Overview

18 Pull requests merged by 12 people

• [docs] Models

  #12248 merged Sep 5, 2025

• [docs] Inference section cleanup

  #12281 merged Sep 5, 2025

• [docs] components_to_quantize

  #12287 merged Sep 5, 2025

• [docs] Sharing pipelines/models

  #12280 merged Sep 4, 2025

• fix some typos

  #12265 merged Sep 3, 2025

• Add AttentionMixin to WanVACETransformer3DModel

  #12268 merged Sep 3, 2025

• [tests] feat: add AoT compilation tests

  #12203 merged Sep 3, 2025

• [Quantization] Add TRT-ModelOpt as a Backend

  #11173 merged Sep 3, 2025

• [docs] AutoPipeline

  #12160 merged Sep 3, 2025

• [core] use kernels to support _flash_3_hub attention backend

  #12236 merged Sep 3, 2025

• [CI] Remove big accelerator requirements from Quanto Tests

  #12266 merged Sep 3, 2025

• [Flux LoRA] fix for prior preservation and mixed precision sampling, follow up on #11873

  #12264 merged Sep 2, 2025

• [docs] Fix VAE scale factor calculation in distributed inference docs

  #12259 merged Sep 2, 2025

• [fix] fix for prior preservation and mixed precision sampling

  #11873 merged Sep 1, 2025

• Fix lora conversion function for ai-toolkit Qwen Image LoRAs

  #12261 merged Sep 1, 2025

• Add Qwen-Image-Edit Inpainting pipeline

  #12225 merged Aug 31, 2025

• Improve performance of NPU FA

  #12260 merged Aug 30, 2025

• [QwenImageEditPipeline] Add image entry in __call__ function

  #12254 merged Aug 30, 2025

14 Pull requests opened by 8 people

• fix incorrect sigmas of DPMSolverMultistepInverseScheduler for flow_prediction

  #12262 opened Aug 30, 2025

• Fix #12116: preserve boolean dtype for attention masks in ChromaPipeline

  #12263 opened Aug 30, 2025

• Support Flux Kontext in modular (T2I and I2I)

  #12269 opened Sep 2, 2025

• Deprecate slicing and tiling methods from `DiffusionPipeline`

  #12271 opened Sep 3, 2025

• [modular] fix flux modular pipelines for t2i and i2i

  #12272 opened Sep 3, 2025

• [quantization] feat: support aobaseconfig classes in `TorchAOConfig`

  #12275 opened Sep 3, 2025

• [tests] disable xformer tests for pipelines it isn't popular.

  #12277 opened Sep 3, 2025

• Bump torch from 2.2.0 to 2.8.0 in /examples/research_projects/realfill

  #12279 opened Sep 3, 2025

• [core] feat: support group offloading at the pipeline level

  #12283 opened Sep 4, 2025

• [docs] Distributed inference

  #12285 opened Sep 4, 2025

• [docs] Pipeline group offloading

  #12286 opened Sep 4, 2025

• [WIP] Add InfiniteTalk Audio-Driven Video Generation Pipeline

  #12288 opened Sep 5, 2025

• Fix many type hint errors

  #12289 opened Sep 5, 2025

• adjust criteria for marigold-intrinsics example on XPU

  #12290 opened Sep 5, 2025

3 Issues closed by 3 people

• Failed to finetune the pre-trained model of 'stable-diffusion-v1-4' on image inpainting task

  #12165 closed Sep 4, 2025

• Questions about train_text_to_image_lora_sdxl.py

  #12282 closed Sep 3, 2025

• [BUG] Misleading ValueError when subclassing StableDiffusionImg2ImgPipeline with a mismatched __init__ signature

  #12255 closed Aug 30, 2025

4 Issues opened by 4 people

• Attention test seems uncompatible with latest xformers

  #12284 opened Sep 4, 2025

• The image is blurry.

  #12276 opened Sep 3, 2025

• New HunyuanVideoI2V PR changes to transformer_hunyuan_video.py

  #12273 opened Sep 3, 2025

• From 0.35.0: DLL load failed while importing libtriton: A dynamic link library (DLL) initialization routine failed.

  #12270 opened Sep 2, 2025

24 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Add Step1X-Edit Pipeline

  #12249 commented on Sep 1, 2025 • 37 new comments

• Context Parallel w/ Ring & Ulysses & Unified Attention

  #11941 commented on Sep 4, 2025 • 22 new comments

• [Modular] Qwen

  #12220 commented on Sep 4, 2025 • 12 new comments

• [docs] Model formats

  #12256 commented on Sep 5, 2025 • 11 new comments

• [quant] allow `components_to_quantize` to be a non-list for single components

  #12234 commented on Sep 5, 2025 • 2 new comments

• Add Wan2.2-S2V

  #12258 commented on Sep 5, 2025 • 1 new comment

• DeepSpeed adaption for flux-kontext

  #12240 commented on Sep 5, 2025 • 0 new comments

• [Refactor] Move transformer blocks out of attention module

  #12229 commented on Sep 3, 2025 • 0 new comments

• [Modular] implement requirements validation for custom blocks

  #12196 commented on Sep 2, 2025 • 0 new comments

• [scheduler] Add SDE Samplers: Curved Euler and Overshoot

  #12192 commented on Sep 1, 2025 • 0 new comments

• Add Dream 7B Diffusion Large Language Model Pipeline

  #12091 commented on Sep 4, 2025 • 0 new comments

• [lora] factor out the overlaps in `save_lora_weights()`.

  #12027 commented on Sep 2, 2025 • 0 new comments

• Add F5 TTS pipeline

  #11958 commented on Sep 5, 2025 • 0 new comments

• Propose to refactor `norm_out` in `SkyReelsV2Transformer3DModel` for device handling in multi-GPU case

  #11850 commented on Sep 3, 2025 • 0 new comments

• AttributeError: _hf_hook caused by delattr in hooks.remove_hook_from_module()

  #10729 commented on Sep 5, 2025 • 0 new comments

• Support for InfiniteTalk

  #12239 commented on Sep 5, 2025 • 0 new comments

• FluxFillPipeline quality deteriorates over multiple iterations of inpainting

  #10204 commented on Sep 3, 2025 • 0 new comments

• Dreambooth finetune FLUX dev CLIPTextModel

  #10925 commented on Sep 3, 2025 • 0 new comments

• `WanVACETransformer3DModel` does not inherit `AttentionMixin` while `WanTransformer3DModel` does

  #12186 commented on Sep 2, 2025 • 0 new comments

• Run FLUX-controlnet zero3 training failed: 'weight' must be 2-D

  #10969 commented on Sep 1, 2025 • 0 new comments

• Qwen-Image-Edit Inferior Results Compared to ComfyUI

  #12216 commented on Sep 1, 2025 • 0 new comments

• stabilityai/stable-diffusion-2-base is missing fp16 files' so that the correct variant file can be added

  #12253 commented on Sep 1, 2025 • 0 new comments

• Error RuntimeError: Invalid buffer size

  #5894 commented on Aug 30, 2025 • 0 new comments

• For qwen-image training file, Maybe "shuffle" of dataloader should be "False" when custom_instance_prompts is not None and cache_latents is False?

  #12124 commented on Aug 30, 2025 • 0 new comments