- the model weights generally have to be patched ahead of time for compilation to work
- the model downloader matches the folder_paths API a bit better
- tweak the logging from the execution node
- GGUF now works and included 4 bit quants, this will allow WAN to run on 24GB VRAM GPUs
- logger only shows full stack for errors
- helper functions for colab notebook
- fix nvcr.io auth error
- lora issues are now reported better
- model downloader will use huggingface cache and symlinking if it is supported on your platform
- torch compile node now correctly patches the model before compilation
- add xet support and add the xet cache to manageable directories
- xet is enabled by default
- fix logging to root in various places
- improve logging about model unloading and loading
- TorchCompileNode now supports the VAE
- torchaudio missing will cause less noise in the logs
- feature flags will assume to be supporting everything in the distributed progress context
- fixes progress notifications
- Wan and Cosmos prompt upsamplers
- Fixed torch.compile issues
- Known models added
- Cosmos, Wan and Hunyuan Video resolutions now supported by Fit Image
to Diffusion Size.
- Better error messages for Ampere and Triton interactions
- Cosmos now fully tested
- Preliminary support for essential Cosmos prompt "upsampler"
- Lumina tests
- Tweaks to language and image resizing nodes
- Fix for #31 all the samplers are now present again
- fix#29 str(model) no longer raises exceptions like with
HyVideoModelLoader
- don't try to format CUDA tensors because that can sometimes raise
exceptions
- cudaAllocAsync has been disabled for now due to 2.6.0 bugs
- improve florence2 support
- add support for paligemma 2. This requires the fix for transformers
that is currently staged in another repo, install with
`uv pip install --no-deps "transformers@git+https://github.com/zucchini-nlp/transformers.git#branch=paligemma-fix-kwargs"`
- triton has been updated
- fix missing __init__.py files
- include detailed runtime instructions for Windows and macOS
- include instructions for agreeing to use Hugging Face repositories
- always create directories by default when run interactively
- model downloader now supports multiple folder names for known models
- improve logging in sd.py
- use their logger when running interactively
- move the extra nodes files to where this fork expects them
- add the mochi checkpoints to known models
- add a mochi workflow test
- Experimental support for sage attention on Linux
- Diffusers loader now supports model indices
- Transformers model management now aligns with updates to ComfyUI
- Flux layers correctly use unbind
- Add float8 support for model loading in more places
- Experimental quantization approaches from Quanto and torchao
- Model upscaling interacts with memory management better
This update also disables ROCm testing because it isn't reliable enough
on consumer hardware. ROCm is not really supported by the 7600.
- move workflows to distinct json files
- add the comfy-org workflows for testing
- fix issues where workflows from windows users would not be compatible
with backends running on linux or macos in light of separator
differences. Because this codebase uses get_or_download wherever
checkpoints, models, etc. are used, this is the only place where the
comparison is gracefully handled for downloading. Validation code
will correctly convert backslashes to forward slashes, assuming that
100% of the places they are used and when comparing with a list, they
are intended to be paths and not strict symbols
