Mirrored_Repos/InvokeAI
1
0
Fork 0
mirror of https://github.com/invoke-ai/InvokeAI synced 2024-08-30 20:32:17 +00:00
Code Issues Packages Projects Releases Wiki Activity
7,182 Commits 274 Branches 239 Tags 397 MiB
421f5b7d75
Commit Graph

450 Commits

Author SHA1 Message Date
blessedcoolant
b229fe19aa
Merge branch 'main' into lstein/configure-max-cache-size 2023-07-07 01:52:12 +12:00
Sergey Borisov
04b57c408f Add clip skip option to prompt node 2023-07-06 16:09:40 +03:00
blessedcoolant
6f1268e2b1
Merge branch 'main' into lstein/more-model-loading-fixes 2023-07-07 00:32:22 +12:00
Lincoln Stein
8f5fcb188c
Merge branch 'main' into lstein/model-manager-router-api 2023-07-05 23:16:43 -04:00
Lincoln Stein
f7daa6e71d all methods now return OPENAPI_MODEL_CONFIGS; convert uses PUT 2023-07-05 23:13:01 -04:00
Lincoln Stein
a7cbcae176 expose max_cache_size to invokeai-configure interface 2023-07-05 20:59:57 -04:00
Lincoln Stein
0a6dccd607 expose max_cache_size to invokeai-configure interface 2023-07-05 20:59:14 -04:00
Lincoln Stein
43c51ff157
Merge branch 'main' into lstein/more-model-loading-fixes 2023-07-05 20:48:15 -04:00
Lincoln Stein
cfa3b2419c partial implementation of merge 2023-07-05 20:25:47 -04:00
Lincoln Stein
d4550b3059 clean up lint errors in lora.py 2023-07-05 19:18:25 -04:00
Lincoln Stein
83d3a043da merge latest changes from main 2023-07-05 19:15:53 -04:00
Lincoln Stein
71dad6d404
Merge branch 'main' into ti-ui 2023-07-05 16:57:31 -04:00
Lincoln Stein
685a47cc7d fix crash during lora application 2023-07-05 16:40:47 -04:00
Lincoln Stein
f8bbec8572 Recognize and load diffusers-style LoRAs (.bin) 
Prevent double-reporting of autoimported models
- closes #3636

Allow autoimport of diffusers-style LoRA models
- closes #3637
 2023-07-05 16:21:23 -04:00
Lincoln Stein
863336acbb Recognize and load diffusers-style LoRAs (.bin) 
Prevent double-reporting of autoimported models
- closes #3636

Allow autoimport of diffusers-style LoRA models
- closes #3637
 2023-07-05 16:19:16 -04:00
Lincoln Stein
90ae8ce26a prevent model install crash "torch needs to be restarted with spawn" 2023-07-05 16:18:20 -04:00
Lincoln Stein
ad5d90aca8 prevent model install crash "torch needs to be restarted with spawn" 2023-07-05 15:38:07 -04:00
Lincoln Stein
5b6dd47b9f add API for model convert 2023-07-05 15:13:21 -04:00
Lincoln Stein
5027d0a603 accept @psychedelicious suggestions above 2023-07-05 14:50:57 -04:00
Lincoln Stein
9f9ce08e44
Merge branch 'main' into lstein/remove-hardcoded-cuda-device 2023-07-05 13:38:33 -04:00
blessedcoolant
9e2d63ef97
Merge branch 'main' into fix/ckpt_convert_scan 2023-07-06 05:01:34 +12:00
Sergey Borisov
0ac9dca926 Fix loading diffusers ti 2023-07-05 19:46:00 +03:00
Lincoln Stein
bd82c4ace0 model installer confirms deletion of models 2023-07-05 09:57:23 -04:00
Lincoln Stein
9edf78dd2e merge with main 2023-07-05 09:12:54 -04:00
Lincoln Stein
6112197edf convert implemented; need router 2023-07-05 09:05:05 -04:00
Sergey Borisov
ee042ab76d Fix ckpt scanning on conversion 2023-07-05 14:18:30 +03:00
Sergey Borisov
2beb8f049e Fix model detection 2023-07-05 09:43:46 +03:00
blessedcoolant
639d88afd6 revert: inference_mode to no_grad 2023-07-05 16:39:15 +12:00
blessedcoolant
c0501ed5c2 fix: Slow loading of Loras 
Co-Authored-By: StAlKeR7779 <7768370+StAlKeR7779@users.noreply.github.com>
 2023-07-05 12:47:34 +10:00
Lincoln Stein
5d099f4a49 update_model working 2023-07-04 17:26:57 -04:00
Lincoln Stein
752b4d50cf model_delete method now working 2023-07-04 10:40:32 -04:00
Lincoln Stein
c1c49d9a76 import model returns 404 for invalid path, 409 for duplicate model 2023-07-04 10:08:10 -04:00
Lincoln Stein
96bf92ead4 add the import model router 2023-07-04 14:35:47 +10:00
Lincoln Stein
fc419546bc
Merge branch 'main' into lstein/remove-hardcoded-cuda-device 2023-07-03 14:10:47 -04:00
Lincoln Stein
d6de11bd56 resolve merge conflict 2023-07-03 12:19:11 -04:00
Lincoln Stein
ed86d0b708 Union[foo, None]=>Optional[foo] 2023-07-03 12:17:45 -04:00
Lincoln Stein
fb2b2a371d Merge branch 'lstein/fix-vae-conversion-crash' into release/invokeai-3-0-alpha 2023-07-03 11:21:16 -04:00
Lincoln Stein
10d513c5f7 add runtime root path to relative vaes and other submodels 2023-07-03 11:19:33 -04:00
Lincoln Stein
877b187a1b Merge branch 'lstein/restore-3.9-compatibility' into release/invokeai-3-0-alpha 2023-07-03 11:01:34 -04:00
Lincoln Stein
ac9ec4e75a restore 3.9 compatibility by replacing | with Union[] 2023-07-03 10:57:40 -04:00
Lincoln Stein
2465c7987b Revert "restore 3.9 compatibility by replacing | with Union[]" 
This reverts commit 76bafeb99e.
 2023-07-03 10:56:41 -04:00
Lincoln Stein
76bafeb99e restore 3.9 compatibility by replacing | with Union[] 2023-07-03 10:55:04 -04:00
Lincoln Stein
6935858ef3 add debugging messages to aid in memory leak tracking 2023-07-02 13:34:53 -04:00
Lincoln Stein
3c2ce51f10 Merge branch 'lstein/remove-hardcoded-cuda-device' into release/invokeai-3-0-alpha 2023-07-01 21:15:58 -04:00
Lincoln Stein
0f02915012 remove hardcoded cuda device in model manager init 2023-07-01 21:15:42 -04:00
Lincoln Stein
f1928d2588 prevent crashes on malformed models 2023-07-01 14:32:58 -04:00
blessedcoolant
c74bb5cdbf
Merge branch 'main' into lstein/fix-vae-convert 2023-07-01 11:18:21 +12:00
Lincoln Stein
1347fc2f00 fix incorrect VAE config file path during conversion of ckpts 2023-06-30 19:14:06 -04:00
Lincoln Stein
ace4f6d586 fix duplicate model key addition when root directory is a relative path 2023-06-28 17:02:03 -04:00
StAlKeR7779
ac46b129bf
Merge branch 'main' into feat/lora_model_patch 2023-06-28 22:43:58 +03:00
Lincoln Stein
79fc708580 warn but do not crash when model scan finds random cruft in models directory 2023-06-28 15:26:42 -04:00
Lincoln Stein
e8ed0fad6c autoimport from embedding/controlnet/lora folders designated in startup file 2023-06-27 12:30:53 -04:00
Lincoln Stein
823e098b7c prompt user for prediction type when autoimporting a v2 model without .yaml file 
don't ask user for prediction type of a config.yaml provided
 2023-06-26 16:30:34 -04:00
Lincoln Stein
011adfc958 merge with main 2023-06-26 13:53:59 -04:00
Lincoln Stein
befd95eb19 rename root_dir to root_path attributes to emphasize return of a Path 2023-06-26 13:52:25 -04:00
Lincoln Stein
a2ddb3823b fix add_model() logic 2023-06-26 13:33:38 -04:00
Sergey Borisov
91c3a58fb6 Fix lycoris layers init 2023-06-26 04:33:37 +03:00
Sergey Borisov
5cebf67ee4 Apply lora by patching lora instead of hooks 2023-06-26 03:57:33 +03:00
Sergey Borisov
1ba94a92b3 Fixes 2023-06-26 03:54:42 +03:00
Sergey Borisov
23c22ac933 Refactor logic/small fixes 2023-06-26 03:07:54 +03:00
Lincoln Stein
160b5d7992 add support for an autoimport models directory scanned at startup time 2023-06-25 18:50:15 -04:00
Lincoln Stein
c91d1eacba Merge branch 'lstein/installer-for-new-model-layout' of github.com:invoke-ai/InvokeAI into lstein/installer-for-new-model-layout 2023-06-25 16:04:48 -04:00
Lincoln Stein
60b37b7ff4 fix model manager documentation 2023-06-25 16:04:43 -04:00
Sergey Borisov
a3c22b5fe6 Remove upcast_attention and prediction_type from stable diffusion model logic, fix ckpt conversion according to this 2023-06-25 21:06:22 +03:00
Lincoln Stein
c3c4a71173 implemented Stalker's suggested improvements 2023-06-24 12:37:26 -04:00
Lincoln Stein
ba1371a88f rename ModelType.Pipeline to ModelType.Main 2023-06-24 11:45:49 -04:00
Lincoln Stein
539d1f3bde remove redundant prediction_type and attention_upscaling flags 2023-06-23 16:54:52 -04:00
Lincoln Stein
466ec3ab5e add router API support for model manager heuristic_import()` 2023-06-23 16:35:39 -04:00
Lincoln Stein
58d1857ab6 merge with main 2023-06-23 13:57:25 -04:00
Lincoln Stein
56bd873d7a make relative model paths work in model manager 2023-06-23 10:52:59 -04:00
Sergey Borisov
5aaaaf64a1 Fix ckpt conversion 2023-06-23 17:29:54 +03:00
StAlKeR7779
9140e2c0f2
Merge branch 'main' into fix/vae_conversion 2023-06-23 15:03:59 +03:00
Lincoln Stein
c7b7e087e4
Merge branch 'main' into lstein/installer-for-new-model-layout 2023-06-23 01:45:05 +01:00
blessedcoolant
bb85608890 Merge branch 'main' into feat/onnx 2023-06-23 05:18:41 +12:00
Sergey Borisov
6c7668aaca Update onnx model structure, change code according 2023-06-22 20:03:17 +03:00
psychedelicious
b937b7da01 feat(models): update model manager service & route to return list of models 2023-06-22 17:34:12 +10:00
Sergey Borisov
21245a0fb2 Set model type to const value in openapi schema, add model format enums to model schema(as they not not referenced in case of Literal definition) 2023-06-22 16:51:53 +10:00
Sergey Borisov
da566b59e8 Update model format field to use enums 2023-06-22 16:51:53 +10:00
Sergey Borisov
e4dc9c5a04 Rename format to model_format(still named format when work with config) 2023-06-22 16:51:53 +10:00
Sergey Borisov
aceadacad4 Remove default model logic 2023-06-22 16:51:53 +10:00
blessedcoolant
727293d722 fix: 2.1 models breaking generation 
Co-Authored-By: StAlKeR7779 <7768370+StAlKeR7779@users.noreply.github.com>
 2023-06-22 16:42:59 +10:00
Sergey Borisov
ef83a2fffe Add name, base_mode, type fields to model info 2023-06-22 16:42:51 +10:00
Sergey Borisov
01d17601b8 Generate config names for openapi 2023-06-22 16:41:19 +10:00
blessedcoolant
bf0d5f4cfc fix: Update missing name types to new names 2023-06-22 16:41:02 +10:00
blessedcoolant
9838dda1b7 chore: Update model config type names 2023-06-22 16:40:40 +10:00
Sergey Borisov
7759b3f75a Small refactor 2023-06-21 04:24:25 +03:00
Sergey Borisov
4d337f6abc ONNX Model/runtime first implementation 2023-06-21 02:12:21 +03:00
Lincoln Stein
90df316835
Merge branch 'main' into lstein/installer-for-new-model-layout 2023-06-20 22:50:41 +01:00
Lincoln Stein
ac6403f877 address some of ebr issues 2023-06-20 11:08:27 -04:00
Sergey Borisov
92c86fd0b8 Set model type to const value in openapi schema, add model format enums to model schema(as they not not referenced in case of Literal definition) 2023-06-20 03:44:58 +03:00
Sergey Borisov
46dc751139 Update model format field to use enums 2023-06-20 03:30:09 +03:00
Sergey Borisov
4cefe37723 Rename format to model_format(still named format when work with config) 2023-06-20 03:25:08 +03:00
Sergey Borisov
82b73c50a0 Remove default model logic 2023-06-20 03:13:10 +03:00
blessedcoolant
b0c4451324 Merge branch 'main' into model-manager-ui-30 2023-06-19 23:02:59 +12:00
Sergey Borisov
7b35162b9e Remove old logic except for inpaint, add support for lora and ti to inpaint node 2023-06-19 15:57:28 +10:00
Sergey Borisov
82091b9a66 Fix vae conversion 2023-06-18 23:46:07 +03:00
Lincoln Stein
e1d53b86f3
Merge branch 'main' into lstein/installer-for-new-model-layout 2023-06-17 16:26:56 -07:00
blessedcoolant
bf0577c882 fix: 2.1 models breaking generation 
Co-Authored-By: StAlKeR7779 <7768370+StAlKeR7779@users.noreply.github.com>
 2023-06-18 08:26:25 +12:00
Sergey Borisov
dc669d1447 Add name, base_mode, type fields to model info 2023-06-17 22:48:44 +03:00
Sergey Borisov
16dc78f6c6 Generate config names for openapi 2023-06-17 17:15:36 +03:00
blessedcoolant
c8dfa49d86 fix: Update missing name types to new names 2023-06-17 22:04:28 +12:00
blessedcoolant
67d05d2066 chore: Update model config type names 2023-06-17 21:28:43 +12:00
Lincoln Stein
f28d50070e configure/install basically working; needs edge case testing 2023-06-16 22:54:36 -04:00
Lincoln Stein
ada7399753 rewrite of widget display - marshalling needs rewrite 2023-06-15 23:32:33 -04:00
Sergey Borisov
5f2d07917d Fix lora import, fix sd2 config, fix list models api 2023-06-15 21:30:15 +03:00
Sergey Borisov
6c5954f9d1 Add controlnet to model manager, fixes 2023-06-14 04:26:21 +03:00
Sergey Borisov
740c05a0bb Save models on rescan, uncache model on edit/delete, fixes 2023-06-14 03:12:12 +03:00
Sergey Borisov
26090011c4 Fix conflict resolve, add model configs to type annotation 2023-06-14 00:26:37 +03:00
Sergey Borisov
e7db6d8120 Fix ckpt and vae conversion, migrate script, remove sd2-base 2023-06-13 18:05:12 +03:00
Lincoln Stein
87ba17a1f5 add migration script and update convert and face restoration paths 2023-06-13 01:27:51 -04:00
Lincoln Stein
1439dc7712 Add SchedulerPredictionType and ModelVariantType enums 2023-06-12 16:07:04 -04:00
Sergey Borisov
36eb1bd893 Fixes 2023-06-12 16:14:09 +03:00
Sergey Borisov
9fa78443de Fixes, add sd variant detection 2023-06-12 05:52:30 +03:00
Lincoln Stein
893f776f1d model_probe working; model_install incomplete 2023-06-11 19:51:53 -04:00
Lincoln Stein
085ab54124 remove modified models.py and migrate code to models/base.py 2023-06-11 16:10:15 -04:00
Lincoln Stein
8e1a56875e remove defunct code 2023-06-11 12:57:06 -04:00
Lincoln Stein
000626ab2e move all installation code out of model_manager 2023-06-11 12:51:50 -04:00
Sergey Borisov
694fd0c92f Fixes, first runable version 2023-06-11 16:42:40 +03:00
Sergey Borisov
738ba40f51 Fixes 2023-06-11 06:12:21 +03:00
Sergey Borisov
3ce3a7ee72 Rewrite model configs, separate models 2023-06-11 04:49:09 +03:00
Lincoln Stein
74b43c9bdf fix incorrect variable/typenames in model_cache 2023-06-10 10:41:48 -04:00
Lincoln Stein
a87d52a389 resolve conflicts between lstein & sttalker changes 2023-06-10 09:59:19 -04:00
Lincoln Stein
959e64c9b3 start removing repo_id support 2023-06-10 09:57:23 -04:00
Sergey Borisov
2c056ead42 New models structure draft 2023-06-10 03:14:10 +03:00
Lincoln Stein
887576d217 add directory scanning for loras, controlnets and textual_inversions 2023-06-08 23:11:53 -04:00
Lincoln Stein
6652f3405b merge with main 2023-06-08 21:08:43 -04:00
Lincoln Stein
2a6d11e645 create databases directory on startup 2023-06-08 07:17:54 -04:00
Lincoln Stein
9ed86a08f1 multiple small fixes 
1. Contents of autoscan directory field are restored after doing an installation.
2. Activate dialogue to choose V2 parameterization when importing from a directory.
3. Remove autoscan directory from init file when its checkbox is unselected.
4. Add widget cycling behavior to install models form.
 2023-06-07 17:32:00 -04:00
Lincoln Stein
04f9757f8d prevent crash when trying to calculate size of missing safety_checker 
- Also fixed up order in which logger is created in invokeai-web
  so that handlers are installed after command-line options are
  parsed (and not before!)
 2023-06-06 22:57:49 -04:00
Lincoln Stein
1f9e1eb964 merge with main 2023-06-06 22:18:41 -04:00
Lincoln Stein
d8d11f9bbb quench fp16 rev id not found warning 2023-06-06 22:01:05 -04:00
Lincoln Stein
f5044c290d fix crash during model conversion 2023-06-06 17:05:29 -04:00
Lincoln Stein
90333c0074 merge with main 2023-06-05 22:03:44 -04:00
Lincoln Stein
cb157ea530 fix crash when install-models launched from config script 2023-06-04 14:55:51 -04:00
Lincoln Stein
1a7fb601dc ask user for v2 variant when model manager can't infer it 2023-06-04 11:27:44 -04:00
Lincoln Stein
31e97ead2a move invokeai.db to ~/invokeai/databases 
- The invokeai.db database file has now been moved into
  `INVOKEAIROOT/databases`. Using plural here for possible
  future with more than one database file.

- Removed a few dangling debug messages that appeared during
  testing.

- Rebuilt frontend to test web.
 2023-06-03 20:25:34 -04:00
Lincoln Stein
72d1e4e404 fix bug in model_manager that prevented import of inpainting models 2023-06-02 22:39:26 -04:00
Lincoln Stein
1390b65a9c new TUI is fully functional; needs some polishing 2023-06-02 17:20:50 -04:00
Lincoln Stein
41f7758977 listing, downloading and deleting LoRAs working; TI support pending 2023-06-02 00:40:15 -04:00
Lincoln Stein
e9821ab711 implemented tabbed model selection; not wired to backend yet 2023-06-01 00:31:46 -04:00
Sergey Borisov
b47786e846 First working TI draft 2023-05-31 02:12:27 +03:00
Sergey Borisov
69ccd3a0b5 Fixes for checkpoint models 2023-05-30 19:12:47 +03:00
Sergey Borisov
79de9047b5 First working lora implementation 2023-05-30 01:11:00 +03:00
Lincoln Stein
2273b3a8c8 fix potential race condition in config system 2023-05-25 20:41:26 -04:00
Sergey Borisov
8e419a4f97 Revert weak references as can be done without it 2023-05-23 04:29:40 +03:00
Sergey Borisov
2533209326 Rewrite cache to weak references 2023-05-23 03:48:22 +03:00
Lincoln Stein
d2dc1ed26f make InvokeAI package installable 
This commit makes InvokeAI 3.0 to be installable via PyPi.org and the
installer script.

Main changes.

1. Move static web pages into `invokeai/frontend/web` and modify the
API to look for them there. This allows pip to copy the files into the
distribution directory so that user no longer has to be in repo root
to launch.

2. Update invoke.sh and invoke.bat to launch the new web application
properly. This also changes the wording for launching the CLI from
"generate images" to "explore the InvokeAI node system," since I would
not recommend using the CLI to generate images routinely.

3. Fix a bug in the checkpoint converter script that was identified
during testing.

4. Better error reporting when checkpoint converter fails.

5. Rebuild front end.
 2023-05-22 17:51:47 -04:00
Lincoln Stein
27241cdde1 port more globals changes over 2023-05-18 17:17:45 -04:00
Lincoln Stein
259d6ec90d fixup cachedir call 2023-05-18 14:52:16 -04:00
Lincoln Stein
a77c4c87b2 fixed logic error in resolution of model path 2023-05-18 14:35:34 -04:00
Lincoln Stein
d96175d127 resolve some undefined symbols in model_cache 2023-05-18 14:31:47 -04:00
Lincoln Stein
b1a99d772c added method to convert vaes 2023-05-18 13:31:11 -04:00
Lincoln Stein
7ea995149e fixes to env parsing, textual inversion & help text 
- Make environment variable settings case InSenSiTive:
  INVOKEAI_MAX_LOADED_MODELS and InvokeAI_Max_Loaded_Models
  environment variables will both set `max_loaded_models`

- Updated realesrgan to use new config system.

- Updated textual_inversion_training to use new config system.

- Discovered a race condition when InvokeAIAppConfig is created
  at module load time, which makes it impossible to customize
  or replace the help message produced with --help on the command
  line. To fix this, moved all instances of get_invokeai_config()
  from module load time to object initialization time. Makes code
  cleaner, too.

- Added `--from_file` argument to `invokeai-node-cli` and changed
  github action to match. CI tests will hopefully work now.
 2023-05-18 10:48:23 -04:00
Sergey Borisov
fd82763412 Model manager draft 2023-05-18 03:56:52 +03:00
Lincoln Stein
e971a7f35c when migrating models.yaml, rename original models.yaml.orig 2023-05-16 22:37:53 -04:00
Lincoln Stein
cd16857f38 fix None in model_type 2023-05-16 00:13:44 -04:00
Lincoln Stein
1442f1cb8d change model filter to None in second place 2023-05-16 00:03:57 -04:00
Lincoln Stein
4fe94a9315 list_models() now returns a dict of {type,{name: info}} 2023-05-15 23:44:08 -04:00
Lincoln Stein
c8f765cc06 improve debugging messages 2023-05-14 18:29:55 -04:00
Lincoln Stein
b9e9087dbe do not manage GPU for pipelines if sequential_offloading is True 2023-05-14 18:09:38 -04:00
Lincoln Stein
63e465eb5c tweaks to get_model() behavior 
1. If an external VAE is specified in config file, then
   get_model(submodel=vae) will return the external VAE, not the one
   burnt into the parent diffusers pipeline.

2. The mechanism in (1) is generalized such that you can now have
   "unet:", "text_encoder:" and similar stanzas in the config file.
   Valid formats of these subsections:

       unet:
          repo_id: foo/bar

       unet:
          path: /path/to/local/folder

       unet:
          repo_id: foo/bar
	  subfolder: unet

    In the near future, these will also be used to attach external
    parts to the pipeline, generalizing VAE behavior.

3. Accommodate callers (i.e. the WebUI) that are passing the
   model key ("diffusers/stable-diffusion-1.5") to get_model()
   instead of the tuple of model_name and model_type.

4. Fixed bug in VAE model attaching code.

5. Rebuilt web front end.
 2023-05-14 16:50:59 -04:00
Lincoln Stein
baf5451fa0
Merge branch 'main' into lstein/new-model-manager 2023-05-13 22:01:34 -04:00
Lincoln Stein
1103ab2844 merge with main 2023-05-13 21:35:19 -04:00
Lincoln Stein
b31a6ff605 fix reversed args in _model_key() call 2023-05-13 21:11:06 -04:00
Sergey Borisov
1f602e6143 Fix - apply precision to text_encoder 2023-05-14 03:46:13 +03:00
Sergey Borisov
039fa73269 Change SDModelType enum to string, fixes(model unload negative locks count, scheduler load error, saftensors convert, wrong logic in del_model, wrong parse metadata in web) 2023-05-14 03:06:26 +03:00
Lincoln Stein
2204e47596 allow submodels to be fetched independent of parent pipeline 2023-05-13 16:54:47 -04:00
Lincoln Stein
d8b1f29066 proxy SDModelInfo so that it can be used directly as context 2023-05-13 16:29:18 -04:00
Lincoln Stein
b23c9f1da5 get Tuple type hint syntax right 2023-05-13 14:59:21 -04:00
Lincoln Stein
72967bf118 convert add_model(), del_model(), list_models() etc to use bifurcated names 2023-05-13 14:44:44 -04:00
Sergey Borisov
3b2a054f7a Add model loader node; unet, clip, vae fields; change compel node to clip field 2023-05-13 04:37:20 +03:00
Sergey Borisov
131145eab1 A big refactor of model manager(according to IMHO) 2023-05-12 23:13:34 +03:00
Kevin Turner
4caa1f19b2 fix(model manager): fix string formatting error on model checksum timer 2023-05-11 19:06:02 -07:00
Lincoln Stein
df5b968954 model manager now running as a service 2023-05-11 21:24:29 -04:00
blessedcoolant
06b5800d28 Add UniPC Scheduler 2023-05-11 22:43:18 +12:00
Lincoln Stein
4627910c5d added a wrapper model_manager_service and model events 2023-05-11 00:09:19 -04:00
Lincoln Stein
99c692f397 check that model name matches format 2023-05-09 23:46:59 -04:00
Lincoln Stein
3d85e769ce clean up ckpt handling 
- remove legacy ckpt loading code from model_cache
- added placeholders for lora and textual inversion model loading
 2023-05-09 22:44:58 -04:00
Lincoln Stein
9cb962cad7 ckpt model conversion now done in ModelCache 2023-05-08 23:39:44 -04:00
Lincoln Stein
a108155544 added StALKeR779's great model size calculating routine 2023-05-08 21:47:03 -04:00
Lincoln Stein
c15b49c805 implement StALKeR7779 requested API for fetching submodels 2023-05-07 23:18:17 -04:00
Lincoln Stein
fd63e36822 optimize subfolder so that it returns submodel if parent is in RAM 2023-05-07 21:39:11 -04:00
Lincoln Stein
4649920074 adjust t2i to work with new model structure 2023-05-07 19:06:49 -04:00
Lincoln Stein
667171ed90 cap model cache size using bytes, not # models 2023-05-07 18:07:28 -04:00
Lincoln Stein
647ffb2a0f defined abstract baseclass for model manager service 2023-05-06 22:41:19 -04:00
Lincoln Stein
05a27bda5e generalize model loading support, include loras/embeds 2023-05-06 15:58:44 -04:00
Lincoln Stein
e0214a32bc mostly ported to new manager API; needs testing 2023-05-06 00:44:12 -04:00
Lincoln Stein
af8c7c7d29 model manager rewritten to use model_cache; API changed! 2023-05-05 19:32:28 -04:00
Lincoln Stein
a4e36bc02a when model is forcibly moved into RAM update loaded_models set 2023-05-04 23:28:03 -04:00
Lincoln Stein
68bc0112fa implement lazy GPU offloading and ref counting 2023-05-04 23:15:32 -04:00
Lincoln Stein
e4196bbe5b adjust non-app modules to use new config system 2023-05-04 00:43:51 -04:00
Lincoln Stein
15ffb53e59 remove globals, args, generate and the legacy CLI 2023-05-03 23:36:51 -04:00
Lincoln Stein
a273bdbdc1
Merge branch 'main' into lstein/new-model-manager 2023-05-03 18:09:29 -04:00
Lincoln Stein
e1fed52c66 work on model cache and its regression test finished 2023-05-03 12:38:18 -04:00
Lincoln Stein
bb959448c1 implement hashing for local & remote models 2023-05-02 16:52:27 -04:00
Lincoln Stein
2e2abf6ea6 caching of subparts working 2023-05-01 22:57:30 -04:00
Lincoln Stein
974841926d logger is a interchangeable service 2023-04-29 10:48:50 -04:00
Lincoln Stein
8db20e0d95 rename log to logger throughout 2023-04-29 09:43:40 -04:00
Lincoln Stein
6b79e2b407 Merge branch 'main' into enhance/invokeai-logs 
- resolve conflicts
- remove unused code identified by pyflakes
 2023-04-28 10:09:46 -04:00
Lincoln Stein
956ad6bcf5 add redesigned model cache for diffusers & transformers 2023-04-28 00:41:52 -04:00
Lincoln Stein
aab262d991
Merge branch 'main' into bugfix/prevent-cli-crash 2023-04-14 20:12:38 -04:00
Lincoln Stein
0b0e6fe448 convert remainder of print() to log.info() 2023-04-14 15:15:14 -04:00
Lincoln Stein
c132dbdefa change "ialog" to "log" 2023-04-11 18:48:20 -04:00
Lincoln Stein
5a4765046e add logging support 
This commit adds invokeai.backend.util.logging, which provides support
for formatted console and logfile messages that follow the status
reporting conventions of earlier InvokeAI versions.

Examples:

   ### A critical error     (logging.CRITICAL)
   *** A non-fatal error    (logging.ERROR)
   ** A warning             (logging.WARNING)
   >> Informational message (logging.INFO)
      | Debugging message   (logging.DEBUG)
 2023-04-11 09:33:28 -04:00
AbdBarho
de189f2db6
Increase chunk size when computing SHAs 2023-04-09 21:53:59 +02:00
Lincoln Stein
8334757af9
Merge branch 'main' into bugfix/prevent-cli-crash 2023-04-07 18:55:54 -04:00
Lincoln Stein
4c339dd4b0 refactor get_submodels() into individual methods 2023-04-06 17:08:23 -04:00
Lincoln Stein
d44151d6ff add a new method to model_manager that retrieves individual pipeline parts 
- New method is ModelManager.get_sub_model(model_name:str,model_part:SDModelComponent)

To use:

```
from invokeai.backend import ModelManager, SDModelComponent as sdmc
manager = ModelManager('/path/to/models.yaml')
vae = manager.get_sub_model('stable-diffusion-1.5', sdmc.vae)
```
 2023-04-05 17:25:42 -04:00
Lincoln Stein
3c4b6d5735
Merge branch 'main' into enhance/heuristic-import-improvements 2023-03-29 16:54:43 -04:00
Lincoln Stein
9a7580dedd fix bugs in online ckpt conversion of 2.0 models 
This commit fixes bugs related to the on-the-fly conversion and loading of
legacy checkpoint models built on SD-2.0 base.

- When legacy checkpoints built on SD-2.0 models were converted
  on-the-fly using --ckpt_convert, generation would crash with a
  precision incompatibility error.
 2023-03-28 00:17:20 -04:00
Lincoln Stein
fe5d9ad171 improve importation and conversion of legacy checkpoint files 
A long-standing issue with importing legacy checkpoints (both ckpt and
safetensors) is that the user has to identify the correct config file,
either by providing its path or by selecting which type of model the
checkpoint is (e.g. "v1 inpainting"). In addition, some users wish to
provide custom VAEs for use with the model. Currently this is done in
the WebUI by importing the model, editing it, and then typing in the
path to the VAE.

To improve the user experience, the model manager's
`heuristic_import()` method has been enhanced as follows:

1. When initially called, the caller can pass a config file path, in
which case it will be used.

2. If no config file provided, the method looks for a .yaml file in the
same directory as the model which bears the same basename. e.g.
```
   my-new-model.safetensors
   my-new-model.yaml
```
   The yaml file is then used as the configuration file for
   importation and conversion.

3. If no such file is found, then the method opens up the checkpoint
   and probes it to determine whether it is V1, V1-inpaint or V2.
   If it is a V1 format, then the appropriate v1-inference.yaml config
   file is used. Unfortunately there are two V2 variants that cannot be
   distinguished by introspection.

4. If the probe algorithm is unable to determine the model type, then its
   last-ditch effort is to execute an optional callback function that can
   be provided by the caller. This callback, named `config_file_callback`
   receives the path to the legacy checkpoint and returns the path to the
   config file to use. The CLI uses to put up a multiple choice prompt to
   the user. The WebUI **could** use this to prompt the user to choose
   from a radio-button selection.

5. If the config file cannot be determined, then the import is abandoned.

The user can attach a custom VAE to the imported and converted model
by copying the desired VAE into the same directory as the file to be
imported, and giving it the same basename. E.g.:

```
    my-new-model.safetensors
    my-new-model.vae.pt
```

For this to work, the VAE must end with ".vae.pt", ".vae.ckpt", or
".vae.safetensors". The indicated VAE will be converted into diffusers
format and stored with the converted models file, so the ".pt" file
can be deleted after conversion.

No facility is currently provided to swap a diffusers VAE at import
time, but this can be done after the fact using the WebUI and CLI's
model editing functions.
 2023-03-27 11:27:45 -04:00
Lincoln Stein
5ac0316c62 fix issue with embeddings being loaded twice 
- as noted by JPPhoto
 2023-03-25 10:45:03 -04:00
Lincoln Stein
9ceec40b76
Merge branch 'main' into feat/use-custom-vaes 2023-03-24 17:45:02 -04:00
Lincoln Stein
85b2822f5e
Merge branch 'main' into security/scan-ckpt-files-main 2023-03-24 08:39:59 -04:00
Lincoln Stein
6e7dbf99f3
Merge branch 'main' into bugfix/dreambooth_ema 2023-03-23 23:24:15 -04:00
Lincoln Stein
deeff36e16
Merge branch 'main' into security/scan-ckpt-files-main 2023-03-23 23:20:52 -04:00
Lincoln Stein
f751dcd245 load embeddings after a ckpt legacy model is converted to diffusers 
- Fixes #2954
- Also improves diagnostic reporting during embedding loading.
 2023-03-23 15:21:58 -04:00
Lincoln Stein
a97107bd90 handle VAEs that do not have a "state_dict" key 2023-03-23 15:11:29 -04:00
Lincoln Stein
b2ce45a417 re-implement model scanning when loading legacy checkpoint files 
- This PR turns on pickle scanning before a legacy checkpoint file
  is loaded from disk within the checkpoint_to_diffusers module.

- Also miscellaneous diagnostic message cleanup.
 2023-03-23 15:03:30 -04:00
Lincoln Stein
4e0b5d85ba convert custom VAEs into diffusers 
- When a legacy checkpoint model is loaded via --convert_ckpt and its
  models.yaml stanza refers to a custom VAE path (using the 'vae:'
  key), the custom VAE will be converted and used within the diffusers
  model. Otherwise the VAE contained within the legacy model will be
  used.

- Note that the heuristic_import() method, which imports arbitrary
  legacy files on disk and URLs, will continue to default to the
  the standard stabilityai/sd-vae-ft-mse VAE. This can be fixed after
  the fact by editing the models.yaml stanza using the Web or CLI
  UIs.

- Fixes issue #2917
 2023-03-23 13:14:19 -04:00
Lincoln Stein
a958ae5e29 Merge branch 'main' into feat/use-custom-vaes 2023-03-23 10:32:56 -04:00
Lincoln Stein
3ca654d256 speculative fix for alternative vaes 2023-03-13 23:27:29 -04:00
jeremy
e0e01f6c50 Reduced Pickle ACE attack surface 
Prior to this commit, all models would be loaded with the extremely unsafe `torch.load` method, except those with the exact extension `.safetensors`. Even a change in casing (eg. `saFetensors`, `Safetensors`, etc) would cause the file to be loaded with torch.load instead of the much safer `safetensors.toch.load_file`.
If a malicious actor renamed an infected `.ckpt` to something like `.SafeTensors` or `.SAFETENSORS` an unsuspecting user would think they are loading a safe .safetensor, but would in fact be parsing an unsafe pickle file, and executing an attacker's payload. This commit fixes this vulnerability by reversing the loading-method decision logic to only use the unsafe `torch.load` when the file extension is exactly `.ckpt`.
 2023-03-13 16:16:30 -04:00
Kyle Schouviller
3ee2798ede [fix] Get the model again if current model is empty 2023-03-12 22:26:11 -04:00
Fabio 'MrWHO' Torchetti
5c5106c14a Add keys when non EMA 2023-03-12 16:22:22 -05:00
Fabio 'MrWHO' Torchetti
c367b21c71 Fix issue #2932 2023-03-12 15:40:33 -05:00
Lincoln Stein
6a77634b34 remove unneeded generate initializer routines 2023-03-11 17:14:03 -05:00
Lincoln Stein
c14241436b move ModelManager initialization into its own module and restore embedding support 2023-03-11 10:56:53 -05:00
Lincoln Stein
d612f11c11 initialize InvokeAIGenerator object with model, not manager 2023-03-11 09:06:46 -05:00
Lincoln Stein
fe75b95464 Merge branch 'refactor/nodes-on-generator' of github.com:invoke-ai/InvokeAI into refactor/nodes-on-generator 2023-03-10 19:36:40 -05:00
Lincoln Stein
95954188b2 remove factory pattern 
Factory pattern is now removed. Typical usage of the InvokeAIGenerator is now:

```
from invokeai.backend.generator import (
    InvokeAIGeneratorBasicParams,
    Txt2Img,
    Img2Img,
    Inpaint,
)
    params = InvokeAIGeneratorBasicParams(
        model_name = 'stable-diffusion-1.5',
        steps = 30,
        scheduler = 'k_lms',
        cfg_scale = 8.0,
        height = 640,
        width = 640
        )
    print ('=== TXT2IMG TEST ===')
    txt2img = Txt2Img(manager, params)
    outputs = txt2img.generate(prompt='banana sushi', iterations=2)

    for i in outputs:
        print(f'image={output.image}, seed={output.seed}, model={output.params.model_name}, hash={output.model_hash}, steps={output.params.steps}')
```

The `params` argument is optional, so if you wish to accept default
parameters and selectively override them, just do this:

```
    outputs = Txt2Img(manager).generate(prompt='banana sushi',
                                        steps=50,
					scheduler='k_heun',
					model_name='stable-diffusion-2.1'
					)
```
 2023-03-10 19:33:04 -05:00
Jonathan
370e8281b3
Merge branch 'main' into refactor/nodes-on-generator 2023-03-10 12:34:00 -06:00
Lincoln Stein
685df33584
fix bug that caused black images when converting ckpts to diffusers in RAM (#2914) 
Cause of the problem was inadvertent activation of the safety checker.

When conversion occurs on disk, the safety checker is disabled during loading.
However, when converting in RAM, the safety checker was not removed, resulting
in it activating even when user specified --no-nsfw_checker.

This PR fixes the problem by detecting when the caller has requested the InvokeAi
StableDiffusionGeneratorPipeline class to be returned and setting safety checker
to None. Do not do this with diffusers models destined for disk because then they
will be incompatible with the merge script!!

Closes #2836
 2023-03-10 18:11:32 +00:00
Lincoln Stein
14c8738a71 fix dangling reference to _model_to_cpu and missing variable model_description 2023-03-09 21:41:45 -05:00
Kevin Turner
ad7b1fa6fb model_manager: model to/from CPU methods are implemented on the Pipeline 2023-03-09 18:15:12 -08:00
Lincoln Stein
b679a6ba37 model manager defaults to consistent values of device and precision 2023-03-09 01:09:54 -05:00
Lincoln Stein
7c60068388
Merge branch 'main' into bugfix/fix-convert-sd-to-diffusers-error 2023-03-06 08:20:29 -05:00
Lincoln Stein
94daaa4abf fix call signature of import_diffuser_model() 2023-03-05 23:37:59 -05:00
Lincoln Stein
2f9dcd7906 support both epsilon and v-prediction v2 inference 
There are actually two Stable Diffusion v2 legacy checkpoint
configurations:

1) "epsilon" prediction type for Stable Diffusion v2 Base
2) "v-prediction" type for Stable Diffusion v2-768

This commit adds the configuration file needed for epsilon prediction
type models as well as the UI that prompts the user to select the
appropriate configuration file when the code can't do so
automatically.
 2023-03-05 22:51:40 -05:00
blessedcoolant
e537b5d8e1 Revert "Merge branch 'main' into bugfix/reenable-ckpt-conversion-to-ram" 
This reverts commit e0e70c9222, reversing
changes made to 0b184913b9.
 2023-03-06 14:29:39 +13:00
blessedcoolant
e0e70c9222 Merge branch 'main' into bugfix/reenable-ckpt-conversion-to-ram 2023-03-06 14:27:30 +13:00
Lincoln Stein
fc187f263e deal with non-directories in diffusers/ 2023-03-05 12:29:52 -05:00
Lincoln Stein
4e9e1b660d respect HF_HOME setting when migrating 2023-03-05 12:08:29 -05:00
Lincoln Stein
d01adedff5 give user chance to back out before migration 2023-03-05 12:04:31 -05:00
Lincoln Stein
b33655b0d6 restore automatic conversion of legacy files to diffusers pipelines 2023-03-05 11:45:25 -05:00
Lincoln Stein
81dee04dc9 during migration do not overwrite symlinks 2023-03-05 08:40:12 -05:00
Lincoln Stein
ef8cf83b28 migrate to new HF diffusers cache location 2023-03-05 08:20:24 -05:00
Lincoln Stein
44400d2a66 fix incorrect import of merge code 2023-03-03 01:07:31 -05:00
Lincoln Stein
60a98cacef all vestiges of ldm.invoke removed 2023-03-03 01:02:00 -05:00
Lincoln Stein
6a990565ff all files migrated; tweaks needed 2023-03-03 00:02:15 -05:00