Lincoln Stein
d73f1c363c
bump version number
2023-05-01 09:28:49 -04:00
Lincoln Stein
e52e7418bb
close #3304
2023-04-29 20:07:21 -04:00
Lincoln Stein
73be58a0b5
fix issue #3293
2023-04-29 11:37:07 -04:00
Lincoln Stein
5a7d11bca8
remove debugging statement
2023-04-27 08:21:26 -04:00
Lincoln Stein
5bbf7fe34a
[Bugfix] Renames in 0.15.0 diffusers ( #3184 )
...
Link to PR in diffusers repository:
https://github.com/huggingface/diffusers/pull/2691
Imports:
`diffusers.models.cross_attention ->
diffusers.models.attention_processor`
Unions:
`AttnProcessor -> AttentionProcessor`
Classes:
| Old name | New name |
| --- | --- |
| CrossAttention | Attention |
| CrossAttnProcessor | AttnProcessor |
| XFormersCrossAttnProcessor | XFormersAttnProcessor |
| CrossAttnAddedKVProcessor | AttnAddedKVProcessor |
| LoRACrossAttnProcessor | LoRAAttnProcessor |
| LoRAXFormersCrossAttnProcessor | LoRAXFormersAttnProcessor |
| FlaxCrossAttention | FlaxAttention |
| AttendExciteCrossAttnProcessor | AttendExciteAttnProcessor |
| Pix2PixZeroCrossAttnProcessor | Pix2PixZeroAttnProcessor |
Also config values no longer sets as attributes of object:
https://github.com/huggingface/diffusers/pull/2849
2023-04-27 11:38:27 +01:00
Lincoln Stein
bfb968bbe8
Merge branch 'v2.3' into fix/new_diffusers_names
2023-04-26 23:54:37 +01:00
Lincoln Stein
6db72f83a2
bump version number to 2.3.5-rc1 ( #3267 )
...
Bump version number for 2.3.5 release candidate.
2023-04-26 23:53:53 +01:00
Sergey Borisov
432e526999
Revert merge changes
2023-04-25 14:49:08 +03:00
Lincoln Stein
830740b93b
remove redundant/buggy restore_default_attention() method
2023-04-25 07:05:07 -04:00
StAlKeR7779
ff3f289342
Merge branch 'v2.3' into fix/new_diffusers_names
2023-04-25 13:21:26 +03:00
Lincoln Stein
34abbb3589
Merge branch 'v2.3' into release/v2.3.5
2023-04-25 04:33:09 +01:00
Lincoln Stein
c0eb1a9921
increase sha256 chunksize when calculating model hash ( #3162 )
...
- Thanks to @abdBarho, who discovered that increasing the chunksize
dramatically decreases the amount of time to calculate the hash.
2023-04-25 04:25:55 +01:00
Lincoln Stein
2ddd0301f4
bump version number to 2.3.5-rc1
2023-04-24 23:24:33 -04:00
Lincoln Stein
ce6629b6f5
Merge branch 'v2.3' into enhance/increase-sha256-chunksize
2023-04-25 03:58:30 +01:00
Lincoln Stein
994a76aeaa
[Enhancement] distinguish v1 from v2 LoRA models ( #3175 )
...
# Distinguish LoRA/LyCORIS files based on what version of SD they were
built on top of
- Attempting to run a prompt with a LoRA based on SD v1.X against a
model based on v2.X will now throw an `IncompatibleModelException`. To
import this exception:
`from ldm.modules.lora_manager import IncompatibleModelException` (maybe
this should be defined in ModelManager?)
- Enhance `LoraManager.list_loras()` to accept an optional integer
argument, `token_vector_length`. This will filter the returned LoRA
models to return only those that match the indicated length. Use:
```
768 => for models based on SD v1.X
1024 => for models based on SD v2.X
```
Note that this filtering requires each LoRA file to be opened by
`torch.safetensors`. It will take ~8s to scan a directory of 40 files.
- Added new static methods to `ldm.modules.kohya_lora_manager`:
- check_model_compatibility()
- vector_length_from_checkpoint()
- vector_length_from_checkpoint_file()
- You can now create subdirectories within the `loras` directory and
organize the model files.
2023-04-25 03:57:45 +01:00
Lincoln Stein
144dfe4a5b
Merge branch 'v2.3' into bugfix/lora-incompatibility-handling
2023-04-25 03:54:46 +01:00
Lincoln Stein
5dbc63e2ae
Revert "improvements to the installation and upgrade processes" ( #3266 )
...
Reverts invoke-ai/InvokeAI#3186
2023-04-25 03:54:04 +01:00
Lincoln Stein
c6ae1edc82
Revert "improvements to the installation and upgrade processes"
2023-04-24 22:53:43 -04:00
Lincoln Stein
0f3c456d59
merge with v2.3
2023-04-24 22:51:48 -04:00
Lincoln Stein
2cd0e036ac
Merge branch 'v2.3' into bugfix/lora-incompatibility-handling
2023-04-25 03:24:25 +01:00
Lincoln Stein
a45b3387c0
Merge branch 'v2.3' into enhance/increase-sha256-chunksize
2023-04-25 03:22:43 +01:00
Lincoln Stein
c088cf0344
improvements to the installation and upgrade processes ( #3186 )
...
- Moved all postinstallation config file and model munging code out of
the CLI and into a separate script named `invokeai-postinstall`
- Fixed two calls to `shutil.copytree()` so that they don't try to
preserve the file mode of the copied files. This is necessary to run
correctly in a Nix environment (see thread at
https://discord.com/channels/1020123559063990373/1091716696965918732/1095662756738371615 )
- Update the installer so that an existing virtual environment will be
updated, not overwritten.
- Pin npyscreen version to see if this fixes issues people have had with
installing this module.
2023-04-25 03:20:58 +01:00
Lincoln Stein
264af3c054
fix crash caused by incorrect conflict resolution
2023-04-24 22:20:12 -04:00
Lincoln Stein
b332432a88
Merge branch 'v2.3' into lstein/bugfix/improve-update-handling
2023-04-25 03:09:12 +01:00
Lincoln Stein
7f7d5894fa
Merge branch 'v2.3' into bugfix/lora-incompatibility-handling
2023-04-25 02:51:27 +01:00
Lincoln Stein
96c39b61cf
Enable LoRAs to patch the text_encoder as well as the unet ( #3214 )
...
Load LoRAs during compel's text embedding encode pass in case there are
requested LoRAs which also want to patch the text encoder.
Also generally cleanup the attention processor patching stuff. It's
still a mess, but at least now it's a *stateless* mess.
2023-04-24 23:22:51 +01:00
Lincoln Stein
40744ed996
Merge branch 'v2.3' into fix_inconsistent_loras
2023-04-22 20:22:32 +01:00
Lincoln Stein
2a2c86896a
pull in diffusers 0.15.1
...
- Change diffusers dependency to `diffusers~=0.15.0` which *should*
enforce non-breaking changes.
2023-04-20 13:29:20 -04:00
Lincoln Stein
f36452d650
rebuild front end
2023-04-20 12:27:08 -04:00
Lincoln Stein
e5188309ec
Merge branch 'v2.3' into bugfix/lora-incompatibility-handling
2023-04-20 17:25:09 +01:00
Lincoln Stein
aabe79686e
Merge branch 'v2.3' into fix/new_diffusers_names
2023-04-20 17:20:33 +01:00
Lincoln Stein
a9e8005a92
CODEOWNERS update - 2.3 branch ( #3230 )
...
Both @mauwii and @keturn have been offline for some time. I am
temporarily removing them from CODEOWNERS so that they will not be
responsible for code reviews until they wish to/are able to re-engage
fully.
Note that I have volunteered @GreggHelt2 to be a codeowner of the
generation backend code, replacing @keturn . Let me know if you're
uncomfortable with this.
2023-04-20 17:19:51 +01:00
Lincoln Stein
c2e6d98e66
Merge branch 'v2.3' into dev/codeowner-fix-2.3
2023-04-20 17:19:30 +01:00
Lincoln Stein
40d9b5dc27
[Feature] Add support for LoKR LyCORIS format ( #3216 )
...
It's like LoHA but use Kronecker product instead of Hadamard product.
https://github.com/KohakuBlueleaf/LyCORIS#lokr
I tested it on this 2 LoKR's:
https://civitai.com/models/34518/unofficial-vspo-yakumo-beni
https://civitai.com/models/35136/mika-pikazo-lokr
More tests hard to find as it's new format)
Better to test with https://github.com/invoke-ai/InvokeAI/pull/3214
Also a bit refactor forward function.
//LyCORIS also have (IA)^3 format, but I can't find examples in this
format and even on LyCORIS page it's marked as experimental. So, until
there some test examples I prefer not to add this.
2023-04-19 22:51:33 +01:00
Lincoln Stein
216b1c3a4a
Merge branch 'v2.3' into fix/new_diffusers_names
2023-04-18 19:37:25 -04:00
Lincoln Stein
1a704efff1
update codeowners in response to team changes
2023-04-18 19:30:52 -04:00
Lincoln Stein
f49d2619be
Merge branch 'v2.3' into fix_inconsistent_loras
2023-04-18 19:09:35 -04:00
Lincoln Stein
da96ec9dd5
Merge branch 'v2.3' into feat/lokr_support
2023-04-18 19:08:03 -04:00
Lincoln Stein
298ccda365
fix the "import from directory" function in console model installer ( #3211 )
...
- This was inadvertently broken when we stopped supporting direct
loading of checkpoint models.
- Now fixed.
- May fix #3209
2023-04-17 23:04:27 -04:00
StAlKeR7779
967d853020
Merge branch 'v2.3' into feat/lokr_support
2023-04-16 23:10:45 +03:00
StAlKeR7779
e91117bc74
Add support for lokr lycoris format
2023-04-16 23:05:13 +03:00
Damian Stewart
4d58444153
fix issues and further cleanup
2023-04-16 17:54:21 +02:00
Damian Stewart
3667eb4d0d
activate LoRAs when generating prompt embeddings; also cleanup attention stuff
2023-04-16 17:03:31 +02:00
Lincoln Stein
203a7157e1
fix the "import from directory" function in console model installer
...
- This was inadvertently broken when we stopped supporting direct
loading of checkpoint models.
- Now fixed.
2023-04-15 21:07:02 -04:00
Lincoln Stein
47883860a6
Merge branch 'v2.3' into enhance/increase-sha256-chunksize
2023-04-13 23:00:34 -04:00
Lincoln Stein
6365a7c790
Merge branch 'v2.3' into lstein/bugfix/improve-update-handling
2023-04-13 22:49:41 -04:00
Lincoln Stein
5fcb3d90e4
fix missing files variable
2023-04-13 22:49:04 -04:00
Lincoln Stein
8f17d17208
Merge branch 'v2.3' into fix/new_diffusers_names
2023-04-13 22:44:05 -04:00
Lincoln Stein
c6ecf3afc5
pin diffusers to 0.15.*, and fix deprecation warning on unet.in_channels
2023-04-13 22:38:50 -04:00
Lincoln Stein
2c449bfb34
Merge branch 'v2.3' into bugfix/lora-incompatibility-handling
2023-04-13 22:23:59 -04:00
