ort integration#916

Closed

prathikr wants to merge 227 commits intohuggingface:mainfrom

prathikr:prathikrao/ort-integration

Commits on Sep 22, 2022

wrapped model in ORTModule
Prathik Rao
committed
bug fix
Prathik Rao
committed

Commits on Oct 19, 2022

bug fixes
Prathik Rao
committed
formatting
Prathik Rao
committed

Commits on Oct 20, 2022

bug fix
Prathik Rao
committed
remove commented blocks
Prathik Rao
committed
bug fix
Prathik Rao
committed
bug fix
Prathik Rao
committed
formatting
Prathik Rao
committed
add onnxruntime.training
Prathik Rao
committed
Merge branch 'main' into prathikrao/ort-integration
prathikr
authored

Commits on Oct 26, 2022

docs: `.md` readability fixups (#619)

ryanrussell
authored and
Prathik Rao
committed
Flax documentation (#589)

authored and
Prathik Rao
committed
fix docs: change sample to images (#613)

AbdullahAlfaraj
authored and
Prathik Rao
committed
refactor: pipelines readability improvements (#622)

ryanrussell
authored and
Prathik Rao
committed
Allow passing session_options for ORT backend (#620)

cloudhan
authored and
Prathik Rao
committed
Fix breaking error: "ort is not defined" (#626)

pcuenca
authored and
Prathik Rao
committed
docs: `src/diffusers` readability improvements (#629)

ryanrussell
authored and
Prathik Rao
committed
Fix formula for noise levels in Karras scheduler and tests (#627)

sgrigory
authored and
Prathik Rao
committed
[CI] Fix onnxruntime installation order (#633)

anton-l
authored and
Prathik Rao
committed
Warning for too long prompts in DiffusionPipelines (Resolve #447) (#472)

shirayu
authored and
Prathik Rao
committed
Fix docs link to train_unconditional.py (#642)

AbdullahAlfaraj
authored and
Prathik Rao
committed
Remove deprecated `torch_device` kwarg (#623)

pcuenca
authored and
Prathik Rao
committed
refactor: `custom_init_isort` readability fixups (#631)

ryanrussell
authored and
Prathik Rao
committed
Remove inappropriate docstrings in LMS docstrings. (#634)

pcuenca
authored and
Prathik Rao
committed
Flax pipeline pndm (#583)

authored and
Prathik Rao
committed
Fix `SpatialTransformer` (#578)

ydshieh
authored and
Prathik Rao
committed
Add training example for DreamBooth. (#554)

authored and
Prathik Rao
committed
rebase off main
Prathik Rao
committed
[examples/dreambooth] don't pass tensor_format to scheduler. (#649)

patil-suraj
authored and
Prathik Rao
committed
[dreambooth] update install section (#650)

patil-suraj
authored and
Prathik Rao
committed
[DDIM, DDPM] fix add_noise (#648)

patil-suraj
authored and
Prathik Rao
committed
[Pytorch] add dep. warning for pytorch schedulers (#651)

kashif
authored and
Prathik Rao
committed
[CLIPGuidedStableDiffusion] remove set_format from pipeline (#653)

patil-suraj
authored and
Prathik Rao
committed
Fix onnx tensor format (#654)

anton-l
authored and
Prathik Rao
committed
Fix `main`: stable diffusion pipelines cannot be loaded (#655)

pcuenca
authored and
Prathik Rao
committed
Fix the LMS pytorch regression (#664)

anton-l
authored and
Prathik Rao
committed
Added script to save during textual inversion training. Issue 524 (#645)

isamu-isozaki
authored and
Prathik Rao
committed
[CLIPGuidedStableDiffusion] take the correct text embeddings (#667)

patil-suraj
authored and
Prathik Rao
committed
Update index.mdx (#670)

tmabraham
authored and
Prathik Rao
committed
[examples] update transfomers version (#665)

patil-suraj
authored and
Prathik Rao
committed
[gradient checkpointing] lower tolerance for test (#652)

patil-suraj
authored and
Prathik Rao
committed
Flax `from_pretrained`: clean up `mismatched_keys`. (#630)

pcuenca
authored and
Prathik Rao
committed
`trained_betas` ignored in some schedulers (#635)

vishnu-anirudh
authored and
Prathik Rao
committed
Renamed x -> hidden_states in resnet.py (#676)

daspartho
authored and
Prathik Rao
committed
Optimize Stable Diffusion (#371)

NouamaneTazi
authored and
Prathik Rao
committed
Allow resolutions that are not multiples of 64 (#505)

authored and
Prathik Rao
committed
refactor: update ldm-bert `config.json` url closes #675 (#680)

ryanrussell
authored and
Prathik Rao
committed
[docs] fix table in fp16.mdx (#683)

NouamaneTazi
authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
Fix slow tests (#689)

NouamaneTazi
authored and
Prathik Rao
committed
Fix BibText citation (#693)

osanseviero
authored and
Prathik Rao
committed
Add callback parameters for Stable Diffusion pipelines (#521)

jamestiotio
authored and
Prathik Rao
committed
[dreambooth] fix applying clip_grad_norm_ (#686)

patil-suraj
authored and
Prathik Rao
committed
Forgot to add the OG!

patrickvonplaten
authored and
Prathik Rao
committed
Flax: add shape argument to `set_timesteps` (#690)

pcuenca
authored and
Prathik Rao
committed
Fix type annotations on StableDiffusionPipeline.__call__ (#682)

tasercake
authored and
Prathik Rao
committed
Fix import with Flax but without PyTorch (#688)

pcuenca
authored and
Prathik Rao
committed
[Support PyTorch 1.8] Remove inference mode (#707)

patrickvonplaten
authored and
Prathik Rao
committed
[CI] Speed up slow tests (#708)

anton-l
authored and
Prathik Rao
committed
[Utils] Add deprecate function and move testing_utils under utils (#659)

patrickvonplaten
authored and
Prathik Rao
committed
Checkpoint conversion script from Diffusers => Stable Diffusion (CompVis) (#701)

authored and
Prathik Rao
committed
[Docs] fix docstring for issue #709 (#710)

kashif
authored and
Prathik Rao
committed
Update schedulers README.md (#694)

authored and
Prathik Rao
committed
add accelerate to load models with smaller memory footprint (#361)

authored and
Prathik Rao
committed
Fix typos (#718)

authored and
Prathik Rao
committed
Add an argument "negative_prompt" (#549)

authored and
Prathik Rao
committed
Fix import if PyTorch is not installed (#715)

pcuenca
authored and
Prathik Rao
committed
Remove comments no longer appropriate (#716)

pcuenca
authored and
Prathik Rao
committed
[train_unconditional] fix applying clip_grad_norm_ (#721)

patil-suraj
authored and
Prathik Rao
committed
renamed x to meaningful variable in resnet.py (#677)

authored and
Prathik Rao
committed
[Tests] Add accelerate to testing (#729)

patrickvonplaten
authored and
Prathik Rao
committed
[dreambooth] Using already created `Path` in dataset (#681)

DrInfiniteExplorer
authored and
Prathik Rao
committed
Include CLIPTextModel parameters in conversion (#695)

kanewallmann
authored and
Prathik Rao
committed
Avoid negative strides for tensors (#717)

shirayu
authored and
Prathik Rao
committed
[Pytorch] pytorch only timesteps (#724)

authored and
Prathik Rao
committed
[Scheduler design] The pragmatic approach (#719)

anton-l
authored and
Prathik Rao
committed
Removing `autocast` for `35-25% speedup`. (`autocast` considered harmful). (#511)

authored and
Prathik Rao
committed
No more use_auth_token=True (#733)

patrickvonplaten
authored and
Prathik Rao
committed
remove use_auth_token from remaining places (#737)

patil-suraj
authored and
Prathik Rao
committed
Replace messages that have empty backquotes (#738)

pcuenca
authored and
Prathik Rao
committed
[Docs] Advertise fp16 instead of autocast (#740)

patrickvonplaten
authored and
Prathik Rao
committed
make style

patrickvonplaten
authored and
Prathik Rao
committed
remove use_auth_token from for TI test (#747)

patil-suraj
authored and
Prathik Rao
committed
allow multiple generations per prompt (#741)

patil-suraj
authored and
Prathik Rao
committed
Add back-compatibility to LMS timesteps (#750)

anton-l
authored and
Prathik Rao
committed
update the clip guided PR according to the new API (#751)

patil-suraj
authored and
Prathik Rao
committed
Raise an error when moving an fp16 pipeline to CPU (#749)

authored and
Prathik Rao
committed
Better steps deprecation for LMS (#753)

anton-l
authored and
Prathik Rao
committed
Actually fix the grad ckpt test (#734)

patil-suraj
authored and
Prathik Rao
committed
Custome Pipelines (#744)

authored and
Prathik Rao
committed
make CI happy

patrickvonplaten
authored and
Prathik Rao
committed
Python 3.7 doesn't like keys() + keys()

patrickvonplaten
authored and
Prathik Rao
committed
[v0.4.0] Temporarily remove Flax modules from the public API (#755)

anton-l
authored and
Prathik Rao
committed
Release: v0.4.0

anton-l
authored and
Prathik Rao
committed
Update clip_guided_stable_diffusion.py

patil-suraj
authored and
Prathik Rao
committed
Bump to v0.4.1.dev0

anton-l
authored and
Prathik Rao
committed
Revert "[v0.4.0] Temporarily remove Flax modules from the public API (#755)"

anton-l
authored and
Prathik Rao
committed
Bump to v0.5.0.dev0

anton-l
authored and
Prathik Rao
committed
Update clip_guided_stable_diffusion.py

patil-suraj
authored and
Prathik Rao
committed
Created using Colaboratory

patil-suraj
authored and
Prathik Rao
committed
[Tests] Lower required memory for clip guided and fix super edge-case git pipeline module bug (#754)

patrickvonplaten
authored and
Prathik Rao
committed
Revert "Bump to v0.5.0.dev0"

anton-l
authored and
Prathik Rao
committed
Change fp16 error to warning (#764)

authored and
Prathik Rao
committed
Release: v0.4.1

patrickvonplaten
authored and
Prathik Rao
committed
Bump to v0.5.0dev0

patrickvonplaten
authored and
Prathik Rao
committed
remove bogus folder

patrickvonplaten
authored and
Prathik Rao
committed
remove bogus folder no.2

patrickvonplaten
authored and
Prathik Rao
committed
Fix push_to_hub for dreambooth and textual_inversion (#748)

YaYaB
authored and
Prathik Rao
committed
Fix ONNX conversion script opset argument type (#739)

justinchuby
authored and
Prathik Rao
committed
Add final latent slice checks to SD pipeline intermediate state tests (#731)

jamestiotio
authored and
Prathik Rao
committed
fix(DDIM scheduler): use correct dtype for noise (#742)

keturn
authored and
Prathik Rao
committed
[schedulers] hanlde dtype in add_noise (#767)

patil-suraj
authored and
Prathik Rao
committed
[img2img, inpainting] fix fp16 inference (#769)

patil-suraj
authored and
Prathik Rao
committed
[Tests] Fix tests (#774)

patrickvonplaten
authored and
Prathik Rao
committed
debug an exception (#638)

authored and
Prathik Rao
committed
Clean up resnet.py file (#780)

Nathan Lambert
authored and
Prathik Rao
committed
add sigmoid betas (#777)

Nathan Lambert
authored and
Prathik Rao
committed
[Low CPU memory] + device map (#772)

authored and
Prathik Rao
committed
Fix gradient checkpointing test (#797)

patrickvonplaten
authored and
Prathik Rao
committed
fix typo docstring in unet2d (#798)

Nathan Lambert
authored and
Prathik Rao
committed
DreamBooth DeepSpeed support for under 8 GB VRAM training (#735)

Ttl
authored and
Prathik Rao
committed
support bf16 for stable diffusion (#792)

patil-suraj
authored and
Prathik Rao
committed
stable diffusion fine-tuning (#356)

authored and
Prathik Rao
committed
Flax: Trickle down `norm_num_groups` (#789)

akash5474
authored and
Prathik Rao
committed
Eventually preserve this typo? :) (#804)

spezialspezial
authored and
Prathik Rao
committed
Fix indentation in the code example (#802)

osanseviero
authored and
Prathik Rao
committed
`mps`: Alternative implementation for `repeat_interleave` (#766)

authored and
Prathik Rao
committed
Update img2img.mdx

patrickvonplaten
authored and
Prathik Rao
committed
Add diffusers version and pipeline class to the Hub UA

anton-l
authored and
Prathik Rao
committed
[Img2Img] Fix batch size mismatch prompts vs. init images (#793)

authored and
Prathik Rao
committed
Minor package fixes (#809)

anton-l
authored and
Prathik Rao
committed
[Dummy imports] Better error message (#795)

authored and
Prathik Rao
committed
Revert an accidental commit

anton-l
authored and
Prathik Rao
committed
add or fix license formatting in models directory (#808)

Nathan Lambert
authored and
Prathik Rao
committed
[train_text2image] Fix EMA and make it compatible with deepspeed. (#813)

patil-suraj
authored and
Prathik Rao
committed
Fix fine-tuning compatibility with deepspeed (#816)

pink-red
authored and
Prathik Rao
committed
Add diffusers version and pipeline class to the Hub UA (#814)

authored and
Prathik Rao
committed
[Flax] Add test (#824)

patrickvonplaten
authored and
Prathik Rao
committed
update flax scheduler API (#822)

patil-suraj
authored and
Prathik Rao
committed
Fix dreambooth loss type with prior_preservation and fp16 (#826)

anton-l
authored and
Prathik Rao
committed
Fix type mismatch error, add tests for negative prompts (#823)

anton-l
authored and
Prathik Rao
committed
Give more customizable options for safety checker (#815)

authored and
Prathik Rao
committed
Flax safety checker (#825)

authored and
Prathik Rao
committed
Align PT and Flax API - allow loading checkpoint from PyTorch configs (#827)

patrickvonplaten
authored and
Prathik Rao
committed
[Flax] Complete tests (#828)

patrickvonplaten
authored and
Prathik Rao
committed
Release: 5.0.0 (#830)

anton-l
authored and
Prathik Rao
committed
[FlaxStableDiffusionPipeline] fix bug when nsfw is detected (#832)

patil-suraj
authored and
Prathik Rao
committed
Release 0 5 1 (#833)

patrickvonplaten
authored and
Prathik Rao
committed
[Community] One step unet (#840)

patrickvonplaten
authored and
Prathik Rao
committed
Remove unneeded use_auth_token (#839)

osanseviero
authored and
Prathik Rao
committed
Bump to 0.6.0.dev0 (#831)

authored and
Prathik Rao
committed
Remove the last of ["sample"] (#842)

anton-l
authored and
Prathik Rao
committed
Fix Flax pipeline: width and height are ignored #838 (#848)

camenduru
authored and
Prathik Rao
committed
[DeviceMap] Make sure stable diffusion can be loaded from older trans… (#860)

patrickvonplaten
authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
Add Stable Diffusion Interpolation Example (#862)

authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
All in one Stable Diffusion Pipeline (#821)

patrickvonplaten
authored and
Prathik Rao
committed
Fix small community pipeline import bug and finish README (#869)

patrickvonplaten
authored and
Prathik Rao
committed
Update README.md

patrickvonplaten
authored and
Prathik Rao
committed
Fix training push_to_hub (unconditional image generation): models were not saved before pushing to hub (#868)

pcuenca
authored and
Prathik Rao
committed
Fix table in community README.md (#879)

nateraw
authored and
Prathik Rao
committed
Add generic inference example to community pipeline readme (#874)

apolinario
authored and
Prathik Rao
committed
Rename frame filename in interpolation community example (#881)

nateraw
authored and
Prathik Rao
committed
Add Apple M1 tests (#796)

authored and
Prathik Rao
committed
Fix autoencoder test (#886)

pcuenca
authored and
Prathik Rao
committed
Rename StableDiffusionOnnxPipeline -> OnnxStableDiffusionPipeline (#887)

anton-l
authored and
Prathik Rao
committed
Fix DDIM on Windows not using int64 for timesteps (#819)

hafriedlander
authored and
Prathik Rao
committed
[dreambooth] allow fine-tuning text encoder (#883)

patil-suraj
authored and
Prathik Rao
committed
Stable Diffusion image-to-image and inpaint using onnx. (#552)

authored and
Prathik Rao
committed
Improve ONNX img2img numpy handling, temporarily fix the tests (#899)

anton-l
authored and
Prathik Rao
committed
make fix copies

patrickvonplaten
authored and
Prathik Rao
committed
[Stable Diffusion Inpainting] Deprecate inpainting pipeline in favor of official one (#903)

authored and
Prathik Rao
committed
[Communit Pipeline] Make sure "mega" uses correct inpaint pipeline (#908)

patrickvonplaten
authored and
Prathik Rao
committed
Stable diffusion inpainting. (#904)

authored and
Prathik Rao
committed
finish tests (#909)

patrickvonplaten
authored and
Prathik Rao
committed
ONNX supervised inpainting (#906)

anton-l
authored and
Prathik Rao
committed
Initial docs update for new in-painting pipeline (#910)

authored and
Prathik Rao
committed
Release: 0.6.0

anton-l
authored and
Prathik Rao
committed
[Community Pipelines] Long Prompt Weighting Stable Diffusion Pipelines (#907)

authored and
Prathik Rao
committed
[Stable Diffusion] Add components function (#889)

patrickvonplaten
authored and
Prathik Rao
committed
[PNDM Scheduler] Make sure list cannot grow forever (#882)

patrickvonplaten
authored and
Prathik Rao
committed
[DiffusionPipeline.from_pretrained] add warning when passing unused k… (#870)

patrickvonplaten
authored and
Prathik Rao
committed
DOC Dreambooth Add --sample_batch_size=1 to the 8 GB dreambooth example script (#829)

leszekhanusz
authored and
Prathik Rao
committed
[Examples] add speech to image pipeline example (#897)

MikailINTech
authored and
Prathik Rao
committed
[dreambooth] dont use safety check when generating prior images (#922)

patil-suraj
authored and
Prathik Rao
committed
Dreambooth class image generation: using unique names to avoid overwriting existing image (#847)

leszekhanusz
authored and
Prathik Rao
committed
fix test_components (#928)

patil-suraj
authored and
Prathik Rao
committed
Fix Compatibility with Nvidia NGC Containers (#919)

tasercake
authored and
Prathik Rao
committed
[Community Pipelines] Fix pad_tokens_and_weights in lpw_stable_diffusion (#925)

SkyTNT
authored and
Prathik Rao
committed
Bump the version to 0.7.0.dev0 (#912)

anton-l
authored and
Prathik Rao
committed
Introduce the copy mechanism (#924)

anton-l
authored and
Prathik Rao
committed
Merge branch 'prathikrao/ort-integration' of https://github.com/prathikr/diffusers into prathikrao/ort-integration
Prathik Rao
committed
updated unet tests
Prathik Rao
committed
[gradient checkpointing] lower tolerance for test (#652)

patil-suraj
authored and
Prathik Rao
committed
[Tests] Add accelerate to testing (#729)

patrickvonplaten
authored and
Prathik Rao
committed
Created using Colaboratory

patil-suraj
authored and
Prathik Rao
committed
remove bogus folder no.2

patrickvonplaten
authored and
Prathik Rao
committed
Add generic inference example to community pipeline readme (#874)

apolinario
authored and
Prathik Rao
committed
Add Apple M1 tests (#796)

authored and
Prathik Rao
committed
Fix autoencoder test (#886)

pcuenca
authored and
Prathik Rao
committed
updated unet tests
Prathik Rao
committed
Merge branch 'prathikrao/ort-integration' of https://github.com/prathikr/diffusers into prathikrao/ort-integration
Prathik Rao
committed
remove random code add
Prathik Rao
committed
remove random code add
Prathik Rao
committed
[Tests] Add accelerate to testing (#729)

patrickvonplaten
authored and
Prathik Rao
committed
Created using Colaboratory

patil-suraj
authored and
Prathik Rao
committed
remove bogus folder no.2

patrickvonplaten
authored and
Prathik Rao
committed
Add generic inference example to community pipeline readme (#874)

apolinario
authored and
Prathik Rao
committed
Add Apple M1 tests (#796)

authored and
Prathik Rao
committed
Fix autoencoder test (#886)

pcuenca
authored and
Prathik Rao
committed
rebase
Prathik Rao
committed
[Tests] Add accelerate to testing (#729)

patrickvonplaten
authored and
Prathik Rao
committed
Created using Colaboratory

patil-suraj
authored and
Prathik Rao
committed
remove bogus folder no.2

patrickvonplaten
authored and
Prathik Rao
committed
Add generic inference example to community pipeline readme (#874)

apolinario
authored and
Prathik Rao
committed
Add Apple M1 tests (#796)

authored and
Prathik Rao
committed
Fix autoencoder test (#886)

pcuenca
authored and
Prathik Rao
committed
rebase
Prathik Rao
committed
remove random code add
Prathik Rao
committed
Merge branch 'prathikrao/ort-integration' of https://github.com/prathikr/diffusers into prathikrao/ort-integration
Prathik Rao
committed

Commits on Oct 31, 2022

seperate script for ort
Prathik Rao
committed
formatting
Prathik Rao
committed
removed random eval statement
Prathik Rao
committed