Issues

• fal FLUX.1 Kontext LoRAs with unprefixed embedder keys fail in _convert_fal_kontext_lora_to_diffusers

  huggingface/diffusers#14917 · Filter by author christopher5106 opened on Sep 30, 2026

• CFG cutoff callbacks keep a single embedding row, so they break for num_images_per_prompt > 1 or several prompts

  huggingface/diffusers#14907 · Filter by author JoeyTan21 opened on Sep 29, 2026
  1

• Required weight pass on a singlefile

  huggingface/diffusers#14903 · Filter by author xalteropsx opened on Sep 29, 2026

• numpy_to_pil removes singleton spatial dimensions from grayscale images

  huggingface/diffusers#14901 · Filter by author Afloat16 opened on Sep 29, 2026
  1

• Legacy final_offload_hook checks are dead in 14 pipelines; KandinskyPriorPipeline skips maybe_free_model_hooks() when negative_prompt is set

  huggingface/diffusers#14896 · Filter by author kv-248 opened on Sep 28, 2026
  1

• Cosmos2VideoToWorldPipeline conditions on the entire input video — no num_conditional_frames, so a full clip is reconstructed instead of continued

  bug
  huggingface/diffusers#14892 · Filter by author rebel-seinpark opened on Sep 28, 2026
  2

• Converted sigma schedules with final_sigmas_type="sigma_min" repeat the last sigma, and the third-order and heun updates of DPMSolverMultistepScheduler and UniPCMultistepScheduler return NaN on that zero-width final step

  huggingface/diffusers#14890 · Filter by author Nicholas022400701 opened on Sep 28, 2026

• QwenImage21 FlexAttention mask fails to compile on MPS because of boolean ~

  huggingface/diffusers#14889 · Filter by author korellas opened on Sep 28, 2026

• UniPCMultistepScheduler fails in torch.linalg.solve under a float64 default dtype: the unit entry of rks takes the default dtype

  huggingface/diffusers#14888 · Filter by author Nicholas022400701 opened on Sep 27, 2026

• UniPCMultistepScheduler returns nan with solver_type="bh1", and inf or nan with lower_order_final=False or predict_x0=False, on the final step to sigma = 0

  huggingface/diffusers#14887 · Filter by author Nicholas022400701 opened on Sep 27, 2026

• [docs] 22 dead external links across the English docs

  huggingface/diffusers#14886 · Filter by author pratikgx opened on Sep 27, 2026

• WanAnimatePipeline.get_i2v_mask() defaults device to "cuda", which raises on non-CUDA accelerators (NPU/XPU/MPS)

  huggingface/diffusers#14881 · Filter by author li-lizhe opened on Sep 27, 2026

• Cosmos denoising loops sync with the host every step

  huggingface/diffusers#14880 · Filter by author adrianrfreedman opened on Sep 27, 2026

• Five Z-Image pipelines miss the precomputed-t_norms DtoH-sync fix from ZImagePipeline

  huggingface/diffusers#14879 · Filter by author adrianrfreedman opened on Sep 27, 2026

• Pipeline.to() offload guards only cover cuda/xpu: Ascend NPU misses the sequential-offload conflict error and the model-offload warning

  huggingface/diffusers#14878 · Filter by author li-lizhe opened on Sep 26, 2026

• apply_group_offloading(use_stream=True) raises on Ascend NPU: no stream branch for torch.npu

  huggingface/diffusers#14877 · Filter by author li-lizhe opened on Sep 26, 2026

• Ming Image 0.1 Design

  huggingface/diffusers#14873 · Filter by author alerikaisattera opened on Sep 25, 2026
  3

• [Feature] Support backward for Ulysses Anything attention

  huggingface/diffusers#14871 · Filter by author NancyFyong opened on Sep 25, 2026

• MiniMax H3 transformer does not expose attention masks for padded joint sequences

  huggingface/diffusers#14870 · Filter by author NancyFyong opened on Sep 25, 2026

• IndexError when preprocessing an empty image or video list

  huggingface/diffusers#14864 · Filter by author MohammadHijjawi97 opened on Sep 24, 2026

• QwenImage21Pipeline on MPS: every edit after the first in the same process returns all-NaN at output_resolution=1024

  huggingface/diffusers#14859 · Filter by author ck71 opened on Sep 23, 2026
  4

• QwenImage21Pipeline: editing an image the model generated degrades at output_resolution=1024 regardless of seed (not noise replay, MPS)

  huggingface/diffusers#14858 · Filter by author ck71 opened on Sep 23, 2026
  11

• Explore llama.cpp/ggml quantization kernels for our GGUF integration

  performance
  huggingface/diffusers#14841 · Filter by author moon-bot-app opened on Sep 22, 2026
  1

• Windows: check_ai.py fails decoding UTF-8 guides with the default locale

  huggingface/diffusers#14837 · Filter by author dlowzzxx opened on Sep 22, 2026

• Enable KV cache offloading for a better speed-memory trade-off

  performance
  huggingface/diffusers#14833 · Filter by author sayakpaul opened on Sep 22, 2026