3fb9d5a85c
fix vae output layer weight cannot cast
2024-07-31 21:20:56 -07:00
0d079a846d
Intergrate Native AutoEncoderKL
2024-07-31 21:10:19 -07:00
2c3afff371
gradio==4.40.0
...
all basic UIs tested
2024-07-31 19:04:02 -07:00
f052fabd4d
make model guess a function that can be patched
2024-07-30 17:26:49 -06:00
c8156fcf41
rework model loader and configs
2024-07-30 13:27:26 -06:00
40dd61ba6c
improve vae key mapping
2024-07-30 09:23:58 -06:00
3289ccb53f
add vae configs
2024-07-30 08:55:22 -06:00
dc62b0d2d7
begin to use new vae impl
2024-07-30 08:43:14 -06:00
abd4d4d83d
unify cast name
2024-07-30 08:42:51 -06:00
9a48c9eff3
rename state_dict method to make it clear
2024-07-30 08:05:36 -06:00
9cb69baf9f
Implement VAE for new backend
2024-07-29 22:26:32 -06:00
e6e18f7edb
Update .gitignore
2024-07-29 17:55:44 -06:00
50da83d985
upgrade diffusers
2024-07-29 17:33:05 -06:00
3bc9402de9
ling
2024-07-29 11:48:55 -06:00
68b672493a
implement attention for new backend
2024-07-29 11:46:16 -06:00
99fcb35506
rename
2024-07-29 11:29:29 -06:00
c4f52d8ffb
add args for new backend
2024-07-29 11:27:58 -06:00
5d291c28ad
implement args for new backend
2024-07-29 11:22:20 -06:00
1b2610db3e
implement stream in new backend
2024-07-29 11:16:59 -06:00
9793b4be0f
implement operations from scratch
2024-07-29 10:59:16 -06:00
6b163843fc
add backend folder
2024-07-29 10:37:18 -06:00
4a9c8f32ee
remove train tab
...
This tab has been broken since 6 months ago and no one even creates an issue for it ; this indicates no one uses it.
Also just tried original webui Train tab and it seems also broken even in Auto's repo.
so we remove it to make room for new tabs and new features.
2024-07-29 10:02:43 -06:00
c3e17f05f3
fix #880
2024-07-29 09:22:44 -06:00
0bfc3beb36
Update README.md
2024-07-29 08:20:13 -06:00
37386592ad
version
2024-07-28 08:08:28 -06:00
918c6af162
upgrade js canvas for #881
2024-07-28 08:07:37 -06:00
f3926ee3b7
resolve t2i upscale button #882 #874
2024-07-28 07:59:53 -06:00
2f87f123b2
less aggressive clip skip
...
to make CivitAI pony image meta works better
2024-07-27 13:23:39 -07:00
6eb23c6d09
fix #861
2024-07-27 07:52:20 -07:00
7e31721803
fix #861
2024-07-27 07:50:59 -07:00
3ba632b64b
version
2024-07-27 06:54:05 -07:00
f94bcae7b0
fix img2img batch upload
2024-07-27 06:53:37 -07:00
d4e7639233
sync change log
2024-07-27 06:47:04 -07:00
04c6ab68bb
1.10.1
2024-07-27 06:39:58 -07:00
8c7da52e01
Update Forge Status
2024-07-27 06:19:21 -07:00
0ba9f7f399
Clip Skip to All Models
...
Technically Correct Implementation of CLIP skip for all models.
Automatic1111 uses a weird logic to ignore SDXL clip skip but added an option to only set CLIP-L Skip with an offset 1, which means if one set skip as 3, then the SDXL CLIP G does not skip but SDXL CLIP L uses 3-1=2 as skip (not 3?!)
Forge now uses technically correct and consistent clip skip for all models. Forge now outweigh technical correctness over reproducing Auto results.
But one can get same results by just do not set clip skip for SDXL.
2024-07-27 05:46:53 -07:00
33e381f16a
align with SAG ui
2024-07-26 15:19:24 -07:00
f6a0c69f0f
fix ADetailer
2024-07-26 15:11:12 -07:00
1673a5ac2d
Perturbed Attention Guidance Integrated
2024-07-26 15:00:31 -07:00
1d185584a7
move hypertile, svd, z123 to legacy extensions
...
https://github.com/lllyasviel/forge-legacy-extensions
2024-07-26 14:28:47 -07:00
bc22a930e5
Update README.md
2024-07-26 13:49:15 -07:00
49ca78a0fa
do not ignore pyi
2024-07-26 13:08:47 -07:00
c2bfa6f6a3
fix gradio
2024-07-26 13:08:15 -07:00
e1082b8799
fix loading order
2024-07-26 12:57:14 -07:00
bb209ae2b2
Update README.md
2024-07-26 12:49:24 -07:00
46ac4bec95
Update README.md
2024-07-26 12:47:14 -07:00
e26abf87ec
Gradio 4 + WebUI 1.10
2024-07-26 12:02:46 -07:00
e95333c556
Update README.md
2024-07-26 07:26:18 -07:00
42ee0b7388
add ESRGAN folder
2024-07-26 07:20:29 -07:00
a9e0c38700
Update README.md
2024-07-21 23:34:37 -07:00
layerdiffusion