Finegrain/refiners
1
0
Fork 0
mirror of https://github.com/finegrain-ai/refiners.git synced 2024-11-22 06:08:46 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
702 commits 35 branches 5 tags 149 MiB
Commit graph

702 commits

This branch
This branch
All branches
Author SHA1 Message Date
Laurent 241abfafe4 deprecate LatentDiffusionAutoencoder's decode_latents 2024-10-15 15:19:36 +02:00
Laurent d8f77dd880 add sdxl.t2i_adapter.depth.zoe + rework mappings of t2i adapters 2024-10-15 15:17:16 +02:00
Laurent 9b19272d75 check conversion deps are installed when using refiners.conversion 2024-10-15 10:51:45 +02:00
Laurent 8f31039c4a remove top-level notebooks directory 2024-10-15 10:23:49 +02:00
Laurent ef0a7525a8 add DINOv2 similarity to compare_images + relax some test constraints 2024-10-14 15:12:59 +02:00
Laurent d498ecd369 update CONTRIBUTING.md 2024-10-14 15:12:59 +02:00
Laurent 316fe6e4f0 update tests to use new fixtures 2024-10-14 15:12:59 +02:00
Laurent 94eeb1afc3 add new weight related test fixtures 2024-10-14 15:12:59 +02:00
Laurent d322e9c5ed add new weight conversion logic 2024-10-14 15:12:59 +02:00
Laurent 4045904b26 remove gdown dependency 2024-10-14 15:12:59 +02:00
Laurent 189cfa1a69 delete old conversion scripts 2024-10-14 15:12:59 +02:00
Pierre Chapuis 2796117d2d properly check for bfloat16 
- we check only the test device, not the machine in general
- we don't want emulated bfloat16 (e.g. CPU)
 2024-10-09 11:20:27 +02:00
Laurent f3d2b6c325 modify some foundational tests to also test in float16 and bfloat16 2024-10-03 11:05:09 +02:00
Laurent b20474f8f5 add various torch.dtype test fixtures 2024-10-03 11:05:09 +02:00
Laurent 16714e6745 gradute comfyui-refiners get_dtype to refiners 2024-10-03 11:05:09 +02:00
Laurent 4360aa046f add test_diffusion_std_random_init_bfloat16 e2e test 2024-10-03 11:05:09 +02:00
Laurent 12622ad114 add pytest-rerunfailures optional test dependency 2024-10-03 11:05:09 +02:00
Pierre Chapuis ae5ef4f14c make sure adapters can access the context of their parent chain 
https://github.com/finegrain-ai/refiners/issues/456
 2024-10-03 10:51:11 +02:00
TyrianOtter 590648ebca fix CLIPTokenizer skipping underscores 2024-10-03 09:38:49 +02:00
Laurent f89c4f720d add test_multi_upscaler_small to check that MultiUpscaler works with small (16x16px) images 2024-10-01 11:14:52 +02:00
Laurent 7b6f97347e update some reference test images 2024-10-01 11:09:57 +02:00
Laurent 18ef5684c0 add missing remove_noise method from the DPMSolver class 2024-10-01 11:09:57 +02:00
Laurent 8bd405661d rewrite project.urls in pyproject.toml and add keywords 2024-10-01 11:09:57 +02:00
Laurent d816d3d24c update requirements.lock 2024-10-01 11:09:57 +02:00
TyrianOtter 8352821984
fix DPMSolver.__call__ docs calling it deterministic (#452) 
Co-authored-by: Cédric Deltheil <355031+deltheil@users.noreply.github.com>
 2024-09-30 10:20:29 +02:00
Laurent 07fc6193ed rework compute_clip_image_embedding overloads + improve docstring 2024-09-27 12:25:45 +02:00
Laurent 9e2a499262 pyright: ignore "tests/repos" 2024-09-27 12:25:45 +02:00
Laurent 5a92285104 add torch.Generator to MultiUpscaler.upscale + make MultiUpscaler.diffuse_targets "stateless"
Some checks failed
CI / lint_and_typecheck (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details
Spell checker / Spell check (push) Has been cancelled
Details
2024-09-26 11:31:00 +02:00
Pierre Chapuis 883a2121f2 fix precision of DPM solver in bfloat16 2024-09-26 08:55:07 +02:00
Pierre Chapuis ed7e2e5964 check timesteps stay the same in bfloat16 2024-09-26 08:55:07 +02:00
Pierre Chapuis f1c8503e68 use default (1e-8) atol for basic DPM solver (used in SD1.5) 2024-09-26 08:55:07 +02:00
Pierre Chapuis 83b931296f fix DPM-Solver with bfloat16
Some checks are pending
CI / lint_and_typecheck (push) Waiting to run
Details
Deploy docs to GitHub Pages / Deploy docs (push) Waiting to run
Details
Spell checker / Spell check (push) Waiting to run
Details
2024-09-25 14:41:19 +02:00
Cédric Deltheil 283bf457dc
README.md: fix typo
Some checks are pending
CI / lint_and_typecheck (push) Waiting to run
Details
Deploy docs to GitHub Pages / Deploy docs (push) Waiting to run
Details
Spell checker / Spell check (push) Waiting to run
Details
2024-09-24 14:46:44 +02:00
Cédric Deltheil ae480dc684
README.md: add https://github.com/chloedia/layerdiffuse 2024-09-24 14:45:01 +02:00
Laurent 425ed96cad (comfyui-refiners) set lower version constraints for transformers and huggingface-hub dependencies
Some checks failed
CI / lint_and_typecheck (push) Has been cancelled
Details
Publish to Comfy registry / Publish Custom Node to registry (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details
Spell checker / Spell check (push) Has been cancelled
Details
2024-09-20 21:05:35 +02:00
Cédric Deltheil 49ed1f32a1 pyproject.toml: silent typos for imaginAIry 2024-09-20 16:07:02 +02:00
Hugues Pouillot 986cb167ff add editor link 2024-09-20 16:07:02 +02:00
Pierre Chapuis a542337a83 fix typing
Some checks are pending
CI / lint_and_typecheck (push) Waiting to run
Details
Deploy docs to GitHub Pages / Deploy docs (push) Waiting to run
Details
Spell checker / Spell check (push) Waiting to run
Details 
see https://github.com/microsoft/pyright/issues/9025
 2024-09-19 15:30:11 +02:00
Laurent cf270885a4 Update comfyui-refiners README.md 2024-09-19 14:17:50 +02:00
Pierre Chapuis 73089a4e2d fix typing issue in dinov2 test
Some checks failed
CI / lint_and_typecheck (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details
Spell checker / Spell check (push) Has been cancelled
Details
2024-09-18 10:44:24 +02:00
Pierre Chapuis 336253f26b remove unused argument in WindowSDPA
Some checks failed
CI / lint_and_typecheck (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details
Spell checker / Spell check (push) Has been cancelled
Details
2024-09-12 16:02:13 +02:00
Pierre Chapuis 31b5f80496 simplify implementation of load_from_safetensors 2024-09-12 15:14:14 +02:00
Laurent 2c0174f50e rename test_unet.py to test_sd15_unet.py + use test_device fixture
Some checks failed
CI / lint_and_typecheck (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details
Spell checker / Spell check (push) Has been cancelled
Details
2024-09-10 09:16:22 +02:00
Laurent 4eceda810c export/expose SD1Autoencoder and SDXLAutoencoder + some formatting 2024-09-10 09:16:22 +02:00
Laurent 602c63ce76 add missing device and dtype to SD1UNet's UpBlocks 2024-09-10 09:16:22 +02:00
Laurent 444882a734 move some tests into the adapters test folder
Some checks are pending
CI / lint_and_typecheck (push) Waiting to run
Details
Deploy docs to GitHub Pages / Deploy docs (push) Waiting to run
Details
Spell checker / Spell check (push) Waiting to run
Details
2024-09-09 17:44:26 +02:00
Laurent c6631c7ca6 implement _add_noise for dpm solver 2024-09-09 17:25:31 +02:00
Laurent 6d58492097 loosen the tolerances for some sam tests (because of a recent pytorch upgrade) 2024-09-09 17:17:12 +02:00
Laurent 2395e666d5 update gaussian_blur fluxion util, see 45e053b2ae 2024-09-09 17:05:15 +02:00
Pierre Chapuis a51d695523 MultiUpscaler: improve resizing logic
Some checks failed
Spell checker / Spell check (push) Has been cancelled
Details
CI / lint_and_typecheck (push) Has been cancelled
Details
Deploy docs to GitHub Pages / Deploy docs (push) Has been cancelled
Details 
- Do not have a parameter to downscale (caller can do it beforehand if they want).
- Do not enforce mod 8 *before* upscale, we need it afterwards only (before SD).
 2024-09-06 18:23:42 +02:00