ComfyUI_TGate

ComfyUI reference implementation for T-GATE.

Example workflows

The examples directory has workflow example. There are images generated with and without T-GATE in the assets folder.

example

Origin result

origin_result

T-GATE result (start at 0.5, only_cross_attention) tgate_result

INSTALL

git clone https://github.com/JettHu/ComfyUI_TGate
git apply tgate.patch   # need patch yet, PR in progress

Major Features

  • Training-Free.
  • Friendly support CNN-based U-Net, Transformer, and Consistency Model
  • 10%-50% speed up for different diffusion models.

Nodes reference

TGateApply

Inputs

  • model, model loaded by Load Checkpoint and other MODEL loaders.

Configuration parameters

  • start_at, this is the timestepping. Defines at what percentage point of the generation to start use the T-GATE cache.
  • only_cross_attention, default only used to cache the cross-attention output, ref to issues

Optional configuration

  • self_attn_start_at, only takes effect when only_cross_attention is false, timestepping too. Defines at what percentage point of the generation to start use the T-GATE cache.

Performance (from T-GATE)

Model MACs Param Latency Zero-shot 10K-FID on MS-COCO
SD-1.5 16.938T 859.520M 7.032s 23.927
SD-1.5 w/ TGATE 9.875T 815.557M 4.313s 20.789
SD-2.1 38.041T 865.785M 16.121s 22.609
SD-2.1 w/ TGATE 22.208T 815.433 M 9.878s 19.940
SD-XL 149.438T 2.570B 53.187s 24.628
SD-XL w/ TGATE 84.438T 2.024B 27.932s 22.738
Pixart-Alpha 107.031T 611.350M 61.502s 38.669
Pixart-Alpha w/ TGATE 65.318T 462.585M 37.867s 35.825
DeepCache (SD-XL) 57.888T - 19.931s 23.755
DeepCache w/ TGATE 43.868T - 14.666s 23.999
LCM (SD-XL) 11.955T 2.570B 3.805s 25.044
LCM w/ TGATE 11.171T 2.024B 3.533s 25.028
LCM (Pixart-Alpha) 8.563T 611.350M 4.733s 36.086
LCM w/ TGATE 7.623T 462.585M 4.543s 37.048

The latency is tested on a 1080ti commercial card.

The MACs and Params are calculated by calflops.

The FID is calculated by PytorchFID.

TODO

  • Result image quality is inconsistent with origin. Now cache attn2 (cross_attention) only.
  • Implement a native version and no longer rely on git patch
S
Description
No description provided
Readme GPL-3.0
5.9 MiB
Languages
Python 100%