huggingface / diffusers

🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.

https://huggingface.co/docs/diffusers

Repository from Github https://github.comhuggingface/diffusers

huggingface/diffusers Issues

UNet1DModel does not converge
Updated 6 months ago
Script to convert FLUX diffusers to safetensors
Updated 6 months ago2
Sage Attention for diffuser library
Updated 6 months ago5
LattePipeline fails with dtype mismatch
Closed 6 months ago
Implement caching on LTX and WAN video models
Updated 6 months ago
Stable Virtual Camera
Updated 6 months ago1
Failed to import diffusers.schedulers.scheduling_ddim because of the following error (look up to see its traceback): numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject
Updated 6 months ago
WanImageToVideoPipeline update to work with offloading
Closed 6 months ago1
WanImageToVideoPipeline broken math when preparing latents
Updated 6 months ago1
Flux FIll lora / dreambooth finetune support
Updated 6 months ago
Inconsistent Inference Results with Diffusers’ Implementation of WAN 2.1 14B I2V
Updated 6 months ago
[PEFT] Loading a LoRA in PeftAdapterMixin does not set _hf_peft_config_loaded to True; causes issues with {enable,disable}_adapters
Closed 6 months ago1
TypeError: BnB4BitDiffusersQuantizer.create_quantized_param() got an unexpected keyword argument 'dtype'
Updated 6 months ago1
[LTX0.9.5] make LTX0.9.5 works with text-to-video
Updated 6 months ago9
[New Pipeline]: SmoothCache: A Universal Inference Acceleration Technique for Diffusion Transformers
Updated 6 months ago2
FlaxUNet2DConditionModel is not initialized with correct dtypes
Updated 6 months ago3
Generated video quality is not up to the mark LTX-Video 0.9.5
Closed 6 months ago4
Pipelines loaded with `dtype=torch.float16` cannot run with `cpu` device.
Updated 6 months ago3
SanaSprintPipeline: Intermediate timesteps for SCM is not supported when num_inference_steps != 2
Closed 6 months ago1
Civit AI flux model razor-8step-rapid-real not working with diffusers single file
Updated 6 months ago4
HunyuanVideoImageToVideoPipeline failures
Updated 6 months ago2
bug while using cogvideox image to video pipeline
Updated 6 months ago
support SageAttention
Updated 6 months ago
Is there a way to generate a single image using multiple GPUs?
Updated 6 months ago9
Limit on resolution for decoding for REMOTEVAE
Updated 6 months ago6
Error in loading Civit AI Lora: LCMTurboMix_Euler_A_fix
Updated 6 months ago1
StableDiffusion3Pipeline RuntimeError: c10::Half != c10::BFloat16
Updated 6 months ago2
ControlNets for Multi-GPU
Updated 6 months ago
channel inconsistency in cogvideo Lora training example
Updated 6 months ago1
AutoencoderKLWan - support grandient_checkpointing
Updated 6 months ago3
LTX-Video-0.9.5 both offload methods are broken
Closed 6 months ago4
TorchAO example broken
Updated 6 months ago3
depth lora causing transformer layers mismatch
Closed 6 months ago2
Which repo should I use for LTX-Video 0.9.5 diffusers
Closed 6 months ago2
Running diffusers with torch-DDP leads to warning on stride mismatch
Closed 6 months ago1
RuntimeError after using apply_group_offloading on diffusers: Input type (torch.cuda.FloatTensor) and weight type (torch.FloatTensor) should be the same
Updated 6 months ago4
add LTX-Video 0.9.5 diffusers support
Closed 6 months ago1
Empty timesteps in StableDiffusionXLImg2ImgPipeline with SDXL-Turbo using strength > 0.5 and 1 inference step
Updated 6 months ago
wan2.1 transformer gguf load error
Updated 6 months ago4
`prepare_image` in Kandinsky pipelines doesn't support `torch.Tensor`
Updated 6 months ago1
About VAE convert script
Updated 6 months ago
Remove VAE encode
Updated 6 months ago
[examples/controlnet/train_controlnet_sd3.py] training SD3.5 large controlnets from stability
Updated 6 months ago1
prepare_attention_mask - incorrect padding?
Updated 6 months ago1
[examples/controlnet/train_controlnet_sd3.py] prompt_embeds and pooled_prompt_embeds not cast to weight_dtype in bf16/fp16 training
Closed 6 months ago2
Output of randn_tensor changes based on dtype, making seeds non-reproducible if the model precision is changed
Updated 6 months ago3
No controlnet pag inpaint model in pipeline
Closed 6 months ago3
Training on unconditional image generation creates colorized images
Updated 6 months ago
Training Wan2.1 LoRAs
Closed 6 months ago2
flux pipeline inference with controlnet, inpainting, plus ip-adapter
Updated 6 months ago