 AUTOMATIC
|
8faac8b963
run basic torch calculation at startup in parallel to reduce the performance impact of first generation
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
028d3f6425
ruff auto fixes
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
5fe0dd79be
rename CPU RNG to RNG source in settings, add infotext and parameters copypaste support to RNG source
|
%!s(int64=2) %!d(string=hai) anos |
 Deciare
|
d40e44ade4
Option to use CPU for random number generation.
|
%!s(int64=2) %!d(string=hai) anos |
 brkirch
|
1b8af15f13
Refactor Mac specific code to a separate file
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
2217331cd1
Refactor MPS fixes to CondFunc
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
7738c057ce
MPS fix is still needed :(
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC1111
|
fecb990deb
Merge pull request #7309 from brkirch/fix-embeddings
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
f9edd578e9
Remove MPS fix no longer needed for PyTorch
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
ada17dbd7c
Refactor conditional casting, fix upscalers
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
9beb794e0b
clarify the option to disable NaN check.
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
d2ac95fa7b
remove the need to place configs near models
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
e3b53fd295
Add UI setting for upcasting attention to float32
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
84d9ce30cb
Add option for float32 sampling with float16 UNet
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC1111
|
aa60fc6660
Merge pull request #6922 from brkirch/cumsum-fix
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
a255dac4f8
Fix cumsum for MPS in newer torch
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
c361b89026
disable the new NaN check for the CI
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
9991967f40
Add a check and explanation for tensor with all NaNs.
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
8111b5569d
Add support for PyTorch nightly and local builds
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
16b4509fa6
Add numpy fix for MPS on PyTorch 1.12.1
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
b6e5edd746
add built-in extension system
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
46b0d230e7
add comment for #4407 and remove seemingly unnecessary cudnn.enabled
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
2651267e3a
fix #4407 breaking UI entirely for card other than ones related to the PR
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC1111
|
681c0003df
Merge pull request #4407 from yoinked-h/patch-1
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
0fddb4a1c0
Rework MPS randn fix, add randn_like fix
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC1111
|
cc90dcc933
Merge pull request #4918 from brkirch/pytorch-fixes
|
%!s(int64=2) %!d(string=hai) anos |
|
AUTOMATIC
|
5b2c316890
eliminate duplicated code from #5095
|
%!s(int64=2) %!d(string=hai) anos |
 Matthew McGoogan
|
c67c40f983
torch.cuda.empty_cache() defaults to cuda:0 device unless explicitly set otherwise first. Updating torch_gc() to use the device set by --device-id if specified to avoid OOM edge cases on multi-GPU systems.
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
e247b7400a
Add fixes for PyTorch 1.12.1
|
%!s(int64=2) %!d(string=hai) anos |
|
brkirch
|
abfa22c16f
Revert "MPS Upscalers Fix"
|
%!s(int64=2) %!d(string=hai) anos |
