fecb990deb
Merge pull request #7309 from brkirch/fix-embeddings
...
Fix embeddings, upscalers, and refactor `--upcast-sampling`
2023-01-28 18:44:36 +03:00
f9edd578e9
Remove MPS fix no longer needed for PyTorch
...
The torch.narrow fix was required for nightly PyTorch builds for a while to prevent a hard crash, but newer nightly builds don't have this issue.
2023-01-28 04:16:27 -05:00
ada17dbd7c
Refactor conditional casting, fix upscalers
2023-01-28 04:16:25 -05:00
9beb794e0b
clarify the option to disable NaN check.
2023-01-27 13:08:00 +03:00
d2ac95fa7b
remove the need to place configs near models
2023-01-27 11:28:12 +03:00
e3b53fd295
Add UI setting for upcasting attention to float32
...
Adds "Upcast cross attention layer to float32" option in Stable Diffusion settings. This allows for generating images using SD 2.1 models without --no-half or xFormers.
In order to make upcasting cross attention layer optimizations possible it is necessary to indent several sections of code in sd_hijack_optimizations.py so that a context manager can be used to disable autocast. Also, even though Stable Diffusion (and Diffusers) only upcast q and k, unfortunately my findings were that most of the cross attention layer optimizations could not function unless v is upcast also.
2023-01-25 01:13:04 -05:00
84d9ce30cb
Add option for float32 sampling with float16 UNet
...
This also handles type casting so that ROCm and MPS torch devices work correctly without --no-half. One cast is required for deepbooru in deepbooru_model.py, some explicit casting is required for img2img and inpainting. depth_model can't be converted to float16 or it won't work correctly on some systems (it's known to have issues on MPS) so in sd_models.py model.depth_model is removed for model.half().
2023-01-25 01:13:02 -05:00
aa60fc6660
Merge pull request #6922 from brkirch/cumsum-fix
...
Improve cumsum fix for MPS
2023-01-19 13:18:34 +03:00
a255dac4f8
Fix cumsum for MPS in newer torch
...
The prior fix assumed that testing int16 was enough to determine if a fix is needed, but a recent fix for cumsum has int16 working but not bool.
2023-01-17 20:54:18 -05:00
c361b89026
disable the new NaN check for the CI
2023-01-17 11:05:01 +03:00
9991967f40
Add a check and explanation for tensor with all NaNs.
2023-01-16 22:59:46 +03:00
8111b5569d
Add support for PyTorch nightly and local builds
2023-01-05 20:54:52 -05:00
16b4509fa6
Add numpy fix for MPS on PyTorch 1.12.1
...
When saving training results with torch.save(), an exception is thrown:
"RuntimeError: Can't call numpy() on Tensor that requires grad. Use tensor.detach().numpy() instead."
So for MPS, check if Tensor.requires_grad and detach() if necessary.
2022-12-17 04:22:58 -05:00
b6e5edd746
add built-in extension system
...
add support for adding upscalers in extensions
move LDSR, ScuNET and SwinIR to built-in extensions
2022-12-03 18:06:33 +03:00
46b0d230e7
add comment for #4407 and remove seemingly unnecessary cudnn.enabled
2022-12-03 16:01:23 +03:00
2651267e3a
fix #4407 breaking UI entirely for card other than ones related to the PR
2022-12-03 15:57:52 +03:00
681c0003df
Merge pull request #4407 from yoinked-h/patch-1
...
Fix issue with 16xx cards
2022-12-03 10:30:34 +03:00
0fddb4a1c0
Rework MPS randn fix, add randn_like fix
...
torch.manual_seed() already sets a CPU generator, so there is no reason to create a CPU generator manually. torch.randn_like also needs a MPS fix for k-diffusion, but a torch hijack with randn_like already exists so it can also be used for that.
2022-11-30 10:33:42 -05:00
cc90dcc933
Merge pull request #4918 from brkirch/pytorch-fixes
...
Fixes for PyTorch 1.12.1 when using MPS
2022-11-27 13:47:01 +03:00
5b2c316890
eliminate duplicated code from #5095
2022-11-27 13:08:54 +03:00
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.
2022-11-26 23:25:16 +00:00
e247b7400a
Add fixes for PyTorch 1.12.1
...
Fix typo "MasOS" -> "macOS"
If MPS is available and PyTorch is an earlier version than 1.13:
* Monkey patch torch.Tensor.to to ensure all tensors sent to MPS are contiguous
* Monkey patch torch.nn.functional.layer_norm to ensure input tensor is contiguous (required for this program to work with MPS on unmodified PyTorch 1.12.1)
2022-11-21 02:07:19 -05:00
abfa22c16f
Revert "MPS Upscalers Fix"
...
This reverts commit 768b95394a
2022-11-17 00:08:21 -05:00
0ab0a50f9a
change formatting to match the main program in devices.py
2022-11-12 10:00:49 +03:00
1130d5df66
Update devices.py
2022-11-12 11:09:28 +08:00
76ab31e188
Fix wrong mps selection below MasOS 12.3
2022-11-12 11:02:40 +08:00
62e9fec3df
actual better fix
...
thanks C43H66N12O12S2
2022-11-08 15:19:09 -08:00
29eff4a194
terrible hack
2022-11-07 18:06:48 -08:00
cd6c55c1ab
16xx card fix
...
cudnn
2022-11-06 17:05:51 -08:00
faed465a0b
MPS Upscalers Fix
...
Get ESRGAN, SCUNet, and SwinIR working correctly on MPS by ensuring memory is contiguous for tensor views before sending to MPS device.
2022-10-25 09:42:53 +03:00
4c24347e45
Remove BSRGAN from --use-cpu, add SwinIR
2022-10-25 09:42:53 +03:00
50b5504401
remove parsing command line from devices.py
2022-10-22 14:04:14 +03:00
57eb54b838
implement CUDA device selection by ID
2022-10-22 00:11:07 +02:00
fdef8253a4
Add 'interrogate' and 'all' choices to --use-cpu
...
* Add 'interrogate' and 'all' choices to --use-cpu
* Change type for --use-cpu argument to str.lower, so that choices are case insensitive
2022-10-14 16:31:39 +03:00
7349088d32
--no-half-vae
2022-10-10 16:16:29 +03:00
e9e2a7ec9a
Merge branch 'master' into cpu-cmdline-opt
2022-10-04 07:42:53 -04:00
6c6ae28bf5
send all three of GFPGAN's and codeformer's models to CPU memory instead of just one for #1283
2022-10-04 12:32:22 +03:00
27ddc24fde
Add BSRGAN to --add-cpu
2022-10-04 05:18:17 -04:00
eeab7aedf5
Add --use-cpu command line option
...
Remove MPS detection to use CPU for GFPGAN / CodeFormer and add a --use-cpu command line option.
2022-10-04 04:24:35 -04:00
b88e4ea7d6
Merge branch 'master' into master
2022-10-04 01:04:19 -04:00
820f1dc96b
initial support for training textual inversion
2022-10-02 15:03:39 +03:00
bdaa36c844
When device is MPS, use CPU for GFPGAN instead
...
GFPGAN will not work if the device is MPS, so default to CPU instead.
2022-09-30 23:53:25 -04:00
9d40212485
first attempt to produce crrect seeds in batch
2022-09-13 21:49:58 +03:00
c7e0e28ccd
changes for #294
2022-09-12 20:09:32 +03:00
b70b51cc72
Allow TF32 in CUDA for increased performance #279
2022-09-12 16:34:13 +03:00
8fb9c57ed6
add half() supporrt for CLIP interrogation
2022-09-11 23:24:24 +03:00
f194457229
CLIP interrogator
2022-09-11 18:48:36 +03:00
b5d1af11b7
Modular device management
2022-09-11 09:49:43 +03:00
AUTOMATIC1111