5afd75fd98
try-except around preloading items to help troubleshoot issues
2023-03-18 22:27:09 -04:00
f01a7354f0
remove sort from dataset due to slowdown on large sets, add contribution readme
2023-03-18 22:24:03 -04:00
92815a869d
doc update
2023-03-15 22:19:17 -04:00
d26520e67b
Merge branch 'main' of https://github.com/victorchall/EveryDream2trainer
2023-03-15 22:18:01 -04:00
59f14656fe
make sure undersized image is utf8
2023-03-15 22:14:45 -04:00
f665a188f3
Merge pull request #113 from qslug/config-fix
...
Assign default value for max_caption_length
2023-03-15 16:55:11 -04:00
48f132554c
Assign default value for MAX_CAPTION_LENGTH
2023-03-15 21:52:32 +01:00
b9f4a6d657
update vram doc since defaults changed
2023-03-15 12:25:43 -04:00
ba687de8b4
add pbar back to preloading, remove cruft from testing loss stuff
2023-03-15 12:06:29 -04:00
d1bc94fe3e
cruft left from experiment
2023-03-15 11:23:47 -04:00
605716a646
conf
2023-03-15 11:22:40 -04:00
da3c183cc5
autofix exif orientation
2023-03-15 11:12:59 -04:00
ce57954a9d
Merge pull request #112 from nawnie/patch-10
...
update sleep timer for disconnet
2023-03-14 01:29:30 -04:00
493afc3f20
Merge pull request #106 from qslug/enhanced-config
...
Add support for enhanced dataset configuration
2023-03-14 01:28:54 -04:00
20740091d3
update sleep timer for disconnet
...
Discord complaints of last.ckpt not appear may be due to auto disconnect only waiting 3 seconds after training this updates it to 40 which is more then enough time to save to the Gdrive (takes about 28 seconds). This will be safe to accept with out further testing due to only timer being changed.
2023-03-14 00:24:29 -05:00
4f451ac987
Update issue templates
2023-03-13 00:26:18 -04:00
6413ada2e5
Add a test for default rating
2023-03-13 01:01:29 +01:00
e70c6cf743
Make manual model download optional
2023-03-13 00:57:27 +01:00
7e20a74586
Fix cond_dropout and rating handling
2023-03-13 00:36:59 +01:00
ce8f8751c9
Back to xformers cowboy update process
2023-03-12 23:00:09 +01:00
dea8c6e862
Refactor dataset to support multiple main_prompts
...
...rather than multiple whole captions
2023-03-12 19:48:15 +01:00
cf767f38b6
Update issue templates
2023-03-12 13:45:00 -04:00
08a0e7a13e
Update issue templates
2023-03-12 13:35:42 -04:00
9389a90c67
defaulting amp to on now
2023-03-10 21:35:47 -05:00
c913824979
fix every_n_epoch>1 logic and remove unnecessary log
2023-03-10 23:13:53 +01:00
c829521c34
Merge pull request #111 from amirothman/patch-1
...
Update TRAINING.md
2023-03-10 12:50:27 -05:00
1dcc9d501a
Update TRAINING.md
...
--ed1_mode is deprecated so the documentation should reflect this.
2023-03-10 18:03:30 +01:00
e5aa70c54f
update lr recommendation
2023-03-09 21:02:05 -05:00
6c41be3539
Merge pull request #110 from nawnie/main
...
added resume_ckpt from huggingface diffusers
2023-03-09 16:57:50 -05:00
b724398e32
Merge pull request #108 from damian0815/val_starts_from_0
...
make validation more comparable across runs
2023-03-09 16:44:54 -05:00
47e90ad865
round rather than ceil
2023-03-09 20:47:37 +01:00
103ab20696
allow sub-epoch validation when every_n_epochs <1
2023-03-09 20:42:17 +01:00
5ccfdf9d03
Update Train_Colab.ipynb
2023-03-09 03:59:16 -06:00
431e1402d3
Created using Colaboratory
2023-03-09 03:37:17 -06:00
f0111a6e2b
make validation more comparable across runs
2023-03-09 08:21:59 +01:00
4c996cb6b5
Update README.md
2023-03-08 23:47:33 -05:00
0716c40ab6
Add support for enhanced dataset configuration
...
Add support for:
* flip_p.txt
* cond_dropout.txt
* local.yaml consolidated config (including default captions)
* global.yaml consolidated config which applies recursively to subfolders
* flip_p, and cond_dropout config per image
* manifest.json with full image-level configuration
2023-03-08 15:02:14 +01:00
2e3d044ba3
Merge branch 'main' of https://github.com/victorchall/EveryDream2trainer
2023-03-04 15:09:31 -05:00
87ffb5413e
log warning when val/loss is showing upward trend
2023-03-04 15:08:47 -05:00
2531c7bd35
Created using Colaboratory
2023-03-03 19:12:04 -06:00
32585a74b2
Merge pull request #103 from damian0815/feat_text_encoder_LR
...
Separate text encoder LR; logging & pbar improvements; misc sample generator tweaks
2023-03-03 17:46:45 -05:00
29c93eca03
if progress bars are disabled, log a short message instead
2023-03-03 10:50:48 +01:00
644c9e6b2a
log to logging.info instead of stdout
2023-03-03 09:52:44 +01:00
c4d37862ba
document new sample generator params
2023-03-02 23:12:47 +01:00
15187ae2e2
fix log leaking color
2023-03-02 22:53:59 +01:00
e2fd45737d
overwrite args.seed with the actual seed if -1 is passed (so it appears in tensorboard)
...
also improve logging when unet training is disabled
2023-03-02 22:52:33 +01:00
bb1fbb329a
tweak docs
2023-03-02 22:45:22 +01:00
97a8a49773
don't log separate text encoder LR if it's the same as unet LR
2023-03-02 22:36:32 +01:00
ae281976ca
documentation of new text encoder LR and aspect_ratio settings
2023-03-02 22:36:11 +01:00
fe0083877f
add aspect_ratio arg to sample generation
2023-03-02 22:16:21 +01:00
Victor Hall