Archives/imaginAIry
2
0
Fork 0
mirror of https://github.com/brycedrennan/imaginAIry synced 2024-11-05 12:00:15 +00:00
Code Issues Packages Projects Releases Wiki Activity
128 Commits 10 Branches 104 Tags 322 MiB
7ae77faf07
Commit Graph

128 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Bryce
cc014685bd fix: leave sample in float32 for all final processing 2022-09-22 10:33:35 -07:00
Bryce
09f00f9652 version: 1.6.1 2022-09-22 08:26:55 -07:00
Bryce
9614a82f12 fix: use float32 at image render time for wider cpu compatibility 
If the x_sample was a bfloat on the gpu but the cpu doesn't support bfloat, that can cause a TypeError
```
  File "/home/stdiff/.local/lib/python3.10/site-packages/imaginairy/api.py", line 292, in imagine
    x_sample.cpu().numpy(), "c h w -> h w c"

TypeError: Got unsupported ScalarType BFloat16`
```
 2022-09-22 08:25:06 -07:00
Bryce Drennan
08fca72033
Merge pull request #18 from brycedrennan/bugfixes 
Bugfixes + per-prompt tile mode
 2022-09-21 22:45:05 -07:00
Bryce
a275d9c9cd version: 1.6.0 + docs 2022-09-21 22:44:49 -07:00
Bryce
cdfeaa4c6f tests: add docker image for testing environment. minor test improvements 2022-09-21 22:38:44 -07:00
Bryce
09bc1c70e6 fix: try to address #13 BFloat16 issue 
Seems to be caused by incompatible types in group_norm when we use autocast.

Patch group_norm to cast the weights to the same type as the inputs

From what I can understand all the other repos just switch to full precision instead
of addressing this.  I think this would make things slower but I'm not sure. So maybe
the patching solution I'm doing is better?

https://github.com/pytorch/pytorch/pull/81852
 2022-09-21 22:03:12 -07:00
Bryce
e23e363bf5 lint 2022-09-21 09:14:39 -07:00
Bryce
ff7fc0dab7 feature: tile mode can now be specified per-prompt 2022-09-20 22:57:03 -07:00
Bryce
41b0ae99b3 version 1.5.4 
minor changes to help with google colab compat
 2022-09-20 21:16:09 -07:00
Bryce
6f77c23c1d version 1.5.4 
minor changes to help with google colab compat
 2022-09-20 21:15:19 -07:00
Bryce
83c809e2b7 version 1.5.3 
bugfix missing config file
 2022-09-20 17:05:11 -07:00
Bryce
3d64867e1f version 1.5.2 
bugfix missing config file
 2022-09-20 17:03:33 -07:00
Bryce
37db8db1a2 version 1.5.1 2022-09-20 08:42:27 -07:00
Bryce
a46424c673 feature: img2img now supported with PLMS (instead of just DDIM) 
Kinda hacky copy/pasting from ddim.  Need to cleanup
 2022-09-20 08:42:00 -07:00
Bryce
3a10a2eb80 docs: osx => macOS 2022-09-20 07:17:04 -07:00
Bryce Drennan
ced6ec8151
Merge pull request #11 from brycedrennan/describe-images 
feature: generate captions for images
 2022-09-19 21:20:20 -07:00
Bryce
4705d182d5 feature: generate captions for images 
- add wip functionality for negative masks
- ci: add code linter that removes unused imports
- add instructions to install rust on osx
 2022-09-19 21:19:22 -07:00
Bryce
60a04a5d68 version: 1.4.0 
we can get the stem of the pear
 2022-09-18 15:46:48 -07:00
Bryce
f040e3bffe feature: allow multiple additive targets for text masking targets 2022-09-18 15:24:31 -07:00
Bryce
a0087c1304 Merge branch 'clip-masking' 2022-09-18 06:12:18 -07:00
Bryce
be85133070 version: 1.3.0 2022-09-18 06:11:56 -07:00
Bryce Drennan
a5e010d01f
Merge pull request #9 from brycedrennan/clip-masking 
Clip masking
 2022-09-18 06:10:39 -07:00
Bryce
930295d840 feature: txt2mask - automated text replacement 
from https://github.com/timojl/clipseg
 2022-09-18 06:07:47 -07:00
Bryce
7087c4a680 feature: automatic mask generation 
mask generation working but not integrated
 2022-09-17 17:08:48 -07:00
Bryce
8238e59067 tests: add more tests 2022-09-17 15:49:38 -07:00
Bryce
d7cbf6e416 refactor: cleanup ddim 
- delete more unused code
- fix some lints
 2022-09-17 14:02:27 -07:00
Bryce
6307a0daf5 refactor: cleanup ddim 2022-09-17 12:24:27 -07:00
Bryce
89908603cf fix: don't repeatedly download the same url 2022-09-16 22:34:42 -07:00
Bryce
59648dbe61 test: add some tests/logging 
add some experiments
 2022-09-16 22:21:38 -07:00
Bryce
19d0b563ac style: lint fixes. remove unused code 2022-09-16 09:24:24 -07:00
Bryce
793df6e9dc docs update 2022-09-16 08:35:38 -07:00
Bryce
33f1e2aeec version 1.2.0 
- add url init images
- don't install protobuf versions that are causing problems
https://github.com/brycedrennan/imaginAIry/issues/7
https://github.com/protocolbuffers/protobuf/issues/10571
 2022-09-16 08:20:29 -07:00
Bryce
af3dc6bacd fix: restrict usage of difficult protobuf versions 2022-09-16 08:18:50 -07:00
Bryce Drennan
6b4c348132
Merge pull request #6 from brycedrennan/bd/init-img-url 
feature: urls as init images
 2022-09-15 23:24:21 -07:00
Bryce
c7a822d701 feature: urls as init images 
- --init-image accepts urls
- cleanup command line code
 2022-09-15 23:06:59 -07:00
Bryce
47db34647b version 1.1.4 2022-09-15 07:55:41 -07:00
Bryce
23f0e2be12 fix: select ddim for img2img 2022-09-15 07:55:09 -07:00
Bryce
da0ba308b2 version: 1.1.3 
fix compat with python 3.8,3.9
 2022-09-15 07:51:38 -07:00
Bryce Drennan
8c064587d9
Merge pull request #1 from wseagar/ws/nsfw_fix 
fix: is_nsfw_img is None rather than false
 2022-09-15 07:46:25 -07:00
wseagar
56fc414724 fix: is_nsfw_img is None rather than false 2022-09-15 20:27:29 +12:00
Bryce
a1675fa599 docs: add dockerfile example 2022-09-14 21:24:45 -07:00
Bryce
fa3673ef56 fix: use context manager syntax compatible with python 3.8 
parentheses supported in python 3.10
 2022-09-14 21:24:05 -07:00
Bryce
51a8b1b8d8 fix: filter logic was wrong 2022-09-14 20:01:12 -07:00
Bryce
0834db63c8 version 1.1.1 2022-09-14 19:44:37 -07:00
Bryce
967eb76365 refactor/test: logging suppression + hashed image test 
- simpler logging suppression for `transformers` library
- suppress logging noise for running tests
- get test running for all samplers on mps and cuda platforms
- refactor safety model env variable to allow classification
 2022-09-14 19:40:50 -07:00
Bryce
8c88f495d2 version 2022-09-14 09:39:25 -07:00
Bryce
bb665b9eb6 fix: make k-diffusion samplers deterministic 
- add test for hashes on mps.  images look same on CUDA but are slightly different.
 2022-09-14 09:37:45 -07:00
Bryce
b4a3b8c2b3 feature: k-diffusion samplers 
- improved image logging functionality. can just stick log_latent wherever you want
- improved some variable naming
- moved all the samplers together
- vendored k-diffusion library
 2022-09-14 00:40:25 -07:00
Bryce
20ac04d9df fix: missing data file 2022-09-13 10:44:08 -07:00