c65909ad16
feat(api): return more data for embeddings
3 years ago
311354c0bb
fix the issue with training on SD2.0
3 years ago
bdbe09827b
changed embedding accepted shape detection to use existing code and support the new alt-diffusion model, and reformatted messages a bit #6149
3 years ago
f55ac33d44
validate textual inversion embeddings
3 years ago
3bf5591efe
fix F541 f-string without any placeholders
3 years ago
c0355caefe
Fix various typos
3 years ago
c9a2cfdf2a
Merge branch 'master' into racecond_fix
3 years ago
a2feaa95fc
Merge pull request #5194 from brkirch/autocast-and-mps-randn-fixes
...
Use devices.autocast() and fix MPS randn issues
3 years ago
119a945ef7
Fix divide by 0 error
...
Fix of the edge case 0 weight that occasionally will pop up in some specific situations. This was crashing the script.
3 years ago
4d5f1691dd
Use devices.autocast instead of torch.autocast
3 years ago
39827a3998
Merge pull request #4688 from parasi22/resolve-embedding-name-in-filewords
...
resolve [name] after resolving [filewords] in training
3 years ago
b48b7999c8
Merge remote-tracking branch 'flamelaw/master'
3 years ago
755df94b2a
set TI AdamW default weight decay to 0
3 years ago
ce6911158b
Add support Stable Diffusion 2.0
3 years ago
89d8ecff09
small fixes
3 years ago
5b57f61ba4
fix pin_memory with different latent sampling method
3 years ago
c81d440d87
moved deepdanbooru to pure pytorch implementation
3 years ago
2d22d72cda
fix random sampling with pin_memory
3 years ago
a4a5735d0a
remove unnecessary comment
3 years ago
bd68e35de3
Gradient accumulation, autocast fix, new latent sampling method, etc
3 years ago
89daf778fb
Merge pull request #4812 from space-nuko/feature/interrupt-preprocessing
...
Add interrupt button to preprocessing
3 years ago
cdc8020d13
change StableDiffusionProcessing to internally use sampler name instead of sampler index
3 years ago
c8c40c8a64
Add interrupt button to preprocessing
3 years ago
9a1aff645a
resolve [name] after resolving [filewords] in training
3 years ago
73776907ec
Merge pull request #4117 from TinkTheBoush/master
...
Adding optional tag shuffling for training
3 years ago
a1e271207d
Update dataset.py
3 years ago
b19af67d29
Update dataset.py
3 years ago
13a2f1dca3
adding tag drop out option
3 years ago
d85c2cb2d5
Merge branch 'master' into gradient-clipping
3 years ago
8011be33c3
move functions out of main body for image preprocessing for easier hijacking
3 years ago
bb832d7725
Simplify grad clip
3 years ago
821e2b883d
change option position to Training setting
3 years ago
39541d7725
Fixes race condition in training when VAE is unloaded
...
set_current_image can attempt to use the VAE when it is unloaded to
the CPU while training
3 years ago
237e79c77d
Merge branch 'master' into gradient-clipping
3 years ago
af6fba2475
Merge branch 'master' into master
3 years ago
cffc240a73
fixed textual inversion training with inpainting models
3 years ago
467cae167a
append_tag_shuffle
3 years ago
890e68aaf7
Fixed minor bug
...
when unloading vae during TI training, generating images after
training will error out
3 years ago
3b0127e698
Merge branch 'master' of https://github.com/AUTOMATIC1111/stable-diffusion-webui into TI_optimizations
3 years ago
006756f9cd
Added TI training optimizations
...
option to use xattention optimizations when training
option to unload vae when training
3 years ago
cd4d59c0de
Merge master
3 years ago
17a2076f72
Merge pull request #3928 from R-N/validate-before-load
...
Optimize training a little
3 years ago
3d58510f21
Fix dataset still being loaded even when training will be skipped
3 years ago
a07f054c86
Add missing info on hypernetwork/embedding model log
...
Mentioned here: https://github.com/AUTOMATIC1111/stable-diffusion-webui/discussions/1528#discussioncomment-3991513
Also group the saving into one
3 years ago
ab05a74ead
Revert "Add cleanup after training"
...
This reverts commit 3ce2bfdf95
3 years ago
a27d19de2e
Additional assert on dataset
3 years ago
3ce2bfdf95
Add cleanup after training
3 years ago
ab27c111d0
Add input validations before loading dataset for training
3 years ago
ef4c94e1cf
Improve lr schedule error message
3 years ago
a5f3adbdd7
Allow trailing comma in learning rate
3 years ago
05e2e40537
Merge branch 'master' into gradient-clipping
3 years ago
810e6a407d
Merge pull request #3858 from R-N/log-csv
...
Fix log off by 1 #3847
3 years ago
9ceef81f77
Fix log off by 1
3 years ago
16451ca573
Learning rate sched syntax support for grad clipping
3 years ago
1618df41ba
Gradient clipping for textual embedding
3 years ago
a0a7024c67
Fix random dataset shuffle on TI
3 years ago
737eb28fac
typo: cmd_opts.embedding_dir to cmd_opts.embeddings_dir
3 years ago
f4e1464217
Implement PR #3625 but for embeddings.
3 years ago
4875a6c217
Implement PR #3309 but for embeddings.
3 years ago
c2dc9bfa89
Implement PR #3189 but for embeddings.
3 years ago
cbb857b675
enable creating embedding with --medvram
3 years ago
df0c5ea29d
update default weights
3 years ago
54f0c14824
download better face detection module dynamically
3 years ago
db8ed5fe5c
Focal crop UI elements
3 years ago
6629446a2f
Merge branch 'master' into focal-point-cropping
3 years ago
3e6c2420c1
improve debug markers, fix algo weighting
3 years ago
18f86e41f6
Removed two unused imports
3 years ago
1be5933ba2
auto cropping now works with non square crops
3 years ago
f49c08ea56
prevent error spam when processing images without txt files for captions
3 years ago
5e9afa5c8a
Merge branch 'master' into fix/train-preprocess-keep-ratio
3 years ago
306e2ff6ab
Update image_embedding.py
3 years ago
d0ea471b0c
Use opts in textual_inversion image_embedding.py for dynamic fonts
3 years ago
7d6b388d71
Merge branch 'ae'
3 years ago
0c5522ea21
Merge branch 'master' into training-help-text
3 years ago
b69c37d25e
Allow datasets with only 1 image in TI
3 years ago
8f59129847
Some changes to the tensorboard code and hypernetwork support
3 years ago
a6d593a6b5
Fixed a typo in a variable
3 years ago
85dd62c4c7
train: ui: added `Split image threshold` and `Split image overlap ratio` to preprocess
3 years ago
9681419e42
train: fixed preprocess image ratio
3 years ago
29e74d6e71
Add support for Tensorboard for training embeddings
3 years ago
0ddaf8d202
improve face detection a lot
3 years ago
858462f719
do caption copy for both flips
3 years ago
59ed744383
face detection algo, configurability, reusability
...
Try to move the crop in the direction of a face if it is present
More internal configuration options for choosing weights of each of the algorithm's findings
Move logic into its module
3 years ago
9b65c4ecf4
pass preprocess_txt_action param
3 years ago
fbcce66601
add existing caption file handling
3 years ago
c3835ec85c
pass overwrite old flag
3 years ago
0087079c2d
allow overwrite old embedding
3 years ago
41e3877be2
fix entropy point calculation
3 years ago
abeec4b630
Add auto focal point cropping to Preprocess images
...
This algorithm plots a bunch of points of interest on the source
image and averages their locations to find a center.
Most points come from OpenCV. One point comes from an
entropy model. OpenCV points account for 50% of the weight and the
entropy based point is the other 50%.
The center of all weighted points is calculated and a bounding box
is drawn as close to centered over that point as possible.
3 years ago
1997ccff13
Merge branch 'master' into test_resolve_conflicts
3 years ago
62edfae257
print list of embeddings on reload
3 years ago
ae0fdad64a
Merge branch 'master' into test_resolve_conflicts
3 years ago
9324cdaa31
ui fix, re organization of the code
3 years ago
0c5fa9a681
do not reload embeddings from disk when doing textual inversion
3 years ago
97ceaa23d0
Merge branch 'master' into test_resolve_conflicts
3 years ago
b6e3b96dab
Change vector size footer label
3 years ago
ddf6899df0
generalise to popular lossless formats
3 years ago
9a1dcd78ed
add webp for embed load
3 years ago
939f16529a
only save 1 image per embedding
3 years ago
9e846083b7
add vector size to embed text
3 years ago
Philpax
AUTOMATIC
Vladimir Mandic
Yuval Aboulafia
Jim Hays
PhytoEpidemic
brkirch
flamelaw
space-nuko
parasi
KyuSeok Jung
Muhammad Rizqi Nur
TinkTheBoush
Fampai
Nerogar
FlameLaw
DepFA
timntorres
captin411
Melan
guaneec
Milly
MalumaDev