COMMITS
May 29, 2026
T
Initializers to Torch (#1549)
torzdf committed
May 22, 2026
T
bugfix: lpips - disable grad
torzdf committed
T
bugfix: Set color order for relevant loss functions
torzdf committed
May 21, 2026
T
extract bugfixes: vgg-obstructed model path + unet-dfl output shape
torzdf committed
T
Migrate optimizers to torch (#1547)
torzdf committed
May 9, 2026
T
Completely remove Keras from loss calculations (#1545)
torzdf committed
May 8, 2026
T
bugfix: setup.py - ensurepip
torzdf committed
April 24, 2026
T
bugfix: Native torch loss functions running in Keras
torzdf committed
T
bugfix: GUI loading projects with removed options
torzdf committed
T
bugfix: Loss function device allocation
torzdf committed
April 23, 2026
T
All loss functions to Torch
torzdf committed
April 21, 2026
T
bugfix: FS2 to 3 SeparableConv2D update
torzdf committed
T
Most loss functions to torch
torzdf committed
April 20, 2026
T
hrnet: Slightly improved weights
torzdf committed
T
lpips: To torch + crop feature map zero-padding
torzdf committed
April 16, 2026
T
Train: Migrate data loading to Torch (#1540)
torzdf committed
April 8, 2026
T
bugfix: alignments tool unit test
torzdf committed
T
bugfix: Aligned faces. Fix padding calculation
torzdf committed
T
bugfix: legacy alignments conversion
torzdf committed
T
April 7, 2026
April 5, 2026
T
bugfix: Align - Fix re-feed OOB cropping + loosen HRNet BBox
torzdf committed
T
bugfix: Manual tool - prevent error on no masks
torzdf committed
T
Alignments: Migrate typed-dicts to dataclasses
torzdf committed
March 31, 2026
T
Bugfix: Manual tool. Correctly detect frames folder on load
torzdf committed
March 30, 2026
T
Remove legacy face set support
torzdf committed
March 29, 2026
T
Bugfix: Extract/Convert - Don't error when an input frame is black
torzdf committed
March 28, 2026
T
convert bugfixes:
torzdf committed
March 27, 2026
T
Fix Preview viewer test
torzdf committed