repo stringclasses 1
value | github_id int64 1.27B 4.61B | github_node_id stringlengths 18 24 | number int64 8 13.9k | 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 2026-06-08 13:44:32 | updated_at stringdate 2022-06-12 22:18:01 2026-06-08 13:47:08 | closed_at stringdate 2022-06-12 22:18:01 2026-06-08 10:50:20 ⌀ | author_association stringclasses 3
values | milestone_title stringclasses 0
values | snapshot_id stringclasses 64
values | extracted_at stringdate 2026-04-07 13:34:13 2026-06-08 14:35:15 | author_login stringlengths 3 28 | author_id int64 1.54k 290M | author_node_id stringlengths 12 20 | author_type stringclasses 3
values | author_site_admin bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
huggingface/diffusers | 2,418,309,358 | I_kwDOHa8MBc6QJHju | 8,903 | https://github.com/huggingface/diffusers/issues/8903 | https://api.github.com/repos/huggingface/diffusers/issues/8903 | aina | ### Model/Pipeline/Scheduler description
modelo asiatica en kyoto
### Open source status
- [x] 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 | 4 | [] | [] | 2024-07-19T08:01:37Z | 2024-08-30T11:06:57Z | 2024-08-30T11:06:29Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patproduuuuu | 176,018,116 | U_kgDOCn3SxA | User | false |
huggingface/diffusers | 2,418,315,965 | I_kwDOHa8MBc6QJJK9 | 8,904 | https://github.com/huggingface/diffusers/issues/8904 | https://api.github.com/repos/huggingface/diffusers/issues/8904 |  | 
_Originally posted by @patproduuuuu in https://github.com/huggingface/diffusers/issues/8903#issuecomment-2238597810_
... | closed | completed | false | 2 | [] | [] | 2024-07-19T08:05:52Z | 2024-08-30T11:07:02Z | 2024-08-30T11:06:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patproduuuuu | 176,018,116 | U_kgDOCn3SxA | User | false |
huggingface/diffusers | 2,418,444,324 | I_kwDOHa8MBc6QJogk | 8,906 | https://github.com/huggingface/diffusers/issues/8906 | https://api.github.com/repos/huggingface/diffusers/issues/8906 | there is no qk_norm in SD3Transformer2DModel. Is that right? | ### Describe the bug
there is no qk_norm in SD3Transformer2DModel. Is that right?
self.attn = Attention(
query_dim=dim,
cross_attention_dim=None,
added_kv_proj_dim=dim,
dim_head=attention_head_dim // num_attention_heads,
heads=num_attention_he... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-07-19T09:18:05Z | 2024-10-31T19:19:24Z | 2024-10-31T19:19:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | heart-du | 115,032,254 | U_kgDOBttAvg | User | false |
huggingface/diffusers | 2,418,553,607 | I_kwDOHa8MBc6QKDMH | 8,907 | https://github.com/huggingface/diffusers/issues/8907 | https://api.github.com/repos/huggingface/diffusers/issues/8907 | [Tests] Improve transformers model test suite coverage | Currently, we have different variants of transformers: https://github.com/huggingface/diffusers/tree/main/src/diffusers/models/transformers/. However, we don't have test suites for each of them: https://github.com/huggingface/diffusers/tree/main/tests/models/transformers/.
We are seeking contributions from the comm... | closed | completed | false | 6 | [
"Good second issue",
"contributions-welcome"
] | [] | 2024-07-19T10:14:34Z | 2024-08-19T03:00:12Z | 2024-08-19T03:00:12Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,419,708,990 | I_kwDOHa8MBc6QOdQ- | 8,909 | https://github.com/huggingface/diffusers/issues/8909 | https://api.github.com/repos/huggingface/diffusers/issues/8909 | Getting problems saving text_encoder_ti when using train_dreambooth_lora_sd15_advanced | ### Describe the bug
Hi,
I encountered an issue while training Stable Diffusion 1.5 using the script `examples/advanced_diffusion_training/train_dreambooth_lora_sd15_advanced.py`.
The problem occurs when trying to save the text_encoder while also using textual inversion.
The following line of code seems to ... | open | null | false | 3 | [
"bug",
"stale"
] | [] | 2024-07-19T19:37:54Z | 2024-09-14T15:05:57Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jleehun | 76,911,999 | MDQ6VXNlcjc2OTExOTk5 | User | false |
huggingface/diffusers | 2,420,541,807 | I_kwDOHa8MBc6QRolv | 8,911 | https://github.com/huggingface/diffusers/issues/8911 | https://api.github.com/repos/huggingface/diffusers/issues/8911 | Bugs in PixArtImageProcessor.resize_and_crop_tensor | ### Describe the bug
`PixArtImageProcessor.resize_and_crop_tensor` returns image with incorrect size
### Reproduction
```python
import torch
from diffusers.image_processor import PixArtImageProcessor
orig_height, orig_width = (832, 1152)
new_height, new_width = (960, 1280)
image = torch.rand((1, 3, orig_hei... | open | null | false | 2 | [
"bug",
"stale"
] | [] | 2024-07-20T03:59:45Z | 2024-09-14T15:05:51Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MagiaSN | 15,558,746 | MDQ6VXNlcjE1NTU4NzQ2 | User | false |
huggingface/diffusers | 2,420,867,203 | I_kwDOHa8MBc6QS4CD | 8,915 | https://github.com/huggingface/diffusers/issues/8915 | https://api.github.com/repos/huggingface/diffusers/issues/8915 | ImportError: cannot import name 'AutoPipelineForText2Image' from 'diffusers' | ### Describe the bug
It seems like an old issue that was mentioned a few times before. I encountered it at version 0.29.2.
I am able to finished the training and get the weights, so probably environment setting is fine. The import error is shown on inference stage from the below script.
### Reproduction
```
... | closed | completed | false | 6 | [
"bug"
] | [] | 2024-07-20T12:00:26Z | 2024-07-26T01:13:53Z | 2024-07-26T01:13:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | blueclowd | 7,544,000 | MDQ6VXNlcjc1NDQwMDA= | User | false |
huggingface/diffusers | 2,421,187,285 | I_kwDOHa8MBc6QUGLV | 8,918 | https://github.com/huggingface/diffusers/issues/8918 | https://api.github.com/repos/huggingface/diffusers/issues/8918 | train_dreambooth_sd3.py still broken | ### Describe the bug
incorrect code/logic, referencing variables issue
### Reproduction
run train_dreambooth_sd3.py with valid args
### Logs
text_input_ids_list referenced before variable assignment
### System Info
diffusers 30 dev (source)
### Who can help?
_No response_ | open | null | false | 6 | [
"bug",
"stale"
] | [] | 2024-07-21T03:04:05Z | 2024-12-14T15:04:55Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | neuron-party | 96,799,331 | U_kgDOBcUKYw | User | false |
huggingface/diffusers | 1,412,479,785 | I_kwDOHa8MBc5UMLsp | 892 | https://github.com/huggingface/diffusers/issues/892 | https://api.github.com/repos/huggingface/diffusers/issues/892 | v0.4.0+ breaks compatibility with Nvidia Deep Learning Containers | ### Describe the bug
[Nvidia's deep learning containers](https://catalog.ngc.nvidia.com/orgs/nvidia/containers/pytorch) are a popular way to run machine learning workloads on top of Docker.
With diffusers 0.4.0+, I'm unable to import `diffusers` inside of this container because `torch.backends.mps` doesn't exist.... | closed | completed | false | 4 | [
"bug"
] | [] | 2022-10-18T02:35:20Z | 2022-10-20T17:23:43Z | 2022-10-20T17:23:43Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tasercake | 13,855,549 | MDQ6VXNlcjEzODU1NTQ5 | User | false |
huggingface/diffusers | 2,421,391,391 | I_kwDOHa8MBc6QU4Af | 8,922 | https://github.com/huggingface/diffusers/issues/8922 | https://api.github.com/repos/huggingface/diffusers/issues/8922 | Stable Diffusion 3 single file safetensors model with 8b quantized T5xxl returns zeroes, black image | ### Describe the bug
Loading `sd3_medium_incl_clips_t5xxlfp8.safetensors` like the official example generates a black image of zeroes.
People in community found same to be happening with fp16 model as well.
Related huggingface thread:
https://huggingface.co/stabilityai/stable-diffusion-3-medium/discussions/161
#... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-07-21T12:55:07Z | 2024-07-22T17:22:12Z | 2024-07-22T13:57:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hnhparitosh | 36,197,416 | MDQ6VXNlcjM2MTk3NDE2 | User | false |
huggingface/diffusers | 2,422,133,208 | I_kwDOHa8MBc6QXtHY | 8,924 | https://github.com/huggingface/diffusers/issues/8924 | https://api.github.com/repos/huggingface/diffusers/issues/8924 | Adding Differential Diffusion to Kolors, Auraflow, HunyuanDiT | Diffusers recently added support for the following models:
- [x] [Kolors](https://github.com/huggingface/diffusers/pull/8812) (@tuanh123789)
- [x] [AuraFlow](https://github.com/huggingface/diffusers/pull/8796)
- [x] [HunyuanDiT](https://github.com/huggingface/diffusers/pull/8240) (@MnCSSJ4x)
A few weeks ago, we a... | closed | completed | false | 28 | [
"good first issue",
"help wanted",
"Good second issue",
"contributions-welcome"
] | [
"tuanh123789",
"devesh-2002"
] | 2024-07-22T07:17:58Z | 2024-10-31T19:18:32Z | 2024-10-31T19:18:32Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | a-r-r-o-w | 72,266,394 | MDQ6VXNlcjcyMjY2Mzk0 | User | false |
huggingface/diffusers | 2,422,932,137 | I_kwDOHa8MBc6QawKp | 8,930 | https://github.com/huggingface/diffusers/issues/8930 | https://api.github.com/repos/huggingface/diffusers/issues/8930 | StableDiffusionXLControlNetImg2ImgPipeline often fails to respect "pose" control images | ### Describe the bug
Hello,
Using [StableDiffusionXLControlNetImg2ImgPipeline](https://huggingface.co/docs/diffusers/en/api/pipelines/controlnet_sdxl#diffusers.StableDiffusionXLControlNetImg2ImgPipeline), and passing a "pose" control image often fails to produce an output image that maintains the pose.
I couldn'... | open | null | false | 14 | [
"bug",
"stale"
] | [] | 2024-07-22T13:48:48Z | 2024-09-21T07:48:04Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Clement-Lelievre | 70,368,164 | MDQ6VXNlcjcwMzY4MTY0 | User | false |
huggingface/diffusers | 2,423,252,231 | I_kwDOHa8MBc6Qb-UH | 8,933 | https://github.com/huggingface/diffusers/issues/8933 | https://api.github.com/repos/huggingface/diffusers/issues/8933 | kolors lora training | please add lora training for kolors | closed | completed | false | 12 | [] | [] | 2024-07-22T16:13:05Z | 2025-01-19T12:16:22Z | 2024-11-18T20:00:24Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | loboere | 15,265,895 | MDQ6VXNlcjE1MjY1ODk1 | User | false |
huggingface/diffusers | 2,423,732,203 | I_kwDOHa8MBc6Qdzfr | 8,935 | https://github.com/huggingface/diffusers/issues/8935 | https://api.github.com/repos/huggingface/diffusers/issues/8935 | Tensor not on the same device for long prompts in `AuraFlowPipeline` | ### Describe the bug
When I generate an image with a long prompt in `AuraFlowPipeline`, it raises this error:
```
File "~/workspace/anaconda3/envs/diffusers/lib/python3.11/site-packages/diffusers/pipelines/aura_flow/pipeline_aura_flow.py", line 507, in __call__
) = self.encode_prompt(
^^^^^^^^^^^^^^^... | closed | completed | false | 1 | [
"bug"
] | [
"sayakpaul"
] | 2024-07-22T20:39:22Z | 2024-07-24T05:49:31Z | 2024-07-24T05:49:31Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lmxyy | 10,475,754 | MDQ6VXNlcjEwNDc1NzU0 | User | false |
huggingface/diffusers | 2,424,341,057 | I_kwDOHa8MBc6QgIJB | 8,939 | https://github.com/huggingface/diffusers/issues/8939 | https://api.github.com/repos/huggingface/diffusers/issues/8939 | OSError: Can't load tokenizer for 'stabilityai/stable-diffusion-3-medium-diffusers'. If you were trying to load it from 'https://huggingface.co/models', make sure you don't have a local directory with the same name. Otherwise, make sure 'stabilityai/stable-diffusion-3-medium-diffusers' is the correct path to a director... | ### Describe the bug
Hi, I'm trying to fine-tuning stabilityai/stable-diffusion-3-medium-diffusers and using [the official diffuser scripts](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_sd3.py). However it couldn't start. I've logged in into huggingface using a FINEGRAIN... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-07-23T06:12:23Z | 2024-07-23T08:49:25Z | 2024-07-23T08:49:25Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dann2333 | 146,356,127 | U_kgDOCLk3nw | User | false |
huggingface/diffusers | 1,412,822,402 | I_kwDOHa8MBc5UNfWC | 894 | https://github.com/huggingface/diffusers/issues/894 | https://api.github.com/repos/huggingface/diffusers/issues/894 | [Community] Training AutoencoderKL | Hi!
I am working on latent diffusion for audio and music. It seems to me that Diffusers 🧨 is the place to be! There is a feature I would like to request: Training AutoencoderKL (Variational Autoencoder).
What I would love to do, is training AutoencoderKL on square and non-square images, either with one or more t... | closed | completed | false | 24 | [
"community-examples",
"stale"
] | [] | 2022-10-18T08:49:36Z | 2025-05-02T06:39:40Z | 2022-11-28T15:03:31Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AI-Guru | 32,195,399 | MDQ6VXNlcjMyMTk1Mzk5 | User | false |
huggingface/diffusers | 2,424,909,815 | I_kwDOHa8MBc6QiS_3 | 8,943 | https://github.com/huggingface/diffusers/issues/8943 | https://api.github.com/repos/huggingface/diffusers/issues/8943 | libtorch_cuda_cu.so: cannot open shared object file, when installing diffusers of main branch | ### Describe the bug
Bug log: libtorch_cuda_cu.so: cannot open shared object file, when installing diffusers of main branch.
### Reproduction
pip install -e ".[torch]"
```python
from diffusers import StableDiffusion3InpaintPipeline
```
### Logs
_No response_
### System Info
python3.9
torch ... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-07-23T11:01:15Z | 2024-07-24T03:01:58Z | 2024-07-24T03:01:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | huangjun12 | 22,365,664 | MDQ6VXNlcjIyMzY1NjY0 | User | false |
huggingface/diffusers | 1,412,854,254 | I_kwDOHa8MBc5UNnHu | 895 | https://github.com/huggingface/diffusers/issues/895 | https://api.github.com/repos/huggingface/diffusers/issues/895 | [Community Pipeline] Imagic: Text-Based Real Image Editing with Diffusion Models | ## Intro
Community Pipelines are introduced in `diffusers==0.4.0` with the idea of allowing the community to quickly add, integrate, and share their custom pipelines on top of `diffusers`.
You can find a guide about Community Pipelines [here](https://github.com/huggingface/diffusers/issues/841). You can also fin... | closed | completed | false | 18 | [
"good first issue",
"community-examples"
] | [] | 2022-10-18T09:11:34Z | 2023-10-01T07:41:47Z | 2022-11-11T08:33:57Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | apolinario | 788,417 | MDQ6VXNlcjc4ODQxNw== | User | false |
huggingface/diffusers | 2,426,478,814 | I_kwDOHa8MBc6QoSDe | 8,950 | https://github.com/huggingface/diffusers/issues/8950 | https://api.github.com/repos/huggingface/diffusers/issues/8950 | [SD3] bug in SD3 LoRA Training | ### Describe the bug
Codes here delete `tokenizers` and `text_encoders` which would be used later for `train_dataset.custom_instance_prompts`
https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_sd3.py#L1457-L1463
The if-else logic here needs to be adjusted to prevent error ... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-07-24T02:58:41Z | 2024-08-19T01:37:42Z | 2024-08-19T01:37:42Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | townwish4git | 143,256,262 | U_kgDOCInqxg | User | false |
huggingface/diffusers | 2,426,689,262 | I_kwDOHa8MBc6QpFbu | 8,953 | https://github.com/huggingface/diffusers/issues/8953 | https://api.github.com/repos/huggingface/diffusers/issues/8953 | Why loading a lora weights so low? | I used diffusers to load lora weights but it much slow to finish.
diffusers version: 0.29.2
I test another version of diffusers 0.23.0 without peft installation, and the time is decent.
```
t1 = time.time()
pipe.load_lora_weights("/data/**/lora_weights/lcm-lora-sdxl/", weight_name="pytorch_lora_weights.safeten... | closed | completed | false | 18 | [
"peft"
] | [] | 2024-07-24T06:16:42Z | 2024-10-15T15:23:34Z | 2024-10-15T15:23:33Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zengjie617789 | 25,050,291 | MDQ6VXNlcjI1MDUwMjkx | User | false |
huggingface/diffusers | 2,426,933,327 | I_kwDOHa8MBc6QqBBP | 8,955 | https://github.com/huggingface/diffusers/issues/8955 | https://api.github.com/repos/huggingface/diffusers/issues/8955 | RuntimeError: Error(s) in loading state_dict for CLIPTextModelWithProjection | ### Describe the bug
Hi, I'm trying to fine-tuning stabilityai/stable-diffusion-3-medium-diffusers and using [the official diffuser scripts](https://github.com/huggingface/diffusers/blob/main/examples/dreambooth/train_dreambooth_lora_sd3.py). The train process was normal expect the loss cannot reduce. I hope to add a ... | closed | completed | false | 11 | [
"bug",
"stale"
] | [] | 2024-07-24T08:35:14Z | 2024-12-16T18:52:16Z | 2024-12-16T18:52:16Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | dann2333 | 146,356,127 | U_kgDOCLk3nw | User | false |
huggingface/diffusers | 2,427,089,396 | I_kwDOHa8MBc6QqnH0 | 8,956 | https://github.com/huggingface/diffusers/issues/8956 | https://api.github.com/repos/huggingface/diffusers/issues/8956 | OVStableDiffusionXLPipeline can't load model | ### Describe the bug
```
Framework not specified. Using pt to export the model.
Keyword arguments {'subfolder': '', 'trust_remote_code': False} are not expected by StableDiffusionXLImg2ImgPipeline and will be ignored.
Loading pipeline components...: 29%|████████████████████████████████▊ ... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-07-24T09:44:16Z | 2024-07-24T14:08:39Z | 2024-07-24T14:08:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | 631068264 | 8,144,089 | MDQ6VXNlcjgxNDQwODk= | User | false |
huggingface/diffusers | 2,428,336,533 | I_kwDOHa8MBc6QvXmV | 8,957 | https://github.com/huggingface/diffusers/issues/8957 | https://api.github.com/repos/huggingface/diffusers/issues/8957 | StableDiffusionSafetyChecker ignores `attn_implementation` load kwarg | ### Describe the bug
`transformers` added `sdpa` and FA2 for CLIP model in https://github.com/huggingface/transformers/pull/31940. It now initializes the vision model like https://github.com/huggingface/transformers/blob/85a1269e19af022e04bc2aad82572cd5a9e8cdd9/src/transformers/models/clip/modeling_clip.py#L1143.
... | closed | completed | false | 8 | [
"bug",
"help wanted",
"Good second issue",
"contributions-welcome"
] | [] | 2024-07-24T19:38:23Z | 2024-11-19T21:06:53Z | 2024-11-19T21:06:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jambayk | 94,929,125 | U_kgDOBaiA5Q | User | false |
huggingface/diffusers | 2,428,649,986 | I_kwDOHa8MBc6QwkIC | 8,958 | https://github.com/huggingface/diffusers/issues/8958 | https://api.github.com/repos/huggingface/diffusers/issues/8958 | Pytorch 2.4.0 - module 'torch.library' has no attribute 'register_fake' | ### Describe the bug
When using diffusers with pytorch 2.4.0, there is an error on import
### Reproduction
Install diffusers and then instantiate StableVideoDiffusionXL
```py
from diffusers import StableVideoDiffusionPipeline
import torch
from diffusers import StableVideoDiffusionPipeline
import PIL.Image
... | closed | completed | false | 9 | [
"bug"
] | [] | 2024-07-24T23:47:17Z | 2025-08-09T09:19:45Z | 2024-07-27T00:47:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jonluca | 13,029,040 | MDQ6VXNlcjEzMDI5MDQw | User | false |
huggingface/diffusers | 2,428,810,624 | I_kwDOHa8MBc6QxLWA | 8,960 | https://github.com/huggingface/diffusers/issues/8960 | https://api.github.com/repos/huggingface/diffusers/issues/8960 | The Diffusers Basic Training example's Gradient Accumulation is Nonfunctional | ### Describe the bug
In the example provided here: https://huggingface.co/docs/diffusers/en/tutorials/basic_training
Setting `TrainingConfig.gradient_accumulation_steps` to anything greater than 1 will result in the "unscale_() has already been called on this optimizer since the last update()" error despite acceler... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-07-25T02:20:02Z | 2024-07-25T03:10:54Z | 2024-07-25T03:10:54Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | RandomGamingDev | 83,996,185 | MDQ6VXNlcjgzOTk2MTg1 | User | false |
huggingface/diffusers | 2,429,537,088 | I_kwDOHa8MBc6Qz8tA | 8,968 | https://github.com/huggingface/diffusers/issues/8968 | https://api.github.com/repos/huggingface/diffusers/issues/8968 | pipelines: loading safetensors | ## 1. Problem
When I tried to load `ShapEPipeline` from .safetensors files like
```py
pipe = DiffusionPipeline.from_pretrained(
"openai/shap-e",
torch_dtype=torch.float16,
use_safetensors=True,
revision="refs/pr/11",
variant="fp16",
force_download=True,
)
```
where [revision "refs/... | closed | completed | false | 4 | [] | [] | 2024-07-25T09:54:13Z | 2024-10-31T19:17:12Z | 2024-10-31T19:17:12Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | townwish4git | 143,256,262 | U_kgDOCInqxg | User | false |
huggingface/diffusers | 2,429,585,146 | I_kwDOHa8MBc6Q0Ib6 | 8,970 | https://github.com/huggingface/diffusers/issues/8970 | https://api.github.com/repos/huggingface/diffusers/issues/8970 | [Kolors] MPS always return black image | ### Describe the bug
Running Kolors with MPS, the returned result is always black image.
### Reproduction
Run this source code. Output image returns black image
```python
pipe = KolorsPipeline.from_pretrained("Kwai-Kolors/Kolors-diffusers", torch_dtype=torch.float16, variant="fp16")
if torch.cuda.is_available(... | closed | completed | false | 5 | [
"bug"
] | [] | 2024-07-25T10:15:50Z | 2024-08-02T10:57:47Z | 2024-08-02T10:57:47Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bqhuyy | 39,697,900 | MDQ6VXNlcjM5Njk3OTAw | User | false |
huggingface/diffusers | 2,430,665,608 | I_kwDOHa8MBc6Q4QOI | 8,974 | https://github.com/huggingface/diffusers/issues/8974 | https://api.github.com/repos/huggingface/diffusers/issues/8974 | diffusers/scripts/convert_diffusers_to_original_sdxl.py doesn't support models with variant = "fp16" | ### Describe the bug
In an SDXL model, several of the components are stored as `model.fp16.safetensors` rather than `model.safetensors`.
The import script seems to always expect `model.safetensors` and thus produces an error when trying to convert a model with components in format `model.fp16.safetensors`.
###... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-07-25T17:55:23Z | 2025-01-27T01:47:19Z | 2025-01-27T01:47:18Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jslegers | 883,647 | MDQ6VXNlcjg4MzY0Nw== | User | false |
huggingface/diffusers | 2,430,692,026 | I_kwDOHa8MBc6Q4Wq6 | 8,975 | https://github.com/huggingface/diffusers/issues/8975 | https://api.github.com/repos/huggingface/diffusers/issues/8975 | ImportError: cannot import name 'LoraLoaderMixin' from 'diffusers.loaders' | ### Describe the bug
Hello,
I have met an ImportError when I run ```from diffusers.loaders import LoraLoaderMixin```. When I searched for the source code of this class ```LoraLoaderMixin```, I could not find the page. Can anyone help me with that? Thanks!
![Screen Shot 2024-07-25 at 11 05 48 AM](https://github.com... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-07-25T18:09:15Z | 2024-10-18T14:39:18Z | 2024-07-25T20:56:49Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | yj373 | 35,936,254 | MDQ6VXNlcjM1OTM2MjU0 | User | false |
huggingface/diffusers | 2,430,838,162 | I_kwDOHa8MBc6Q46WS | 8,977 | https://github.com/huggingface/diffusers/issues/8977 | https://api.github.com/repos/huggingface/diffusers/issues/8977 | Merge pipelines and/or checkpoints | ### Is your feature request related to a problem? Please describe.
I can't find a way to merge pipelines and/or checkpoints with each other.
### Describe the solution you'd like.
I'd like to merge different Stable Diffusion models using `diffusers`.
AFAIK, `diffusers` doesn't provide that feature yet.
Right... | open | null | false | 19 | [
"stale"
] | [] | 2024-07-25T19:24:47Z | 2024-12-13T15:05:11Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | jslegers | 883,647 | MDQ6VXNlcjg4MzY0Nw== | User | false |
huggingface/diffusers | 1,413,637,719 | I_kwDOHa8MBc5UQmZX | 898 | https://github.com/huggingface/diffusers/issues/898 | https://api.github.com/repos/huggingface/diffusers/issues/898 | [Community] Exception in device=TPU:0: Got unsupported ScalarType BFloat16 | ### Describe the bug
Trying to train Stable Diffusion on TPU with configuring accelerator to use TPU and changing fp16 to bf16. Having the following issue:
```
Exception in device=TPU:0: Got unsupported ScalarType BFloat16
Traceback (most recent call last):
File "/usr/local/lib/python3.8/dist-packages/torch_xla/... | closed | completed | false | 2 | [
"bug",
"help wanted",
"Good second issue"
] | [] | 2022-10-18T18:05:13Z | 2024-11-21T15:26:22Z | 2024-11-21T15:26:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ssusie | 8,882,373 | MDQ6VXNlcjg4ODIzNzM= | User | false |
huggingface/diffusers | 2,431,285,753 | I_kwDOHa8MBc6Q6nn5 | 8,982 | https://github.com/huggingface/diffusers/issues/8982 | https://api.github.com/repos/huggingface/diffusers/issues/8982 | Masked IP Adapter Plus Input | null | closed | completed | false | 0 | [] | [] | 2024-07-26T02:21:54Z | 2024-07-26T03:13:42Z | 2024-07-26T03:13:42Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xiankgx | 4,113,258 | MDQ6VXNlcjQxMTMyNTg= | User | false |
huggingface/diffusers | 2,432,048,545 | I_kwDOHa8MBc6Q9h2h | 8,989 | https://github.com/huggingface/diffusers/issues/8989 | https://api.github.com/repos/huggingface/diffusers/issues/8989 | Sequential offloading bug with Stable Audio | ### Describe the bug
Sequential offloading doesn't work when using `pytest`, but does seem to work outside of tests.
This is an issue, because we can't properly test sequential offloading on Stable Audio.
**Additional information:** I've managed to pinpoint where the first `NaN` appears. It's in the `autoencod... | open | null | false | 5 | [
"bug",
"stale"
] | [] | 2024-07-26T11:38:27Z | 2024-12-13T15:05:07Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ylacombe | 52,246,514 | MDQ6VXNlcjUyMjQ2NTE0 | User | false |
huggingface/diffusers | 2,432,951,581 | I_kwDOHa8MBc6RA-Ud | 8,994 | https://github.com/huggingface/diffusers/issues/8994 | https://api.github.com/repos/huggingface/diffusers/issues/8994 | Generation steps stuck at 0 | ### Describe the bug
When I run my python code everything gets loaded correctly, but the image generation gets stuck at 0 steps and doesn't advance.
### Reproduction
Fresh new Anaconda environment, install the necessary libraries: _pip install diffusers transformers accelerate opencv-python torch torchvision t... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-07-26T20:42:49Z | 2024-07-27T10:01:46Z | 2024-07-27T10:01:46Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Sanfee18 | 73,636,601 | MDQ6VXNlcjczNjM2NjAx | User | false |
huggingface/diffusers | 2,433,775,492 | I_kwDOHa8MBc6REHeE | 8,997 | https://github.com/huggingface/diffusers/issues/8997 | https://api.github.com/repos/huggingface/diffusers/issues/8997 | error in using EDMDPMSolverMultistepScheduler in kolors model | Hi
I want to change kolors default sampler ( euler ) to EDM according [link](https://huggingface.co/Kwai-Kolors/Kolors-diffusers) in below code but my result output image is noise.
```
from diffusers import KolorsPipeline,EDMDPMSolverMultistepScheduler
import torch
pipe = KolorsPipeline.from_pretrained(
"K... | open | null | false | 1 | [
"stale"
] | [] | 2024-07-28T06:29:55Z | 2024-09-14T15:05:10Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | saeedkhanehgir | 65,589,645 | MDQ6VXNlcjY1NTg5NjQ1 | User | false |
huggingface/diffusers | 2,434,591,627 | I_kwDOHa8MBc6RHOuL | 9,001 | https://github.com/huggingface/diffusers/issues/9001 | https://api.github.com/repos/huggingface/diffusers/issues/9001 | Aspect Ratios in Auraflow pipeline | ### Describe the bug
passing Different Height or Width other than 1024 by 1024 leads to an error
### Reproduction
```
from diffusers import AuraFlowPipeline
import torch
pipeline = AuraFlowPipeline.from_pretrained(
"fal/AuraFlow-v0.2",
torch_dtype=torch.float16,
variant="fp16",
).to("cuda")
i... | closed | completed | false | 7 | [
"bug"
] | [] | 2024-07-29T06:46:26Z | 2024-10-15T17:34:41Z | 2024-10-15T17:34:41Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Muawizodux | 88,239,592 | MDQ6VXNlcjg4MjM5NTky | User | false |
huggingface/diffusers | 2,435,085,503 | I_kwDOHa8MBc6RJHS_ | 9,003 | https://github.com/huggingface/diffusers/issues/9003 | https://api.github.com/repos/huggingface/diffusers/issues/9003 | Ability to select the strength of the safety_checker | **Is your feature request related to a problem? Please describe.**
Safety_checker can currently only be turned on/off, and we would like to be able to configure it in more detail.
**Describe alternatives you've considered.**
[I made it based on StableDiffusionPipeline.](https://github.com/suzukimain/diffusers/tr... | open | null | false | 1 | [
"stale"
] | [] | 2024-07-29T10:50:01Z | 2024-09-14T15:05:04Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | suzukimain | 131,413,573 | U_kgDOB9U2RQ | User | false |
huggingface/diffusers | 2,436,139,539 | I_kwDOHa8MBc6RNIoT | 9,007 | https://github.com/huggingface/diffusers/issues/9007 | https://api.github.com/repos/huggingface/diffusers/issues/9007 | Black output in DPMSolverSinglestepScheduler with 3 order | ### Describe the bug
If you run 3n-1 steps(11, 14, ...) generation using `DPMSolverSinglestepScheduler` with `solver_order=3`, it will produce NaN output on final step.
I tried to debug what went wrong and found that in this case executes second order update on zero sigma:
`[singlestep_dpm_solver_second_order_upda... | open | null | false | 1 | [
"bug",
"stale"
] | [] | 2024-07-29T19:05:56Z | 2024-09-14T15:04:58Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | StAlKeR7779 | 7,768,370 | MDQ6VXNlcjc3NjgzNzA= | User | false |
huggingface/diffusers | 2,436,364,975 | I_kwDOHa8MBc6RN_qv | 9,009 | https://github.com/huggingface/diffusers/issues/9009 | https://api.github.com/repos/huggingface/diffusers/issues/9009 | UNET slower by a factor of batch_size | ### Describe the bug
I was expecting to get faster inferences by batching images together. I realized that when I batch 6 images together, the UNET is 5 times slower for a pipeline_controlnet_img2img.py...
Is it possible or normal ? Do I miss anything ? Thanks for your help
### Reproduction
Image dim 1024.... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-07-29T21:01:25Z | 2024-07-30T07:37:51Z | 2024-07-30T07:37:50Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 1,414,487,530 | I_kwDOHa8MBc5UT13q | 901 | https://github.com/huggingface/diffusers/issues/901 | https://api.github.com/repos/huggingface/diffusers/issues/901 | Is there a reference for the model/architecture used by diffusers anywhere? It doesn't seem to match to the original stable-diffusion repo | **What API design would you like to have changed or added to the library? Why?**
Hey there!
I'm not sure if this is the right section to post this, but I have a request/question for a write up on the inference configuration used by diffusers. Similar to a config.yaml in other model repos.
Recently I have bee... | closed | completed | false | 4 | [
"stale"
] | [] | 2022-10-19T07:55:08Z | 2022-11-27T15:03:04Z | 2022-11-27T15:03:04Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tonetechnician | 13,471,797 | MDQ6VXNlcjEzNDcxNzk3 | User | false |
huggingface/diffusers | 2,436,929,616 | I_kwDOHa8MBc6RQJhQ | 9,012 | https://github.com/huggingface/diffusers/issues/9012 | https://api.github.com/repos/huggingface/diffusers/issues/9012 | SD3ControlNetModel.from_transformer load weights error | ### Describe the bug
When specifying `num_layers` in `SD3ControlNetModel.from_transformer`, the function results in an error of unexpected key(s). I think this is because that in `SD3ControlNetModel`, there is only `num_layers` of transformer blocks, while in `SD3Transformer2DModel`, there are 24 transformer blocks. I... | open | null | false | 3 | [
"bug",
"stale"
] | [] | 2024-07-30T05:27:37Z | 2024-09-14T15:04:55Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | FYRichie | 59,409,544 | MDQ6VXNlcjU5NDA5NTQ0 | User | false |
huggingface/diffusers | 1,414,648,614 | I_kwDOHa8MBc5UUdMm | 902 | https://github.com/huggingface/diffusers/issues/902 | https://api.github.com/repos/huggingface/diffusers/issues/902 | Potential regression in deterministic outputs | ### Describe the bug
I've started noticing different outputs ~~in the latest version of diffusers~~ starting from diffusers `0.4.0` when compared against `0.3.0`. This is my test code (extracted from a notebook):
```Python
import diffusers
from diffusers import StableDiffusionPipeline, DDIMScheduler
import tor... | closed | completed | false | 20 | [
"bug",
"stale"
] | [
"patrickvonplaten"
] | 2022-10-19T09:45:42Z | 2023-01-02T15:03:51Z | 2023-01-02T15:03:51Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcuenca | 1,177,582 | MDQ6VXNlcjExNzc1ODI= | User | false |
huggingface/diffusers | 2,439,486,284 | I_kwDOHa8MBc6RZ5tM | 9,023 | https://github.com/huggingface/diffusers/issues/9023 | https://api.github.com/repos/huggingface/diffusers/issues/9023 | Sample code cannot run | ### Describe the bug
Using https://huggingface.co/docs/diffusers/conceptual/evaluation
to measure the clip_score clip_similarity, FID
clip_similarity have problem
It looks the array dimension issue
Change all idx to following
dataset[idx]["image"] to dataset["image"][idx]
also need to take the compatibility ve... | closed | completed | false | 2 | [
"bug",
"stale"
] | [] | 2024-07-31T08:47:49Z | 2024-12-02T20:48:52Z | 2024-12-02T20:48:52Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Jay19751103 | 146,709,562 | U_kgDOCL6cOg | User | false |
huggingface/diffusers | 2,439,733,457 | I_kwDOHa8MBc6Ra2DR | 9,025 | https://github.com/huggingface/diffusers/issues/9025 | https://api.github.com/repos/huggingface/diffusers/issues/9025 | how to use FlaxStableDiffusionPipeline with from_single_file in kaggle tpu vm | ### Describe the bug
I have single safetensors file and work on diffusers.StableDiffusionPipeline.from_single_file
Now I want to use FlaxStableDiffusionPipeline but there are not .from_single_file member function in FlaxStableDiffusionPipeline
I need to
```
pipeline = diffusers.StableDiffusionPipeline.from_single_... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-07-31T10:44:48Z | 2024-08-01T03:59:51Z | 2024-08-01T03:59:51Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,439,760,067 | I_kwDOHa8MBc6Ra8jD | 9,026 | https://github.com/huggingface/diffusers/issues/9026 | https://api.github.com/repos/huggingface/diffusers/issues/9026 | [v0.29.2] TypeError: load_checkpoint_and_dispatch() got an unexpected keyword argument 'force_hook' | ### Describe the bug
An error occurs when setting `device_map='balanced'` in `StableDiffusionPipeline.from_pretrained` to run the model on multiple GPUs. In the latest release, version 0.29.2, the parameter name in the `accelerate.load_checkpoint_and_dispatch` function ([line 822](https://github.com/huggingface/diff... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-07-31T10:59:16Z | 2024-08-01T07:53:48Z | 2024-08-01T07:53:48Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | TsingZ0 | 24,207,672 | MDQ6VXNlcjI0MjA3Njcy | User | false |
huggingface/diffusers | 2,439,784,839 | I_kwDOHa8MBc6RbCmH | 9,027 | https://github.com/huggingface/diffusers/issues/9027 | https://api.github.com/repos/huggingface/diffusers/issues/9027 | consistency in argument names | ### Describe the bug
`tokens` should be `token` for consistency with load method:
https://huggingface.co/docs/diffusers/en/api/loaders/textual_inversion#diffusers.loaders.TextualInversionLoaderMixin.unload_textual_inversion
Load method https://huggingface.co/docs/diffusers/en/api/loaders/textual_inversion#diffuse... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-07-31T11:11:40Z | 2024-11-18T17:11:29Z | 2024-11-18T17:11:29Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | christopher5106 | 6,875,375 | MDQ6VXNlcjY4NzUzNzU= | User | false |
huggingface/diffusers | 2,441,390,140 | I_kwDOHa8MBc6RhKg8 | 9,031 | https://github.com/huggingface/diffusers/issues/9031 | https://api.github.com/repos/huggingface/diffusers/issues/9031 | how to disable safty_checker in FlaxStableDiffusionPipeline | ### Describe the bug
```
! python3 -m pip install -U tensorflow-cpu
import diffusers, os
pipeline = diffusers.StableDiffusionPipeline.from_single_file('https://huggingface.co/chaowenguo/pal/blob/main/chilloutMix-Ni.safetensors')
pipeline.save_pretrained('chilloutMix', safe_serialization=False)
pipeline, params = ... | open | null | false | 4 | [
"bug",
"stale"
] | [] | 2024-08-01T03:48:27Z | 2024-10-13T15:03:54Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,441,402,168 | I_kwDOHa8MBc6RhNc4 | 9,032 | https://github.com/huggingface/diffusers/issues/9032 | https://api.github.com/repos/huggingface/diffusers/issues/9032 | how to get the minimun working example of FlaxStableDiffusionPipeline in google colab with tpu runtime | ### Describe the bug
I try the code in google colab with tpu runtime
```
! python3 -m pip install -U diffusers[flax]
import diffusers, os
pipeline = diffusers.StableDiffusionPipeline.from_single_file('https://huggingface.co/chaowenguo/pal/blob/main/chilloutMix-Ni.safetensors')
pipeline.save_pretrained('chilloutMi... | open | null | false | 4 | [
"bug",
"stale"
] | [] | 2024-08-01T03:58:34Z | 2024-11-04T15:04:13Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,441,690,187 | I_kwDOHa8MBc6RiTxL | 9,034 | https://github.com/huggingface/diffusers/issues/9034 | https://api.github.com/repos/huggingface/diffusers/issues/9034 | AutoPipelineForText2Image requiring sentencepiece, which isn't a diffusers requirement | ### Describe the bug
Importing AutoPipelineForText2Image throws an error if sentencepiece isn't installed, even though sentencepiece is not required for users who just use AutoPipelineForText2Image for SD 1.5 / SD XL and isn't a core diffusers requirement.
### Reproduction
git clone --depth 1 https://github.com/hugg... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-08-01T07:22:59Z | 2024-08-05T18:04:53Z | 2024-08-05T18:04:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | evanscho | 1,657,105 | MDQ6VXNlcjE2NTcxMDU= | User | false |
huggingface/diffusers | 2,442,123,288 | I_kwDOHa8MBc6Rj9gY | 9,038 | https://github.com/huggingface/diffusers/issues/9038 | https://api.github.com/repos/huggingface/diffusers/issues/9038 | how to use prompt weight in FlaxStableDiffusionPipeline | ### Describe the bug
I can see there are prompt_embeds in StableDiffusionPipeline to support Prompt weighting, But how to do that in FlaxStableDiffusionPipeline? there are not prompt_embeds in StableDiffusionPipeline
### Reproduction
N/A
### Logs
_No response_
### System Info
kaggle tpu vm
### Who can help?
_... | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-08-01T10:44:37Z | 2024-10-14T18:25:55Z | 2024-10-14T18:25:55Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,442,201,295 | I_kwDOHa8MBc6RkQjP | 9,039 | https://github.com/huggingface/diffusers/issues/9039 | https://api.github.com/repos/huggingface/diffusers/issues/9039 | how to load_lora_weights in FlaxStableDiffusionPipeline | ### Describe the bug
how to load lora in FlaxStableDiffusionPipeline, there are no load_lora_weights in FlaxStableDiffusionPipeline
### Reproduction
N/A
### Logs
_No response_
### System Info
kaggle tpu vm
### Who can help?
_No response_ | closed | completed | false | 5 | [
"bug",
"stale"
] | [] | 2024-08-01T11:23:52Z | 2024-10-15T03:23:54Z | 2024-10-15T03:23:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | ghost | 10,137 | MDQ6VXNlcjEwMTM3 | User | false |
huggingface/diffusers | 2,442,598,255 | I_kwDOHa8MBc6Rlxdv | 9,042 | https://github.com/huggingface/diffusers/issues/9042 | https://api.github.com/repos/huggingface/diffusers/issues/9042 | Add Flux pipeline (Black Forest Labs) | ### Model/Pipeline/Scheduler description
Blog: https://blackforestlabs.ai/announcing-black-forest-labs/
Implementation: https://github.com/black-forest-labs
Weights: https://huggingface.co/black-forest-labs
Seems like it's just "bigger SD3".
### Open source status
- [X] The model implementation is availab... | closed | completed | false | 4 | [] | [] | 2024-08-01T14:15:43Z | 2024-08-02T07:59:58Z | 2024-08-02T07:07:22Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AmericanPresidentJimmyCarter | 110,263,573 | U_kgDOBpJ9FQ | User | false |
huggingface/diffusers | 2,442,707,163 | I_kwDOHa8MBc6RmMDb | 9,044 | https://github.com/huggingface/diffusers/issues/9044 | https://api.github.com/repos/huggingface/diffusers/issues/9044 | OSError: stabilityai/stable-diffusion-xl-base-1.0 does not appear to have a file named diffusion_pytorch_model.bin. | ### Describe the bug
I use lora to fintune the model of 'stabilityai/stable-diffusion-xl-base-1.0'，get fellow error:
OSError: stabilityai/stable-diffusion-xl-base-1.0 does not appear to have a file named diffusion_pytorch_model.bin.
### Reproduction
python train_text_to_image_lora_sdxl.py --pretrained_model_name_or... | open | null | false | 2 | [
"bug",
"stale"
] | [] | 2024-08-01T15:03:38Z | 2024-09-14T15:04:24Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | limouren2000 | 42,027,714 | MDQ6VXNlcjQyMDI3NzE0 | User | false |
huggingface/diffusers | 2,443,333,410 | I_kwDOHa8MBc6Rok8i | 9,045 | https://github.com/huggingface/diffusers/issues/9045 | https://api.github.com/repos/huggingface/diffusers/issues/9045 | StableDiffusio3Pipeline cannot convert to CUDA device with accelerate & peft | ### Describe the bug
`accelerate` with `peft` will load the sd3 pipeline on device `meta` which cannot be moved to device `cuda` later like in the official sd3 blog [post](https://huggingface.co/docs/diffusers/main/en/api/pipelines/stable_diffusion/stable_diffusion_3). Uninstalling `accelerate` & `peft` solve the is... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-01T20:15:05Z | 2024-09-05T18:30:46Z | 2024-09-05T18:30:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | shingjan | 11,846,349 | MDQ6VXNlcjExODQ2MzQ5 | User | false |
huggingface/diffusers | 2,443,765,330 | I_kwDOHa8MBc6RqOZS | 9,046 | https://github.com/huggingface/diffusers/issues/9046 | https://api.github.com/repos/huggingface/diffusers/issues/9046 | Flux VAE latent_channels is 64？ | ### Describe the bug
<img width="689" alt="image" src="https://github.com/user-attachments/assets/fff2b2c8-6b6a-4e10-ba4f-514a2c7bd38d">
After converting the VAE using scripts/convert_flux_to_diffusers.py, it seems that the corresponding channels are not aligned? It still looks like 16. But the pipeline is fine wit... | closed | completed | false | 1 | [
"bug"
] | [] | 2024-08-02T01:02:15Z | 2024-08-02T01:34:40Z | 2024-08-02T01:34:12Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | wangqixun | 46,553,287 | MDQ6VXNlcjQ2NTUzMjg3 | User | false |
huggingface/diffusers | 2,443,792,057 | I_kwDOHa8MBc6RqU65 | 9,047 | https://github.com/huggingface/diffusers/issues/9047 | https://api.github.com/repos/huggingface/diffusers/issues/9047 | flux does not work on MPS devices | ### Describe the bug
```
scale = torch.arange(0, dim, 2, dtype=torch.float64, device=pos.device) / dim
TypeError: Cannot convert a MPS Tensor to float64 dtype as the MPS framework doesn't support float64. Please use float32 instead.
```
### Reproduction
```py
import torch
from diffusers import FluxPi... | closed | completed | false | 41 | [
"bug",
"stale"
] | [] | 2024-08-02T01:36:32Z | 2024-09-20T14:24:37Z | 2024-09-16T15:06:03Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 1,414,888,418 | I_kwDOHa8MBc5UVXvi | 905 | https://github.com/huggingface/diffusers/issues/905 | https://api.github.com/repos/huggingface/diffusers/issues/905 | [Community Pipeline] image2image inpainting? | **Is your feature request related to a problem? Please describe.**
Imagine you have two images defined, a logo and a t-shirt. we want to apply the logo to the t-shirt in such a way that it matches the physical surface of the t-shirt and the logo deforms correctly.
**Describe the solution you'd like**
How it shou... | closed | completed | false | 9 | [
"community-examples"
] | [] | 2022-10-19T12:30:41Z | 2022-11-07T20:06:53Z | 2022-11-07T20:06:53Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | batrlatom | 11,804,368 | MDQ6VXNlcjExODA0MzY4 | User | false |
huggingface/diffusers | 2,444,085,655 | I_kwDOHa8MBc6RrcmX | 9,051 | https://github.com/huggingface/diffusers/issues/9051 | https://api.github.com/repos/huggingface/diffusers/issues/9051 | `FluxPipeline` does not support batch infernce | ### Describe the bug
The `FLUX.1-dev` model does not support batch inference. When I input two prompts, the inference raised some errors.
### Reproduction
```python
import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", revision="refs/pr/3", tor... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-02T05:29:02Z | 2024-08-02T12:41:17Z | 2024-08-02T12:41:17Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | lmxyy | 10,475,754 | MDQ6VXNlcjEwNDc1NzU0 | User | false |
huggingface/diffusers | 2,444,187,444 | I_kwDOHa8MBc6Rr1c0 | 9,052 | https://github.com/huggingface/diffusers/issues/9052 | https://api.github.com/repos/huggingface/diffusers/issues/9052 | ```from_single_file``` fails on certain civit checkpoints. | ### Describe the bug
Trying to load [this](https://civitai.com/models/403751/dreamshaper-xl-lightning-inpainting) or this [this](https://civitai.com/api/download/models/456538?type=Model&format=SafeTensor&size=pruned&fp=fp16) results in a shape mismatch error with key sizes (640,640) and (640,640,1,1).
### Reproduct... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-02T06:38:33Z | 2024-08-20T06:39:03Z | 2024-08-20T06:39:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Gothos | 95,531,133 | U_kgDOBbGwfQ | User | false |
huggingface/diffusers | 2,444,190,466 | I_kwDOHa8MBc6Rr2MC | 9,053 | https://github.com/huggingface/diffusers/issues/9053 | https://api.github.com/repos/huggingface/diffusers/issues/9053 | Flux-fp8 needs "from_single_file" support in the FluxPipeline | The FLUX weights can be used on 12 GB using these pf8 weights: https://huggingface.co/Kijai/flux-fp8
However .from_single_file is not supported in the FluxPipeline, and afaik can't these weights be used with Diffusers in any other way?
@sayakpaul | closed | completed | false | 57 | [] | [] | 2024-08-02T06:40:12Z | 2024-10-18T07:35:05Z | 2024-08-09T02:19:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | tin2tin | 1,322,593 | MDQ6VXNlcjEzMjI1OTM= | User | false |
huggingface/diffusers | 2,444,336,297 | I_kwDOHa8MBc6RsZyp | 9,055 | https://github.com/huggingface/diffusers/issues/9055 | https://api.github.com/repos/huggingface/diffusers/issues/9055 | ImportError: cannot import name 'StableDiffusionLoraLoaderMixin' from 'diffusers.loaders' | ### Describe the bug
I get this error in diffusers versions 25,26,27,28,29, how can I solve it?
### Reproduction
import ast
import gc
import inspect
import math
import warnings
from collections.abc import Iterable
from typing import Any, Callable, Dict, List, Optional, Union
import torch
import torch.nn.... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-02T07:58:16Z | 2024-08-02T09:32:12Z | 2024-08-02T09:32:12Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MehmetcanTozlu | 65,121,799 | MDQ6VXNlcjY1MTIxNzk5 | User | false |
huggingface/diffusers | 2,444,427,575 | I_kwDOHa8MBc6RswE3 | 9,056 | https://github.com/huggingface/diffusers/issues/9056 | https://api.github.com/repos/huggingface/diffusers/issues/9056 | The density_for_timestep_sampling and loss_weighting for SD3 Training！！！ | Thanks to Rafie Walker's code we can try to train SD3 models with flow-matching!
But some places don't seem to match what's in the paper.
Rafie Walker's code is below:
```
def compute_density_for_timestep_sampling(
weighting_scheme: str, batch_size: int, logit_mean: float = None, logit_std: float = None, mode... | open | null | false | 17 | [
"stale"
] | [] | 2024-08-02T08:47:30Z | 2025-06-10T06:20:41Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DidiD1 | 107,545,725 | U_kgDOBmkEfQ | User | false |
huggingface/diffusers | 2,444,656,461 | I_kwDOHa8MBc6Rtn9N | 9,058 | https://github.com/huggingface/diffusers/issues/9058 | https://api.github.com/repos/huggingface/diffusers/issues/9058 | Support loading quantized models from quanto directly | With more and more and larger and larger diffusion transformer models coming up, I think it makes sense to support the direct loading of quantized models from quanto.
Currently, the workflow to load a quantized model with quanto is simple:
```python
from diffusers import DiffusionPipeline
from optimum.quanto i... | closed | completed | false | 2 | [
"stale"
] | [] | 2024-08-02T10:26:51Z | 2024-09-26T04:24:03Z | 2024-09-26T04:24:02Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,444,796,584 | I_kwDOHa8MBc6RuKKo | 9,059 | https://github.com/huggingface/diffusers/issues/9059 | https://api.github.com/repos/huggingface/diffusers/issues/9059 | Remove debug snippet | Is this a debug snippet? It's more clean if we remove it.
https://github.com/huggingface/diffusers/blob/b1f43d71897ad2c73cb891d2e92d23bc7d46a4be/src/diffusers/models/transformers/transformer_flux.py#L376 | closed | completed | false | 0 | [] | [] | 2024-08-02T11:53:11Z | 2024-08-03T06:17:16Z | 2024-08-03T06:17:16Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | haofanwang | 18,741,068 | MDQ6VXNlcjE4NzQxMDY4 | User | false |
huggingface/diffusers | 2,445,237,843 | I_kwDOHa8MBc6Rv15T | 9,062 | https://github.com/huggingface/diffusers/issues/9062 | https://api.github.com/repos/huggingface/diffusers/issues/9062 | cannot import name 'StableAudioPipeline' from 'diffusers' | ### Describe the bug
Hi, I'm getting an error cannot import name 'StableAudioPipeline' from 'diffusers' (c:\Users\Garin\anaconda3\envs\DiffusersEdit1.1\lib\site-packages\diffusers_init_.py). Have only just installed diffusers so should be up to date.
I checked in the init.py and it's not listed there so tried to add ... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-02T15:33:34Z | 2024-10-20T22:59:07Z | 2024-10-20T22:58:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Gcurtis95 | 78,544,287 | MDQ6VXNlcjc4NTQ0Mjg3 | User | false |
huggingface/diffusers | 2,445,836,399 | I_kwDOHa8MBc6RyIBv | 9,063 | https://github.com/huggingface/diffusers/issues/9063 | https://api.github.com/repos/huggingface/diffusers/issues/9063 | Error creating pre-trained FluxPipeline with diffusers "Cannot instantiate this tokenizer from a slow version" | ### Describe the bug
I'm receiving an error trying to use the `diffusers` module to run the `flux` model.
> ValueError: Cannot instantiate this tokenizer from a slow version. If it's based on sentencepiece, make sure you have sentencepiece installed
https://huggingface.co/black-forest-labs/FLUX.1-schnell
### Re... | closed | completed | false | 22 | [
"bug",
"stale"
] | [] | 2024-08-02T22:26:16Z | 2025-08-26T05:27:30Z | 2025-08-26T05:27:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | pcgeek86 | 466,713 | MDQ6VXNlcjQ2NjcxMw== | User | false |
huggingface/diffusers | 2,446,038,158 | I_kwDOHa8MBc6Ry5SO | 9,067 | https://github.com/huggingface/diffusers/issues/9067 | https://api.github.com/repos/huggingface/diffusers/issues/9067 | No module named 'diffusers.hub_utils' | ### Describe the bug
Training with Diffusers
Installing the dependencies
%%capture
!pip install git+https://github.com/huggingface/diffusers.git#egg=diffusers[training]
!pip install accelerate
! pip install datasets
from accelerate import Accelerator
from diffusers.hub_utils import init_git... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-03T03:48:46Z | 2024-08-07T14:09:44Z | 2024-08-07T14:09:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | elliotthwang | 22,218,711 | MDQ6VXNlcjIyMjE4NzEx | User | false |
huggingface/diffusers | 2,446,338,352 | I_kwDOHa8MBc6R0Ckw | 9,069 | https://github.com/huggingface/diffusers/issues/9069 | https://api.github.com/repos/huggingface/diffusers/issues/9069 | TypeError: expected np.ndarray (got numpy.ndarray) | ### Describe the bug
```
import torch
from diffusers import FluxPipeline
pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)
pipe.to("cuda")
prompt = "A cat holding a sign that says hello world"
# Depending on the variant being used, the pipeline call will slightly ... | closed | completed | false | 11 | [
"bug"
] | [] | 2024-08-03T12:45:03Z | 2024-10-27T06:43:32Z | 2024-10-15T21:06:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | xiangyumou | 60,832,304 | MDQ6VXNlcjYwODMyMzA0 | User | false |
huggingface/diffusers | 2,447,073,600 | I_kwDOHa8MBc6R22FA | 9,076 | https://github.com/huggingface/diffusers/issues/9076 | https://api.github.com/repos/huggingface/diffusers/issues/9076 | Add a better version of 'callback_on_step_end' for FluxPipeline | **Is your feature request related to a problem? Please describe.**
There is a huge delay before starting the inference and once the 4th step is complete and there is no callback for that and it feels like it is stuck, just want a more responsive version.
```
prompt = "A cat holding a sign that says hello world"
ima... | closed | completed | false | 3 | [
"stale"
] | [] | 2024-08-04T10:34:04Z | 2024-11-23T00:24:14Z | 2024-11-23T00:24:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nayan-dhabarde | 24,761,808 | MDQ6VXNlcjI0NzYxODA4 | User | false |
huggingface/diffusers | 2,447,217,344 | I_kwDOHa8MBc6R3ZLA | 9,078 | https://github.com/huggingface/diffusers/issues/9078 | https://api.github.com/repos/huggingface/diffusers/issues/9078 | Flux Can't Load LoRA | I'm getting this error attempting to load lora for flux trained with SimpleTuner:
`AttributeError: 'FluxPipeline' object has no attribute 'load_adapter'`
Desperate to get this working
```
import torch
from diffusers import DiffusionPipeline
# Load the pre-trained model and adapter
model_id = 'black-forest-labs... | closed | completed | false | 17 | [] | [] | 2024-08-04T16:58:27Z | 2025-04-29T09:29:28Z | 2024-08-05T04:13:54Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | markuryy | 20,881,659 | MDQ6VXNlcjIwODgxNjU5 | User | false |
huggingface/diffusers | 2,447,619,511 | I_kwDOHa8MBc6R47W3 | 9,081 | https://github.com/huggingface/diffusers/issues/9081 | https://api.github.com/repos/huggingface/diffusers/issues/9081 | add ControlNet support for Flux.1 | Flux has been released almost a week ago (which is like 1 year in AI time) and appears to be the new SOTA for open txt2img! Please add ControlNet support. Pretty please! | closed | completed | false | 4 | [] | [] | 2024-08-05T05:17:13Z | 2024-08-17T09:55:42Z | 2024-08-17T09:53:59Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | geroldmeisinger | 112,266,044 | U_kgDOBrELPA | User | false |
huggingface/diffusers | 2,448,098,113 | I_kwDOHa8MBc6R6wNB | 9,084 | https://github.com/huggingface/diffusers/issues/9084 | https://api.github.com/repos/huggingface/diffusers/issues/9084 | SD3 inpaint pipeline can not use EulerA scheduler. bug: The `scale_model_input` function should be called before `step` to ensure correct denoising. | ### Describe the bug
SD3 inpaint pipeline can not use EulerA scheduler.
### Reproduction
```python
pipe = StableDiffusion3InpaintPipeline.from_pretrained(xxx)
pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
# then run the pipe
out_images = pipe(
prompt=caption,
... | open | null | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-05T09:41:08Z | 2024-09-14T15:04:01Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | huangjun12 | 22,365,664 | MDQ6VXNlcjIyMzY1NjY0 | User | false |
huggingface/diffusers | 2,448,102,780 | I_kwDOHa8MBc6R6xV8 | 9,085 | https://github.com/huggingface/diffusers/issues/9085 | https://api.github.com/repos/huggingface/diffusers/issues/9085 | controlnet training script for flux | controlnet training script for flux | closed | completed | false | 13 | [] | [] | 2024-08-05T09:43:09Z | 2024-11-10T17:56:16Z | 2024-10-15T15:05:44Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | universewill | 6,790,730 | MDQ6VXNlcjY3OTA3MzA= | User | false |
huggingface/diffusers | 2,449,499,503 | I_kwDOHa8MBc6SAGVv | 9,088 | https://github.com/huggingface/diffusers/issues/9088 | https://api.github.com/repos/huggingface/diffusers/issues/9088 | Script to export Flux to diffusers. | ### Describe the bug
Hello,
I am trying to cast Flux model on FP8, as the original version of Flux need to be converted in order to be used by diffusers i tried to use the script provided but i have an error. It's all good with the transformer, i succeeded to export it, but i have an issue with VAE. Tried sevral t... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-05T21:19:12Z | 2024-08-06T20:18:45Z | 2024-08-06T20:18:45Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | mrrfr | 102,073,980 | U_kgDOBhWGfA | User | false |
huggingface/diffusers | 2,449,541,386 | I_kwDOHa8MBc6SAQkK | 9,089 | https://github.com/huggingface/diffusers/issues/9089 | https://api.github.com/repos/huggingface/diffusers/issues/9089 | FluxPipeline not found in diffusers | ### Describe the bug
I am trying: https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux
and: https://huggingface.co/black-forest-labs/FLUX.1-dev
Either: !pip install -U diffusers
and
!pip install git+https://github.com/huggingface/diffusers.git
However:
```
--------------------------------------... | closed | completed | false | 4 | [
"bug"
] | [] | 2024-08-05T21:52:04Z | 2024-08-08T21:28:58Z | 2024-08-08T21:28:58Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Twenkid | 23,367,640 | MDQ6VXNlcjIzMzY3NjQw | User | false |
huggingface/diffusers | 2,449,634,918 | I_kwDOHa8MBc6SAnZm | 9,090 | https://github.com/huggingface/diffusers/issues/9090 | https://api.github.com/repos/huggingface/diffusers/issues/9090 | PNDMScheduler Inference Timesteps Bug => Repeated Inference Timestep | ### Describe the bug
```python3
from diffusers import PNDMScheduler
scheduler = PNDMScheduler.from_pretrained('runwayml/stable-diffusion-v1-5', subfolder="scheduler")
scheduler.set_timesteps(20)
print(scheduler.timesteps)
# there are 21 timesteps with repeated step
# tensor([951, 901, 901, 851, 801, 751, 701... | open | null | false | 2 | [
"bug",
"stale"
] | [
"yiyixuxu"
] | 2024-08-05T23:23:50Z | 2024-09-14T15:03:55Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | liming-ai | 73,068,772 | MDQ6VXNlcjczMDY4Nzcy | User | false |
huggingface/diffusers | 2,449,637,444 | I_kwDOHa8MBc6SAoBE | 9,091 | https://github.com/huggingface/diffusers/issues/9091 | https://api.github.com/repos/huggingface/diffusers/issues/9091 | Torch Compile support for runwayml/stable-diffusion-v1-5 VAE Decoder | ### Describe the bug
I'd like to run the runwayml/stable-diffusion-v1-5 pipeline using Torch Compile in the reduce-overhead mode. However, the optimized VAE decoder model complains that Torch is not found. I've attached a script to reproduce the error in the logs.
An initial debugging on my end revealed that th... | closed | completed | false | 10 | [
"bug"
] | [] | 2024-08-05T23:26:41Z | 2024-09-20T22:45:40Z | 2024-09-20T22:45:40Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | asfiyab-nvidia | 117,682,710 | U_kgDOBwOyFg | User | false |
huggingface/diffusers | 2,449,762,913 | I_kwDOHa8MBc6SBGph | 9,092 | https://github.com/huggingface/diffusers/issues/9092 | https://api.github.com/repos/huggingface/diffusers/issues/9092 | Fluxpipeline report model_index.json not found | ### Describe the bug
I use the Fluxpipeline and report no file model_index.json.
I read other issue and set the `revision="refs/pr/3"`,but it doesn't work, how can i do to solve this problem and how to use the T5xxl as text encoder? thanks for your help
### Reproduction
```
import torch
from diffusers impor... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-06T01:48:40Z | 2024-08-06T02:25:03Z | 2024-08-06T01:51:50Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chongxian | 18,145,013 | MDQ6VXNlcjE4MTQ1MDEz | User | false |
huggingface/diffusers | 2,450,042,064 | I_kwDOHa8MBc6SCKzQ | 9,095 | https://github.com/huggingface/diffusers/issues/9095 | https://api.github.com/repos/huggingface/diffusers/issues/9095 | FLUX's inference speed is so slow | I run FLUX1.-dev on V100-32G GPU card, the inference code is like:
`
pipe = FluxPipeline.from_pretrained("checkpoints/FLUX.1-dev",
torch_dtype=torch.bfloat16,
low_cpu_mem_usage=True)
pipe.enable_model_cpu_offload()
image = pipe(
p... | closed | completed | false | 7 | [] | [] | 2024-08-06T06:20:20Z | 2024-10-29T08:17:01Z | 2024-10-29T08:17:01Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | chenbinghui1 | 11,517,207 | MDQ6VXNlcjExNTE3MjA3 | User | false |
huggingface/diffusers | 2,450,047,597 | I_kwDOHa8MBc6SCMJt | 9,096 | https://github.com/huggingface/diffusers/issues/9096 | https://api.github.com/repos/huggingface/diffusers/issues/9096 | Flux inference with torch.half outputs NaN values | ### Describe the bug
When running Flux inference with torch.half (i.e. for faster inference on Turing GPUs), there are NaN outputs in the resulting image (will appear as black in matplotlib).
### Reproduction
```python
from diffusers import FluxPipeline
import matplotlib.pyplot as plt
import torch
torch.backends... | closed | completed | false | 0 | [
"bug"
] | [] | 2024-08-06T06:23:48Z | 2024-08-07T05:24:21Z | 2024-08-07T05:24:21Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | latentCall145 | 75,278,258 | MDQ6VXNlcjc1Mjc4MjU4 | User | false |
huggingface/diffusers | 2,450,529,210 | I_kwDOHa8MBc6SEBu6 | 9,098 | https://github.com/huggingface/diffusers/issues/9098 | https://api.github.com/repos/huggingface/diffusers/issues/9098 | Speed up FLUX inferences | Hey guys what do you think are the best ways to speed up flux inferences?
| closed | completed | false | 2 | [
"stale"
] | [] | 2024-08-06T10:29:30Z | 2024-10-31T19:14:39Z | 2024-10-31T19:14:39Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Swarzox | 16,492,753 | MDQ6VXNlcjE2NDkyNzUz | User | false |
huggingface/diffusers | 2,451,219,873 | I_kwDOHa8MBc6SGqWh | 9,101 | https://github.com/huggingface/diffusers/issues/9101 | https://api.github.com/repos/huggingface/diffusers/issues/9101 | InstructPix2Pix training script for SD3 | **Is your feature request related to a problem? Please describe.**
It would be awesome to see a training script for InstructPix2Pix based on Stable Diffusion 3 such as [this](https://github.com/huggingface/diffusers/blob/main/examples/instruct_pix2pix/train_instruct_pix2pix_sdxl.py).
**Describe the solution you'd l... | open | null | false | 22 | [
"stale"
] | [] | 2024-08-06T15:56:06Z | 2025-06-04T00:29:14Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | raghavaggsss | 22,764,471 | MDQ6VXNlcjIyNzY0NDcx | User | false |
huggingface/diffusers | 2,452,115,721 | I_kwDOHa8MBc6SKFEJ | 9,106 | https://github.com/huggingface/diffusers/issues/9106 | https://api.github.com/repos/huggingface/diffusers/issues/9106 | No Attributte: when i run pixart-alpha with time_step=1, error occured. | ### Describe the bug
File "/home/luor/miniconda3/envs/aigc-inf/lib/python3.9/site-packages/diffusers/pipelines/pixart_alpha/pipeline_pixart_alpha.py", line 942, in __call__
latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs).pred_original_sample
AttributeError: 'SchedulerOutput' object h... | open | null | false | 3 | [
"bug",
"stale"
] | [] | 2024-08-07T02:42:15Z | 2024-09-14T15:03:39Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MaltoseFlower | 57,850,152 | MDQ6VXNlcjU3ODUwMTUy | User | false |
huggingface/diffusers | 2,452,330,934 | I_kwDOHa8MBc6SK5m2 | 9,108 | https://github.com/huggingface/diffusers/issues/9108 | https://api.github.com/repos/huggingface/diffusers/issues/9108 | can't resume training with quantized base model | ### Describe the bug
We have a KeyError when the state dict goes to load into the transformer.
### Reproduction
```py
import torch
from diffusers.models import FluxTransformer2DModel
from peft import LoraConfig, set_peft_model_state_dict
from diffusers import FluxPipeline
from diffusers.utils import conve... | open | null | false | 45 | [
"bug",
"stale"
] | [] | 2024-08-07T04:25:34Z | 2024-09-16T15:03:11Z | null | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | bghira | 59,658,056 | MDQ6VXNlcjU5NjU4MDU2 | User | false |
huggingface/diffusers | 2,452,455,079 | I_kwDOHa8MBc6SLX6n | 9,109 | https://github.com/huggingface/diffusers/issues/9109 | https://api.github.com/repos/huggingface/diffusers/issues/9109 | operator torchvision::nms does not exist | ### Describe the bug
```
---------------------------------------------------------------------------
RuntimeError Traceback (most recent call last)
[/usr/local/lib/python3.10/dist-packages/transformers/utils/import_utils.py](https://localhost:8080/#) in _get_module(self, module_name)
... | closed | completed | false | 6 | [
"bug",
"stale"
] | [] | 2024-08-07T05:18:49Z | 2025-04-17T13:35:40Z | 2024-09-28T03:31:30Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | andysingal | 20,493,493 | MDQ6VXNlcjIwNDkzNDkz | User | false |
huggingface/diffusers | 1,415,154,275 | I_kwDOHa8MBc5UWYpj | 911 | https://github.com/huggingface/diffusers/issues/911 | https://api.github.com/repos/huggingface/diffusers/issues/911 | Dreambooth started to return bad config | ### Describe the bug
Hello,
In current master whenever I train dreambooth it returns model's folder that cannot be seamlessly utilized by `StableDiffusionPipeline.from_pretrained`. Previous training worked good.
Now command on the pre-trained model folder results in
```
ValueError: Pipeline <class 'diffusers... | closed | completed | false | 4 | [
"bug"
] | [] | 2022-10-19T15:11:50Z | 2022-10-19T17:09:09Z | 2022-10-19T17:09:09Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | zetyquickly | 25,350,960 | MDQ6VXNlcjI1MzUwOTYw | User | false |
huggingface/diffusers | 2,452,915,548 | I_kwDOHa8MBc6SNIVc | 9,111 | https://github.com/huggingface/diffusers/issues/9111 | https://api.github.com/repos/huggingface/diffusers/issues/9111 | 'BitDiffusion' object has no attribute 'scheduler' | ### Describe the bug
Running the [Bit Diffusion](https://github.com/huggingface/diffusers/blob/main/examples/community/README.md#bit-diffusion) example code from the Community Pipeline Examples fails
### Reproduction
```
from diffusers import DiffusionPipeline
pipe = DiffusionPipeline.from_pretrained("google... | open | null | false | 3 | [
"bug",
"stale"
] | [
"a-r-r-o-w"
] | 2024-08-07T08:41:45Z | 2024-11-25T15:04:56Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | BioGeek | 59,344 | MDQ6VXNlcjU5MzQ0 | User | false |
huggingface/diffusers | 2,453,102,499 | I_kwDOHa8MBc6SN1-j | 9,112 | https://github.com/huggingface/diffusers/issues/9112 | https://api.github.com/repos/huggingface/diffusers/issues/9112 | Auraflow inference error: The size of tensor a (264) must match the size of tensor b (4096) at non-singleton dimension 1 | ### Describe the bug
I am using the auraflow piepline and keep getting the same error as described.
### Reproduction
`import os
import torch
from PIL import Image
from diffusers import AuraFlowPipeline
# Set the device
device = "cuda" if torch.cuda.is_available() else "cpu"
# Load the model
model = AuraFl... | closed | completed | false | 2 | [
"bug"
] | [] | 2024-08-07T10:10:27Z | 2024-08-08T07:37:13Z | 2024-08-08T07:37:13Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | anushka0415 | 84,956,432 | MDQ6VXNlcjg0OTU2NDMy | User | false |
huggingface/diffusers | 2,453,156,026 | I_kwDOHa8MBc6SODC6 | 9,113 | https://github.com/huggingface/diffusers/issues/9113 | https://api.github.com/repos/huggingface/diffusers/issues/9113 | Basic Training Background color issue | ### Describe the bug
Hello!
Thanks so much for the tutorial. Here is what I have noticed:
There is a color background issue with the huggingface diffusion tutorial: https://huggingface.co/docs/diffusers/tutorials/basic_training
This was the output from the tutorial, you can see the background colour slightly in e... | closed | completed | true | 1 | [
"bug"
] | [] | 2024-08-07T10:37:42Z | 2024-08-07T10:44:08Z | 2024-08-07T10:44:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | IsaacAkintaro | 44,300,230 | MDQ6VXNlcjQ0MzAwMjMw | User | false |
huggingface/diffusers | 2,453,915,155 | I_kwDOHa8MBc6SQ8YT | 9,115 | https://github.com/huggingface/diffusers/issues/9115 | https://api.github.com/repos/huggingface/diffusers/issues/9115 | Error using AnimateDiff-Lightning and SDXL motion adapter | ### Describe the bug
Hello,
I asked a question in the HF Discord to see if it was possible to use `"ByteDance/AnimateDiff-Lightning"` and the SDXL motion adapter together. I was told to come here and make an issue.
When attempting to implement these to models together, on `diffusers==0.30`, I get an error saying... | closed | completed | false | 3 | [
"bug"
] | [] | 2024-08-07T16:48:41Z | 2024-08-08T08:17:24Z | 2024-08-08T08:17:23Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | MonkeeMan1 | 98,274,841 | U_kgDOBduOGQ | User | false |
huggingface/diffusers | 2,454,349,754 | I_kwDOHa8MBc6SSme6 | 9,120 | https://github.com/huggingface/diffusers/issues/9120 | https://api.github.com/repos/huggingface/diffusers/issues/9120 | [ar] Translating docs to Arabic (العربية) | <!--
Note: Please search to see if an issue already exists for the language you are trying to translate.
-->
Hi!
Let's bring the documentation to all the <languageName>-speaking community 🌐.
Who would want to translate? Please follow the 🤗 [TRANSLATING guide](https://github.com/huggingface/diffusers/blob/m... | closed | completed | false | 2 | [] | [] | 2024-08-07T21:04:54Z | 2024-10-29T08:14:24Z | 2024-10-29T08:14:19Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | AhmedAlmaghz | 53,489,256 | MDQ6VXNlcjUzNDg5MjU2 | User | false |
huggingface/diffusers | 2,454,666,349 | I_kwDOHa8MBc6STzxt | 9,123 | https://github.com/huggingface/diffusers/issues/9123 | https://api.github.com/repos/huggingface/diffusers/issues/9123 | (feature) diffusers.StableDiffusionAdapterPipeline could inherit FromSingleFileMixin | This is probably not a high traffic pipeline, but I have noticed it is probably possible for this pipeline to inherit ``FromSingleFileMixin`` harmlessly, it currently does not.
I am using this snippet to support single file loads with this pipeline and it seems to work fine when it inherits this mixin.
```python
... | closed | completed | false | 3 | [
"good first issue",
"contributions-welcome"
] | [] | 2024-08-08T01:30:44Z | 2025-01-29T17:18:49Z | 2025-01-29T17:18:49Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | Teriks | 14,919,098 | MDQ6VXNlcjE0OTE5MDk4 | User | false |
huggingface/diffusers | 2,454,735,059 | I_kwDOHa8MBc6SUEjT | 9,124 | https://github.com/huggingface/diffusers/issues/9124 | https://api.github.com/repos/huggingface/diffusers/issues/9124 | can't use the negative_prompt module in the fluxpipeline | The performance would be better if negative_prompt module be added,
I run fluxpipeline as below:
pipe = FluxPipeline.from_pretrained(model_path, torch_dtype=torch.bfloat16)
image = pipe(
prompt="a cat",
negative_prompt="ugly, messy, watermark",
width = OUTPUT_WIDTH,
... | closed | completed | false | 12 | [] | [] | 2024-08-08T02:21:37Z | 2024-09-21T12:07:20Z | 2024-09-21T12:07:20Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | DaaadShot | 88,389,446 | MDQ6VXNlcjg4Mzg5NDQ2 | User | false |
huggingface/diffusers | 2,455,198,116 | I_kwDOHa8MBc6SV1mk | 9,127 | https://github.com/huggingface/diffusers/issues/9127 | https://api.github.com/repos/huggingface/diffusers/issues/9127 | flux.1-dev device_map didn't work | I try to use device_map to use multiple gpu's, but it not worked, how can I use all my gpus?
| closed | completed | false | 33 | [] | [] | 2024-08-08T08:30:33Z | 2024-11-26T02:11:03Z | 2024-11-26T02:11:03Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | hznnnnnn | 92,504,688 | U_kgDOBYOCcA | User | false |
huggingface/diffusers | 1,415,296,258 | I_kwDOHa8MBc5UW7UC | 913 | https://github.com/huggingface/diffusers/issues/913 | https://api.github.com/repos/huggingface/diffusers/issues/913 | Improve tests | Our tests suite exploded and we need a new structure.
We should do the following:
```
Create a `PipelineTesterMixin` class that is identical in spirit to ModelTesterMixin and then we make sure our pipeline tests reflect 1-to-1 the pipeline naming e.g. we should have:
- tests
- pipelines
- stable_d... | closed | completed | false | 1 | [
"stale"
] | [
"anton-l"
] | 2022-10-19T16:48:10Z | 2022-11-27T15:03:02Z | 2022-11-27T15:03:02Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 2,457,019,996 | I_kwDOHa8MBc6ScyZc | 9,134 | https://github.com/huggingface/diffusers/issues/9134 | https://api.github.com/repos/huggingface/diffusers/issues/9134 | Issue with Flux LoRAs trained with SimpleTuner | @bghira we're seeing issues when doing:
```python
import torch
from diffusers import DiffusionPipeline
base_model = "black-forest-labs/FLUX.1-dev"
pipe = DiffusionPipeline.from_pretrained(base_model, torch_dtype=torch.bfloat16).to("cuda")
pipe.load_lora_weights("pzc163/flux-lora-littletinies")
generator = ... | closed | completed | false | 34 | [] | [] | 2024-08-09T03:29:20Z | 2024-09-08T04:08:55Z | 2024-08-12T01:39:59Z | MEMBER | null | 20260407T133413Z | 2026-04-07T13:34:13Z | sayakpaul | 22,957,388 | MDQ6VXNlcjIyOTU3Mzg4 | User | false |
huggingface/diffusers | 2,457,211,605 | I_kwDOHa8MBc6SdhLV | 9,136 | https://github.com/huggingface/diffusers/issues/9136 | https://api.github.com/repos/huggingface/diffusers/issues/9136 | IP adapter output on some resolutions suffers in quality? | ### Describe the bug
I am running IP adapter for 768x1344 which is one of the sdxl listed resolutions. I find that the output quality is much less than say regular 768x768 generations. I've attached sample images and code below. In this experiment 1080x768 seemed to get best output, but its not one of the supported re... | open | null | false | 2 | [
"bug",
"stale"
] | [] | 2024-08-09T06:36:39Z | 2024-09-14T15:03:17Z | null | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | darshats | 56,063,876 | MDQ6VXNlcjU2MDYzODc2 | User | false |
huggingface/diffusers | 2,457,371,231 | I_kwDOHa8MBc6SeIJf | 9,138 | https://github.com/huggingface/diffusers/issues/9138 | https://api.github.com/repos/huggingface/diffusers/issues/9138 | TypeError: '>' not supported between instances of 'int' and 'str' | ### Describe the bug
I train controlnet by script `examples/train_controlnet.py` use the commad like:
```bash
export MODEL_DIR="runwayml/stable-diffusion-v1-5"
export DATA_DIR="/home/zhangnan/codes/diffusers/examples/controlnet/fill50k_bak"
export OUTPUT_DIR="./output"
accelerate launch train_controlnet.py \
... | closed | completed | false | 7 | [
"bug",
"stale"
] | [] | 2024-08-09T08:12:20Z | 2024-09-18T15:05:08Z | 2024-09-18T15:05:08Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | nannanmath | 7,345,714 | MDQ6VXNlcjczNDU3MTQ= | User | false |
huggingface/diffusers | 1,415,301,932 | I_kwDOHa8MBc5UW8ss | 914 | https://github.com/huggingface/diffusers/issues/914 | https://api.github.com/repos/huggingface/diffusers/issues/914 | Improve the precision of our integration tests | We currently have a rather low precision when testing our pipeline due to due reasons.
1. - Our reference is an image and not a numpy array. This means that when we created our reference image we lost float precision which is unnecessary
2. - We only test for `.max() < 1e-2` . IMO we should test for `.max() < 1e-4` ... | closed | completed | false | 7 | [
"stale"
] | [
"patrickvonplaten"
] | 2022-10-19T16:52:57Z | 2023-01-02T15:03:50Z | 2023-01-02T15:03:50Z | CONTRIBUTOR | null | 20260407T133413Z | 2026-04-07T13:34:13Z | patrickvonplaten | 23,423,619 | MDQ6VXNlcjIzNDIzNjE5 | User | false |
huggingface/diffusers | 2,457,518,311 | I_kwDOHa8MBc6SesDn | 9,140 | https://github.com/huggingface/diffusers/issues/9140 | https://api.github.com/repos/huggingface/diffusers/issues/9140 | Diffusers model not working as good as repo ckpt model | Hi,
When I try to run the models stable diffusion v1-5 or Instructpix2pix through the diffusers pipeline and use .from_pretrained() it downloads the models from hugging face and I'm using the code to run inference given in hugging face, the results are not good at all in the sense that there is still noise in the gener... | closed | completed | false | 6 | [
"stale"
] | [] | 2024-08-09T09:34:30Z | 2024-12-14T12:13:15Z | 2024-12-14T12:13:14Z | NONE | null | 20260407T133413Z | 2026-04-07T13:34:13Z | kunalkathare | 76,777,272 | MDQ6VXNlcjc2Nzc3Mjcy | User | false |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.