feat: allow disk offload for diffuser models #3285
patrickvonplaten
merged
85 commits into
huggingface:main
from
hari10599:hkr/disk-offload
allow disk offload for diffuser models
8511176f
sort import
26e322d7
hari10599
changed the title 
patrickvonplaten
approved these changes
on 2023-05-03
sayakpaul
approved these changes
on 2023-05-03
add max_memory argument
6653047c
Changed sample[0] to images[0] (#3304)
87c39683
Typo in tutorial (#3295)
4a16aab3
Torch compile graph fix (#3286)
73e9c24c
Postprocessing refactor img2img (#3268)
1786d33d
[Torch 2.0 compile] Fix more torch compile breaks (#3313)
05c6060c
fix: scale_lr and sync example readme and docs. (#3299)
7a5811b8
Update stable_diffusion.mdx (#3310)
f7b042c0
Fix missing variable assign in DeepFloyd-IF-II (#3315)
2d3067e6
Correct doc build for patch releases (#3316)
add3f977
Add Stable Diffusion RePaint to community pipelines (#3320)
7be5b0a5
Fix multistep dpmsolver for cosine schedule (suitable for deepfloyd-i…
22bfb080
[docs] Improve LoRA docs (#3311)
ab0f138d
Added input pretubation (#3292)
4f32c186
Update write_own_pipeline.mdx (#3323)
5faa00f7
update controlling generation doc with latest goodies. (#3321)
6316e880
[Quality] Make style (#3341)
ffe37384
Fix config dpm (#3343)
880a83b3
Add the SDE variant of DPM-Solver and DPM-Solver++ (#3344)
ae6444dc
Add upsample_size to AttnUpBlock2D, AttnDownBlock2D (#3275)
95327190
Rename --only_save_embeds to --save_as_full_pipeline (#3206)
7315d6d9
[AudioLDM] Generalise conversion script (#3328)
de969901
Fix TypeError when using prompt_embeds and negative_prompt (#2982)
7161fbd9
Fix pipeline class on README (#3345)
75ed7898
Inpainting: typo in docs (#3331)
3572fd8e
Add `use_Karras_sigmas` to LMSDiscreteScheduler (#3351)
5c51eabd
Batched load of textual inversions (#3277)
9c6ae6c6
make fix-copies
bd9cf768
[docs] Fix docstring (#3334)
361b62fd
if dreambooth lora (#3360)
92c2b0c3
Postprocessing refactor all others (#3337)
39e6998b
[docs] Improve safetensors docstring (#3368)
fc4e4f42
add: a warning message when using xformers in a PT 2.0 env. (#3365)
af6e35aa
StableDiffusionInpaintingPipeline - resize image w.r.t height and wid…
7fa90fbb
make style
45b86c9d
[docs] Adapt a model (#3326)
dbe33168
[docs] Load safetensors (#3333)
61743256
make style
9c99e8f4
[Docs] Fix stable_diffusion.mdx typo (#3398)
1e490737
Support ControlNet v1.1 shuffle properly (#3340)
7275de1f
[Tests] better determinism (#3374)
6e4b195f
[docs] Add transformers to install (#3388)
5b8ba3a3
[deepspeed] partial ZeRO-3 support (#3076)
3bab713a
Add omegaconf for tests (#3400)
8b665340
Fix various bugs with LoRA Dreambooth and Dreambooth script (#3353)
574b6c8c
Fix docker file (#3402)
137a7a3a
fix: deepseepd_plugin retrieval from accelerate state (#3410)
24581198
[Docs] Add `sigmoid` beta_scheduler to docstrings of relevant Schedul…
ce28477d
Don't install accelerate and transformers from source (#3415)
c54bea12
Don't install transformers and accelerate from source (#3414)
70090a10
Improve fast tests (#3416)
0cb751cf
attention refactor: the trilogy (#3387)
0f3cedaf
[Docs] update the PT 2.0 optimization doc with latest findings (#3370)
42c4bfe8
Fix style rendering (#3433)
341f9074
unCLIP scheduler do not use note (#3417)
a1ce3e7a
Replace deprecated command with environment file (#3409)
fff76605
fix warning message pipeline loading (#3446)
2592d97b
add stable diffusion tensorrt img2img pipeline (#3419)
728db023
Refactor controlnet and add img2img and inpaint (#3386)
a2b64782
[Scheduler] DPM-Solver (++) Inverse Scheduler (#3335)
34e8868e
[Docs] Fix incomplete docstring for resnet.py (#3438)
83280ac6
fix tiled vae blend extent range (#3384)
f84485b8
Small update to "Next steps" section (#3443)
bf0b0e35
Allow arbitrary aspect ratio in IFSuperResolutionPipeline (#3298)
fa9a44a9
Adding 'strength' parameter to StableDiffusionInpaintingPipeline (#3…
8804feef
[WIP] Bugfix - Pipeline.from_pretrained is broken when the pipeline i…
2306dc4e
Fix gradient checkpointing bugs in freezing part of models (requires_…
32715314
Make dreambooth lora more robust to orig unet (#3462)
0bde5e45
Reduce peak VRAM by releasing large attention tensors (as soon as the…
1e677882
Add min snr to text2img lora training script (#3459)
e3cfd354
Add inpaint lora scale support (#3460)
78771bf6
[From ckpt] Fix from_ckpt (#3466)
4fd6e7ba
Update full dreambooth script to work with IF (#3425)
327b94bc
Add IF dreambooth docs (#3470)
13b2226a
parameterize pass single args through tuple (#3477)
74e6eb6d
attend and excite tests disable determinism on the class level (#3478)
0691dec6
dreambooth docs torch.compile note (#3471)
b05d5b42
add: if entry in the dreambooth training docs. (#3472)
7683b56d
[docs] Textual inversion inference (#3473)
0562e740
[docs] Distributed inference (#3376)
0e1f339d
[{Up,Down}sample1d] explicit view kernel size as number elements in f…
573b5d44
mps & onnx tests rework (#3449)
dfc25493
Merge branch 'huggingface:main' into hkr/disk-offload
49a56d01
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub