psychedelicious 89c999ca58 fix(backend): remove mps_fixes ...

The fixes in this module monkeypatched `torch` to resolve some issues with FP16 on macOS. These issues have long since been resolved.

Included in the now-removed fixes is `CustomSlicedAttentionProcessor`, which is intended to reduce memory requirements for MPS. This overrides `diffusers`' own `SlicedAttentionProcessor`.

Unfortunately, `attention_type: sliced` produces hot garbage with the fixes and black images without the fixes. So this class appears to now be a moot point.

Regardless, SDPA is supported on MPS and very efficient, so sliced attention is largely obsolete.

2025-04-04 18:42:13 +11:00

..

flux

fix(mm): handle FLUX models w/ diff in_channels keys

2025-03-31 12:32:55 +11:00

image_util

Revert some files

2025-03-26 13:18:50 +11:00

ip_adapter

fix(backend): revert non-blocking device transfer

2024-07-16 08:59:42 +10:00

model_hash

Run ruff

2024-06-26 21:46:59 +10:00

model_manager

Merge branch 'main' into lora-classification

2025-03-31 16:43:26 +11:00

onnx

fix(backend): revert non-blocking device transfer

2024-07-16 08:59:42 +10:00

patches

Make ruff happy

2025-03-27 17:47:57 +11:00

quantization

Update GGMLTensor with ops necessary to work with ConcatenatedLoRALayer.

2025-01-28 14:51:35 +00:00

sd3

Update SD3 InpaintExtension with gradient adjustment to match FLUX.

2024-11-07 22:55:30 +00:00

sig_lip

Fix SigLipPipeline model size calculation.

2025-03-06 10:31:17 +11:00

stable_diffusion

fix(app): add trusted classes to torch safe globals to prevent errors when loading them

2025-04-04 18:42:13 +11:00

tiles

feat(nodes): extract LATENT_SCALE_FACTOR to constants.py

2024-03-01 10:42:33 +11:00

util

fix(backend): remove mps_fixes

2025-04-04 18:42:13 +11:00

__init__.py

consolidate model manager parts into a single class

2024-03-01 10:42:33 +11:00

llava_onevision_model.py

fix(mm): vllm model calculates its own size

2025-03-27 09:36:14 +11:00

model_patcher.py

We should not trust the value of since the model could be partially-loaded.

2025-01-07 19:22:31 -05:00

raw_model.py

Merge branch 'main' into ryan/spandrel-upscale

2024-07-16 15:14:08 -04:00

spandrel_image_to_image_model.py

Add tiling support to the SpoandrelImageToImage node.

2024-07-10 14:25:19 -04:00

textual_inversion.py

Add util functions calc_tensor_size(...) and calc_tensors_size(...).

2024-09-15 04:39:56 +03:00