AI/diffusers
1
0
Fork 0
mirror of https://github.com/huggingface/diffusers.git synced 2026-01-29 07:22:12 +03:00
Code Activity
1,440 Commits 742 Branches 94 Tags
459b8ca81a32799aebca681c0ab97e29ddca318f
Commit Graph

1440 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Patrick von Platen
459b8ca81a Research folder (#1553) 
* Research folder

* Update examples/research_projects/README.md

* up
 2022-12-05 17:58:35 +01:00
Suraj Patil
bce65cd13a [refactor] make set_attention_slice recursive (#1532) 
* make attn slice recursive

* remove set_attention_slice from blocks

* fix copies

* make enable_attention_slicing base class method of DiffusionPipeline

* fix set_attention_slice

* fix set_attention_slice

* fix copies

* add tests

* up

* up

* up

* update

* up

* uP

Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
 2022-12-05 17:31:04 +01:00
Adalberto
e289998932 fix mask discrepancies in train_dreambooth_inpaint (#1529) 
The mask and instance image were being cropped in different ways without --center_crop, causing the model to learn to ignore the mask in some cases. This PR fixes that and generate more consistent results.
 2022-12-05 17:26:36 +01:00
Suraj Patil
634be6e53d [examples] use from_pretrained to load scheduler (#1549) 
us from_pretrained to load scheduler
 2022-12-05 15:32:24 +01:00
allo-
d1bcbf38ca [textual_inversion] Add an option for only saving the embeddings (#781) 
[textual_inversion] Add an option to only save embeddings

Add an command line option --only_save_embeds to the example script, for
not saving the full model. Then only the learned embeddings are saved,
which can be added to the original model at runtime in a similar way as
they are created in the training script.
Saving the full model is forced when --push_to_hub is used. (Implements #759)
 2022-12-05 14:45:13 +01:00
Patrick von Platen
df7cd5fe3f Update bug-report.yml 2022-12-05 14:39:35 +01:00
Naga Sai Abhinay
c28d6945b8 [Community Pipeline] Checkpoint Merger based on Automatic1111 (#1472) 
* Add checkpoint_merger pipeline

* Added missing docs for a parameter.

* Fomratting fixes.

* Fixed code quality issues.

* Bug fix: Off by 1 index

* Added docs for pipeline
 2022-12-05 14:36:55 +01:00
Patrick von Platen
5177e65ff0 Update bug-report.yml 2022-12-05 14:17:04 +01:00
Patrick von Platen
60ac5fc235 Update bug-report.yml 2022-12-05 14:13:02 +01:00
Patrick von Platen
19b01749f0 Update bug-report.yml 2022-12-05 14:10:25 +01:00
Patrick von Platen
a980ef2f08 Update bug-report.yml (#1548) 
* Update bug-report.yml

* Update bug-report.yml

* Update bug-report.yml
 2022-12-05 14:03:54 +01:00
Patrick von Platen
7932971542 [Upscaling] Fix batch size (#1525) 2022-12-05 13:28:55 +01:00
Benjamin Lefaudeux
720dbfc985 Compute embedding distances with torch.cdist (#1459) 
small but mighty
 2022-12-05 12:37:05 +01:00
Patrick von Platen
513fc68104 [Stable Diffusion Inpaint] Allow tensor as input image & mask (#1527) 
up
 2022-12-05 12:18:02 +01:00
Anton Lozhkov
cc22bda5f6 [CI] Add slow MPS tests (#1104) 
* [CI] Add slow MPS tests

* fix yml

* temporarily resolve caching

* Tests: fix mps crashes.

* Skip test_load_pipeline_from_git on mps.

Not compatible with float16.

* Increase tolerance, use CPU generator, alt. slices.

* Move to nightly

* style

Co-authored-by: Pedro Cuenca <pedro@huggingface.co>
 2022-12-05 11:50:24 +01:00
Ilmari Heikkinen
daebee0963 Add xformers attention to VAE (#1507) 
* Add xformers attention to VAE

* Simplify VAE xformers code

* Update src/diffusers/models/attention.py

Co-authored-by: Ilmari Heikkinen <ilmari@fhtr.org>
Co-authored-by: Suraj Patil <surajp815@gmail.com>
 2022-12-03 15:08:11 +01:00
Matthieu Bizien
ae368e42d2 [Proposal] Support saving to safetensors (#1494) 
* Add parameter safe_serialization to DiffusionPipeline.save_pretrained

* Add option safe_serialization on ModelMixin.save_pretrained

* Add test test_save_safe_serialization

* Black

* Re-trigger the CI

* Fix doc-builder

* Validate files are saved as safetensor in test_save_safe_serialization
 2022-12-02 18:33:16 +01:00
Patrick von Platen
cf4664e885 fix tests 2022-12-02 17:27:58 +00:00
Patrick von Platen
7222a8eadf make style 2022-12-02 17:18:50 +00:00
bachr
155d272cc1 Update FlaxLMSDiscreteScheduler (#1474) 
- Add the missing `scale_model_input` method to `FlaxLMSDiscreteScheduler`
- Use `jnp.append` for appending to `state.derivatives`
- Use `jnp.delete` to pop from `state.derivatives`
 2022-12-02 18:18:30 +01:00
Adalberto
2b30b1090f Create train_dreambooth_inpaint.py (#1091) 
* Create train_dreambooth_inpaint.py

train_dreambooth.py adapted to work with the inpaint model, generating random masks during the training

* Update train_dreambooth_inpaint.py

refactored train_dreambooth_inpaint with black

* Update train_dreambooth_inpaint.py

* Update train_dreambooth_inpaint.py

* Update train_dreambooth_inpaint.py

Fix prior preservation

* add instructions to readme, fix SD2 compatibility
 2022-12-02 18:06:57 +01:00
Antoine Bouthors
3ad49eeedd Fixed mask+masked_image in sd inpaint pipeline (#1516) 
* Fixed mask+masked_image in sd inpaint pipeline

Those were left unset when inputs are not PIL images

* Fixed formatting
 2022-12-02 17:51:51 +01:00
Patrick von Platen
769f0be8fb Finalize 2nd order schedulers (#1503) 
* up

* up

* finish

* finish

* up

* up

* finish
 2022-12-02 16:38:35 +01:00
Pedro Gabriel Gengo Lourenço
4f596599f4 Fix training docs to install datasets (#1476) 
Fixed doc to install from training packages
 2022-12-02 15:52:04 +01:00
Dhruv Naik
f57a2e0745 Fix Imagic example (#1520) 
fix typo, remove incorrect arguments from .train()
 2022-12-02 15:06:04 +01:00
Pedro Cuenca
3ceaa280bd Do not use torch.long in mps (#1488) 
* Do not use torch.long in mps

Addresses #1056.

* Use torch.int instead of float.

* Propagate changes.

* Do not silently change float -> int.

* Propagate changes.

* Apply suggestions from code review

Co-authored-by: Anton Lozhkov <anton@huggingface.co>

Co-authored-by: Anton Lozhkov <anton@huggingface.co>
 2022-12-02 13:10:17 +01:00
Benjamin Lefaudeux
a816a87a09 [refactor] Making the xformers mem-efficient attention activation recursive (#1493) 
* Moving the mem efficiient attention activation to the top + recursive

* black, too bad there's no pre-commit ?

Co-authored-by: Benjamin Lefaudeux <benjamin@photoroom.com>
 2022-12-02 12:30:01 +01:00
Patrick von Platen
f21415d1d9 Update conversion script to correctly handle SD 2 (#1511) 
* Conversion SD 2

* finish
 2022-12-02 12:28:01 +01:00
Patrick von Platen
22b9cb086b [From pretrained] Allow returning local path (#1450) 
Allow returning local path
 2022-12-02 12:26:39 +01:00
Will Berman
25f850a23b [docs] [dreambooth training] num_class_images clarification (#1508) 2022-12-02 12:12:28 +01:00
Will Berman
b25ae2e6ab [docs] [dreambooth training] accelerate.utils.write_basic_config (#1513) 2022-12-02 12:11:18 +01:00
Suraj Patil
0f1c24664c fix heun scheduler (#1512) 2022-12-01 22:39:57 +01:00
Anton Lozhkov
e65b71aba4 Add an explicit --image_size to the conversion script (#1509) 
* Add an explicit `--image_size` to the conversion script

* style
 2022-12-01 19:22:48 +01:00
Akash Gokul
a6a25ceb61 Fix Flax flip_sin_to_cos (#1369) 
* Fix Flax flip_sin_to_cos

* Adding flip_sin_to_cos

Co-authored-by: Kashif Rasul <kashif.rasul@gmail.com>
 2022-12-01 18:57:01 +01:00
Suraj Patil
b85bb0753e support v prediction in other schedulers (#1505) 
* support v prediction in other schedulers

* v heun

* add tests for v pred

* fix tests

* fix test euler a

* v ddpm
 2022-12-01 18:10:39 +01:00
fboulnois
52eb0348e5 Standardize on using image argument in all pipelines (#1361) 
* feat: switch core pipelines to use image arg

* test: update tests for core pipelines

* feat: switch examples to use image arg

* docs: update docs to use image arg

* style: format code using black and doc-builder

* fix: deprecate use of init_image in all pipelines
 2022-12-01 16:55:22 +01:00
Suraj Patil
2bbf8b67a7 simplyfy AttentionBlock (#1492) 2022-12-01 16:40:59 +01:00
Patrick von Platen
5a5bf7ef5a [Deprecate] Correct stacklevel (#1483) 
* Correct stacklevel

* fix
 2022-12-01 16:28:10 +01:00
Anton Lozhkov
9276b1e148 Replace deprecated hub utils in train_unconditional_ort (#1504) 
* Replace deprecated hub utils in `train_unconditional_ort`

* typo
 2022-12-01 16:00:52 +01:00
regisss
2579d42158 Add doc for Stable Diffusion on Habana Gaudi (#1496) 
* Add doc for Stable Diffusion on Habana Gaudi

* Make style

* Add benchmark

* Center-align columns in the benchmark table
 2022-12-01 15:43:48 +01:00
Anton Lozhkov
999044596a Bump to 0.10.0.dev0 + deprecations (#1490) 2022-11-30 15:27:56 +01:00
Pedro Cuenca
eeeb28a9ad Remove reminder comment (#1489) 
Remove reminder comment.
 2022-11-30 14:59:54 +01:00
Patrick von Platen
c05356497a Add better docs xformers (#1487) 
* Add better docs xformers

* update

* Apply suggestions from code review

* fix
 2022-11-30 13:57:45 +01:00
Patrick von Platen
1d4ad34af0 [Dreambooth] Make compatible with alt diffusion (#1470) 
* [Dreambooth] Make compatible with alt diffusion

* make style

* add example
 2022-11-30 13:48:17 +01:00
Patrick von Platen
20ce68f945 Fix dtype model loading (#1449) 
* Add test

* up

* no bfloat16 for mps

* fix

* rename test
 2022-11-30 11:31:50 +01:00
Patrick von Platen
110ffe2589 Allow saving trained betas (#1468) 2022-11-30 10:05:51 +01:00
Anton Lozhkov
0b7225e918 Add ort_nightly_directml to the onnxruntime candidates (#1458) 
* Add `ort_nightly_directml` to the `onnxruntime` candidates

* style
 2022-11-29 14:00:41 +01:00
Anton Lozhkov
db7b7bd983 [Train unconditional] Unwrap model before EMA (#1469) 2022-11-29 13:45:42 +01:00
Rohan Taori
6a0a312370 Fix bug in half precision for DPMSolverMultistepScheduler (#1349) 
* cast to float for quantile method

* add fp16 test for DPMSolverMultistepScheduler fix

* formatting update
 2022-11-29 13:29:23 +01:00
Ilmari Heikkinen
c28d3c82ce StableDiffusion: Decode latents separately to run larger batches (#1150) 
* StableDiffusion: Decode latents separately to run larger batches

* Move VAE sliced decode under enable_vae_sliced_decode and vae.enable_sliced_decode

* Rename sliced_decode to slicing

* fix whitespace

* fix quality check and repository consistency

* VAE slicing tests and documentation

* API doc hooks for VAE slicing

* reformat vae slicing tests

* Skip VAE slicing for one-image batches

* Documentation tweaks for VAE slicing

Co-authored-by: Ilmari Heikkinen <ilmari@fhtr.org>
 2022-11-29 13:28:14 +01:00