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | codinggosu | 16,798,331 | MDQ6VXNlcjE2Nzk4MzMx | User | false |
huggingface/diffusers | 1,877,409,010 | I_kwDOHa8MBc5v5vzy | 4,866 | https://github.com/huggingface/diffusers/issues/4866 | https://api.github.com/repos/huggingface/diffusers/issues/4866 | Failed to establish a new connection | I have cached the model locally before. But I get this error.
```sh
Traceback (most recent call last):
File "/root/data/xyf/Personalization/sd-scripts/inpaint_img.py", line 26, in <module>
pipe = StableDiffusionInpaintPipeline.from_single_file(
File "/root/miniconda3/envs/py310/lib/python3.10/site-packag... | closed | completed | false | 9 | [
"stale"
] | [] | 2023-09-01T13:15:26Z | 2024-01-17T13:31:49Z | 2023-10-30T15:07:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | YoucanBaby | 43,912,540 | MDQ6VXNlcjQzOTEyNTQw | User | false |
huggingface/diffusers | 1,878,053,760 | I_kwDOHa8MBc5v8NOA | 4,868 | https://github.com/huggingface/diffusers/issues/4868 | https://api.github.com/repos/huggingface/diffusers/issues/4868 | AttributeError: 'super' object has no attribute '__getattr__'. Did you mean: '__setattr__'? | ### Describe the bug
I'm getting this error after running quickstart code on Arch:
```
Traceback (most recent call last):
File "/home/kuchizu/Desktop/StableDiffusion/main.py", line 4, in <module>
pipeline = DiffusionPipeline.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=torch.float16)
... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-09-01T21:16:29Z | 2023-09-01T23:33:58Z | 2023-09-01T23:33:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Kuchizu | 70,284,260 | MDQ6VXNlcjcwMjg0MjYw | User | false |
huggingface/diffusers | 1,370,496,494 | I_kwDOHa8MBc5RsB3u | 487 | https://github.com/huggingface/diffusers/issues/487 | https://api.github.com/repos/huggingface/diffusers/issues/487 | `StableDiffusionPipeline` does not work on M1 CPU | ### Describe the bug
When trying to run `StableDiffusionPipeline` locally on my M1 Mac (used CPU), I obtain torch type mismatch. Is it expected to be able to run the example on CPU?
### Reproduction
The example from `README.md`
```python
# make sure you're logged in with `huggingface-cli login`
from torch... | closed | completed | false | 3 | [
"bug"
] | [] | 2022-09-12T20:44:40Z | 2022-09-13T07:48:20Z | 2022-09-13T07:48:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | stancld | 46,073,029 | MDQ6VXNlcjQ2MDczMDI5 | User | false |
huggingface/diffusers | 1,878,124,442 | I_kwDOHa8MBc5v8eea | 4,871 | https://github.com/huggingface/diffusers/issues/4871 | https://api.github.com/repos/huggingface/diffusers/issues/4871 | How to run "StableDiffusionXLPipeline.from_single_file"? | I got an error when I ran the following code and it got an error on the line "pipe = StableDiffusionXLPipeline." and how to solve it?
notes:
I don't have a model refiner, I just want to run a model with a DIffuser XL
```
from diffusers import StableDiffusionXLPipeline, StableDiffusionXLImg2ImgPipeline
import t... | closed | completed | false | 1 | [] | [] | 2023-09-01T22:42:25Z | 2023-09-09T03:35:53Z | 2023-09-09T03:35:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Damarcreative | 106,896,172 | U_kgDOBl8bLA | User | false |
huggingface/diffusers | 1,878,478,925 | I_kwDOHa8MBc5v91BN | 4,872 | https://github.com/huggingface/diffusers/issues/4872 | https://api.github.com/repos/huggingface/diffusers/issues/4872 | CUDA out of memory on T4 Colab in SDXL finetuning | ### Describe the bug
I'm trying to fine tuning a SDXL model using a T4 (Colab notebook)
here the command:
!accelerate launch train_text_to_image_lora_sdxl.py \
--pretrained_model_name_or_path="stabilityai/stable-diffusion-xl-base-1.0" \
--pretrained_vae_model_name_or_path="madebyollin/sdxl-vae-f... | closed | completed | false | 2 | [
"bug"
] | [] | 2023-09-02T08:08:00Z | 2023-09-02T09:23:26Z | 2023-09-02T08:50:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | EnricoBeltramo | 22,332,651 | MDQ6VXNlcjIyMzMyNjUx | User | false |
huggingface/diffusers | 1,878,482,343 | I_kwDOHa8MBc5v912n | 4,873 | https://github.com/huggingface/diffusers/issues/4873 | https://api.github.com/repos/huggingface/diffusers/issues/4873 | Why `train_controlnet_webdataset.py` uses `EulerDiscreteScheduler`? | I noticed that the train_controlnet_webdataset.py script in our codebase uses EulerDiscreteScheduler for training, whereas most other training scripts use DDPMScheduler. Furthermore, I observed that there are some differences between this script and the original train_controlnet_sdxl.py script.
I'm interested in und... | closed | completed | false | 4 | [] | [] | 2023-09-02T08:17:28Z | 2023-10-20T01:24:03Z | 2023-09-05T01:16:00Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | okotaku | 24,734,142 | MDQ6VXNlcjI0NzM0MTQy | User | false |
huggingface/diffusers | 1,878,610,468 | I_kwDOHa8MBc5v-VIk | 4,874 | https://github.com/huggingface/diffusers/issues/4874 | https://api.github.com/repos/huggingface/diffusers/issues/4874 | StableDiffusionUpscalePipeline low_res_scheduler bug when using from local | ### Describe the bug
I'm following the example for using an upscaler from here [StableDiffusionUpscalePipeline](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/upscale#diffusers.StableDiffusionUpscalePipeline.__call__.example). But when I run the code I keep getting the following error: ... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2023-09-02T10:54:38Z | 2023-10-30T15:07:20Z | 2023-10-30T15:07:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | gate3 | 3,597,437 | MDQ6VXNlcjM1OTc0Mzc= | User | false |
huggingface/diffusers | 1,878,669,710 | I_kwDOHa8MBc5v-jmO | 4,875 | https://github.com/huggingface/diffusers/issues/4875 | https://api.github.com/repos/huggingface/diffusers/issues/4875 | DPM++ SDE sampling crashes if batch size changes between invocations | ### Describe the bug
When using `DPMSolverSDEScheduler`, the internal `self.noise_sampler` is initialized to return noise with a batch size matching the first batch passed, and then cached. Subsequent usages with the same batch size function, but if the batch size changes `RuntimeError` is raised. For example, with ... | closed | completed | false | 4 | [
"bug"
] | [
"yiyixuxu"
] | 2023-09-02T12:32:32Z | 2023-09-14T16:48:37Z | 2023-09-14T16:48:37Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | damian0815 | 144,366 | MDQ6VXNlcjE0NDM2Ng== | User | false |
huggingface/diffusers | 1,878,868,221 | I_kwDOHa8MBc5v_UD9 | 4,876 | https://github.com/huggingface/diffusers/issues/4876 | https://api.github.com/repos/huggingface/diffusers/issues/4876 | It would be amazing if the std of lora layers was an argument. | **Is your feature request related to a problem? Please describe.**
A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]
**Describe the solution you'd like**
A clear and concise description of what you want to happen.
**Describe alternatives you've considered**
A clear an... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-09-02T22:43:20Z | 2023-10-30T15:07:18Z | 2023-10-30T15:07:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aiXander | 24,993,604 | MDQ6VXNlcjI0OTkzNjA0 | User | false |
huggingface/diffusers | 1,878,868,472 | I_kwDOHa8MBc5v_UH4 | 4,877 | https://github.com/huggingface/diffusers/issues/4877 | https://api.github.com/repos/huggingface/diffusers/issues/4877 | Make noise_init_strength of lora layer weights an input arg | I'm curious why the std of the weights is set to 1/rank by default?
https://github.com/huggingface/diffusers/blob/main/src/diffusers/models/lora.py#L40 | closed | completed | false | 1 | [] | [] | 2023-09-02T22:44:47Z | 2023-09-07T06:20:37Z | 2023-09-07T06:20:37Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | aiXander | 24,993,604 | MDQ6VXNlcjI0OTkzNjA0 | User | false |
huggingface/diffusers | 1,879,086,030 | I_kwDOHa8MBc5wAJPO | 4,878 | https://github.com/huggingface/diffusers/issues/4878 | https://api.github.com/repos/huggingface/diffusers/issues/4878 | Reuse weight on pipeline that loaded using from_single_file | The title pretty much.
I used this and it failed:
```
pipe=StableDiffusionRubberPipeline.from_single_file(
"./checkpoint/lyriel.safetensors", torch_dtype=torch.float32,local_files_only=True,safety_checker=None, requires_safety_checker=False,
).to("cuda")
tempipe=StableDiffusionRubberPipeline.from_single_f... | closed | completed | false | 9 | [
"stale"
] | [] | 2023-09-03T12:55:14Z | 2023-11-22T15:06:43Z | 2023-11-22T15:06:43Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
huggingface/diffusers | 1,879,099,616 | I_kwDOHa8MBc5wAMjg | 4,879 | https://github.com/huggingface/diffusers/issues/4879 | https://api.github.com/repos/huggingface/diffusers/issues/4879 | multi subject dreambooth,py | 
Given 2 subjects , tried in Google collab , getting this error Could you please check and if possible could you please give the support for SDXL multi subject @Kopsahlong
Thank yo... | closed | completed | false | 2 | [
"stale"
] | [] | 2023-09-03T13:28:56Z | 2023-10-30T15:07:15Z | 2023-10-30T15:07:15Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sandeep-Narahari | 46,187,761 | MDQ6VXNlcjQ2MTg3NzYx | User | false |
huggingface/diffusers | 1,370,663,284 | I_kwDOHa8MBc5Rsql0 | 488 | https://github.com/huggingface/diffusers/issues/488 | https://api.github.com/repos/huggingface/diffusers/issues/488 | Diffusion Pipeline numpy_to_pil fails for grayscale/single channel images | ### Describe the bug
This is my first time creating an issue and I'm just starting to use the diffusers library. Really enjoying it! Also, apologies if I'm using the `DiffusionPipeline` incorrectly.
I'm trying to follow along the [unconditional image generation example notebook](https://github.com/huggingface/diff... | closed | completed | false | 2 | [
"bug"
] | [
"anton-l"
] | 2022-09-12T23:52:24Z | 2022-10-27T20:44:36Z | 2022-10-27T20:44:36Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | st-howard | 97,485,469 | U_kgDOBc-CnQ | User | false |
huggingface/diffusers | 1,879,100,345 | I_kwDOHa8MBc5wAMu5 | 4,880 | https://github.com/huggingface/diffusers/issues/4880 | https://api.github.com/repos/huggingface/diffusers/issues/4880 | StableDiffusionControlNetPipeline single controlnet TypeError: image must be passed and be one of PIL image, torch tensor, list of PIL images, or list of torch tensors | ### Describe the bug
When I use `StableDiffusionControlNetPipeline` with single controlnet. `pipe` does not accept single `PIL` image.
### Reproduction
```python
from PIL import Image
from diffusers import ControlNetModel, StableDiffusionControlNetPipeline
from diffusers.utils import load_image
device: s... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2023-09-03T13:31:18Z | 2024-05-08T02:05:55Z | 2023-10-30T15:07:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elcolie | 18,206,728 | MDQ6VXNlcjE4MjA2NzI4 | User | false |
huggingface/diffusers | 1,879,779,239 | I_kwDOHa8MBc5wCyen | 4,884 | https://github.com/huggingface/diffusers/issues/4884 | https://api.github.com/repos/huggingface/diffusers/issues/4884 | Can't import StableDiffusionXLControlNetImg2ImgPipeline | Hey !
I'm trying to use the class `StableDiffusionXLControlNetImg2ImgPipeline` but I can't find it in v0.20.2.
Is it not contain in the last release ? If yes is it possible to use it anyway ?
Might me a random noob python question, sorry haha
Thanks ! | closed | completed | false | 1 | [] | [] | 2023-09-04T08:31:05Z | 2023-09-04T10:30:34Z | 2023-09-04T10:30:34Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Wraken | 13,450,846 | MDQ6VXNlcjEzNDUwODQ2 | User | false |
huggingface/diffusers | 1,880,459,811 | I_kwDOHa8MBc5wFYoj | 4,885 | https://github.com/huggingface/diffusers/issues/4885 | https://api.github.com/repos/huggingface/diffusers/issues/4885 | StableDiffusionXLControlNetImg2ImgPipeline doesn't load LoRA weights | ### Describe the bug
I get `object has no attribute` error when I use `load_lora_weights()` in the `StableDiffusionXLControlNetImg2ImgPipeline`
Other pipelines such as `StableDiffusionXLPipeline` are working.
### Reproduction
```python
import torch
from diffusers import (
ControlNetModel,
Autoenco... | closed | completed | false | 1 | [
"bug"
] | [] | 2023-09-04T14:57:08Z | 2023-09-13T09:05:22Z | 2023-09-13T09:05:22Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clixff | 10,634,323 | MDQ6VXNlcjEwNjM0MzIz | User | false |
huggingface/diffusers | 1,880,772,264 | I_kwDOHa8MBc5wGk6o | 4,888 | https://github.com/huggingface/diffusers/issues/4888 | https://api.github.com/repos/huggingface/diffusers/issues/4888 | Load model safetensors completely offline | ### Describe the bug
the issue in
Load model safetensors completely offline
I see that `FromSingleFileMixin.from_single_file` does not pars `original_config_file`
to pass it to `download_from_original_stable_diffusion_ckpt`
### Reproduction
load any safetensors model without any internet
### Logs
```shell
ur... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-09-04T19:19:40Z | 2023-09-17T12:09:49Z | 2023-09-17T12:09:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | femto-coder | 9,624,109 | MDQ6VXNlcjk2MjQxMDk= | User | false |
huggingface/diffusers | 1,880,880,690 | I_kwDOHa8MBc5wG_Yy | 4,889 | https://github.com/huggingface/diffusers/issues/4889 | https://api.github.com/repos/huggingface/diffusers/issues/4889 | ValueError: Key `down_blocks.0.attentions.0.transformer_blocks.0.attn1.processor` is invalid, expected torch.Tensor but received <class 'dict'> | ### Describe the bug
After completing the training I got this error while training custom diffusion
### Reproduction
`!accelerate launch train_custom_diffusion.py \
--pretraine... | closed | completed | false | 3 | [
"bug"
] | [] | 2023-09-04T21:26:27Z | 2023-09-06T18:49:40Z | 2023-09-06T18:49:40Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sandeep-Narahari | 46,187,761 | MDQ6VXNlcjQ2MTg3NzYx | User | false |
huggingface/diffusers | 1,370,792,761 | I_kwDOHa8MBc5RtKM5 | 489 | https://github.com/huggingface/diffusers/issues/489 | https://api.github.com/repos/huggingface/diffusers/issues/489 | Is it possible to convert the onnx model to fp16 model? | The torch example gives parameter ```revision="fp16"```, can onnx model do the same optimization? Current onnx inference(using CUDAExecutionProvider) is slower than torch version, and used more gpu memory than torch version(12G vs 4G). | closed | completed | false | 27 | [
"stale"
] | [
"anton-l"
] | 2022-09-13T03:18:39Z | 2023-04-04T15:04:09Z | 2023-04-04T15:04:09Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yuananf | 4,101,738 | MDQ6VXNlcjQxMDE3Mzg= | User | false |
huggingface/diffusers | 1,880,971,270 | I_kwDOHa8MBc5wHVgG | 4,892 | https://github.com/huggingface/diffusers/issues/4892 | https://api.github.com/repos/huggingface/diffusers/issues/4892 | [Pipeline] TokenFlow: Consistent Diffusion Features for Consistent Video Editing | ### Model/Pipeline/Scheduler description
TokenFlow is a framework that enables consistent video editing, using a pre-trained text-to-image diffusion model, without any further training or finetuning via nearest-neighbor matching of (noised) latent diffusion features between frames.
The generative AI revolution has ... | closed | completed | false | 3 | [
"stale"
] | [] | 2023-09-04T23:53:42Z | 2023-10-30T15:07:09Z | 2023-10-30T15:07:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | clarencechen | 8,482,341 | MDQ6VXNlcjg0ODIzNDE= | User | false |
huggingface/diffusers | 1,881,130,498 | I_kwDOHa8MBc5wH8YC | 4,894 | https://github.com/huggingface/diffusers/issues/4894 | https://api.github.com/repos/huggingface/diffusers/issues/4894 | "from_single_file" consumes a lot of CPU RAM and cannot be allocated to GPU/CUDA. | reproduction code https://colab.research.google.com/drive/1v0xiESZi187acMP-aHSJLvtVm8MK5UTI
Unlike from_pretrained, even when using "torch_dtype=torch.float16" and the .to("cuda") command, it is still processed using CPU RAM and consumes very high amounts, crashing with over 12GB of CPU RAM usage.
I have tried it... | closed | completed | false | 22 | [] | [] | 2023-09-05T03:41:27Z | 2024-02-16T08:27:50Z | 2023-12-02T15:06:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | putuoka | 20,880,839 | MDQ6VXNlcjIwODgwODM5 | User | false |
huggingface/diffusers | 1,881,327,678 | I_kwDOHa8MBc5wIsg- | 4,895 | https://github.com/huggingface/diffusers/issues/4895 | https://api.github.com/repos/huggingface/diffusers/issues/4895 | Can I train an existing stable diffusion model to improve the existing model by adding new datasheets? | Previously I created a new model using "sd-easy-mode" with the same code in this repository.
Can I improve my existing model with the addition of new datashets?
Here are my models:
https://huggingface.co/tensor-diffusion/melaura-v1-0/tree/main | closed | completed | false | 1 | [] | [] | 2023-09-05T06:57:56Z | 2023-09-09T03:24:56Z | 2023-09-09T03:24:56Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Damarcreative | 106,896,172 | U_kgDOBl8bLA | User | false |
huggingface/diffusers | 1,881,394,322 | I_kwDOHa8MBc5wI8yS | 4,898 | https://github.com/huggingface/diffusers/issues/4898 | https://api.github.com/repos/huggingface/diffusers/issues/4898 | I'd like to use vae-ft-mse-840000-ema-pruned, but I don't have config.... should be usable with safetensors only | Hi, I'm trying to use vae-ft-mse-840000-ema-pruned, but diffusers require a config.js file. At first, I downloaded the safetensors file from civitai, then tried this: stabilityai/sd-vae-ft-mse-original which didn't work because it doesn't have a config file. I looked through the community discussion and someone said to... | closed | completed | false | 2 | [] | [] | 2023-09-05T07:41:35Z | 2023-09-07T07:35:57Z | 2023-09-07T07:35:57Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | alexblattner | 15,870,094 | MDQ6VXNlcjE1ODcwMDk0 | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.