Commit Graph

Select branches

Hide Pull Requests

main

patch-1

0.0.1

0.0.2

0.1.0

0.1.0a

0.1.1

0.1.2

0.1.3

0.1.4

0.1.5

0.10.0

0.10.1

0.11.0

0.11.1

0.11.2

0.12.0

0.12.1

0.14.0

0.14.1

0.14.2

0.14.3

0.15.0

0.15.1

0.15.2

0.15.3

0.15.4

0.15.6

0.15.7

0.2.0

0.2.1

0.2.2

0.2.3

0.2.4

0.26.3

0.26.4

0.26.5

0.27.0

0.27.1

0.27.10

0.27.11

0.27.12

0.27.2

0.27.3

0.27.4

0.27.5

0.27.6

0.27.8

0.27.9

0.28.0

0.29.0

0.29.1

0.3.0

0.3.1

0.3.2

0.30.0

0.31.0

0.31.1

0.32.0

0.4.0

0.5.0

0.5.1

0.5.2

0.6.0

0.6.1

0.6.2

0.6.3

0.6.4

0.6.5

0.6.6

0.6.7

0.6.8

0.6.9

0.7.0

0.7.1

0.8.0

0.8.1

0.9.0

0.9.1

0.9.2

1.16.6

v0.16.0

v0.16.1

v0.16.2

v0.16.3

v0.16.4

v0.16.5

v0.16.6

v0.16.7

v0.17.0

v0.17.1

v0.17.2

v0.17.3

v0.17.4

v0.17.5

v0.17.6

v0.17.7

v0.18.0

v0.18.1

v0.18.2

v0.18.3

v0.18.4

v0.19.0

v0.19.1

v0.19.2

v0.20.0

v0.20.1

v0.20.2

v0.20.2a

v0.21.0

v0.21.1

v0.21.2

v0.22.0

v0.23.0

v0.23.1

v0.23.2

v0.23.3

v0.23.4

v0.24.0

v0.24.1

v0.24.2

v0.24.3

v0.24.4

v0.25.0

v0.25.1

v0.25.2

v0.25.3

v0.26.0

v0.26.1

Mono Color

• 88922929cb Update classifier_free_guidance.py main wassname 2022-12-22 11:01:37 +08:00
• edc6d5136a Update denoising_diffusion_pytorch_1d.py patch-1 wassname 2022-12-22 11:00:56 +08:00
• 207d23af86 Update denoising_diffusion_pytorch.py wassname 2022-12-22 10:57:30 +08:00
• cf4f44909f 0.1.5 0.1.5 Phil Wang 2022-12-21 08:35:59 -08:00
• a7f2d670bb always set results_folder on Trainer Phil Wang 2022-12-21 08:34:34 -08:00
• 6ab29d5cea start saving version of library in the checkpoints, so in case of breaking changes, researchers can return to the right working version 0.1.0a Phil Wang 2022-11-28 12:11:11 -08:00
• d220624a30 start saving version of library in the checkpoints, so in case of breaking changes, researchers can return to the right working version 1.16.6 Phil Wang 2022-11-28 12:08:49 -08:00
• ddc31bc489 trust paper 0.32.0 Phil Wang 2022-11-27 10:05:55 -08:00
• e872ec3618 tweak readme Phil Wang 2022-11-21 10:08:56 -08:00
• 60f760ead5 add classifier free guidance example for @ajrheng at https://github.com/lucidrains/denoising-diffusion-pytorch/issues/120 0.31.1 Phil Wang 2022-11-11 20:01:42 -08:00
• 6205e20e5e ocd Phil Wang 2022-11-05 17:09:34 -07:00
• aece9c2a2a help researchers in need of 1d ddpm, ex. https://github.com/lucidrains/denoising-diffusion-pytorch/issues/116 Phil Wang 2022-11-05 11:42:00 -07:00
• caf9a2dc75 help researchers in need of 1d ddpm, ex. https://github.com/lucidrains/denoising-diffusion-pytorch/issues/116 0.31.0 Phil Wang 2022-11-05 11:40:27 -07:00
• 12079cadee add the v-parameterization from Salimans et al for the discrete case, allow for distillation 0.30.0 Phil Wang 2022-10-30 09:53:28 -07:00
• 0ffff59ca0 add option for random fourier features, given misinterpretation of Katherine's code, thanks to @tmabraham for addressing this in https://github.com/lucidrains/denoising-diffusion-pytorch/issues/112 0.29.1 Phil Wang 2022-10-30 09:32:06 -07:00
• aadaa7d288 bring in the continuous time v-parameterized ddpm, validated to work locally, and which will be used for imagen-video replication 0.29.0 Phil Wang 2022-10-23 19:16:31 -07:00
• 23fd887a5f switch back to regular attention, given @rromb results Phil Wang 2022-10-16 08:43:10 -07:00
• f762d33c17 switch back to regular attention, given @rromb results 0.28.0 Phil Wang 2022-10-16 08:36:17 -07:00
• dfbafee555 0.27.12 0.27.12 Phil Wang 2022-10-05 13:50:54 -07:00
• 40dd8ba1de Merge pull request #102 from npielawski/main Phil Wang 2022-10-05 13:50:39 -07:00
• 2ac3f94a80 Added gradient clipping. Nicolas Pielawski 2022-10-05 11:29:49 -07:00
• 98f2eeac35 link to flax implementation from @yiyixuxu Phil Wang 2022-09-27 11:21:23 -07:00
• 6e8a0f2082 fix auto-conversion of images to mode in dataset 0.27.11 Phil Wang 2022-09-20 19:29:17 -07:00
• 8c36559295 0.27.10 0.27.10 Phil Wang 2022-09-16 17:15:02 -07:00
• f74f536339 Merge pull request #90 from kashif/patch-1 Phil Wang 2022-09-16 17:14:48 -07:00
• d85b8bbe2e fix torch.cumprod Kashif Rasul 2022-09-16 17:19:00 +02:00
• e0a1bed31a 0.27.9 0.27.9 Phil Wang 2022-09-05 02:43:11 -07:00
• 82b67fc00a Merge pull request #85 from RyannDaGreat/main Phil Wang 2022-09-05 02:42:49 -07:00
• 7c0cd05c27 Trainer.load can use GPU's other than cuda:0 Ryan Burgert 2022-09-04 22:11:35 -04:00
• 6dda508ff6 in ddim, clip x0 before calculation of predicted noise, thanks to @lukovnikov again for pointing out this inconsistency with glides implementation 0.27.8 Phil Wang 2022-09-01 09:34:50 -07:00
• 9ec8d27217 0.27.7 Phil Wang 2022-08-31 07:36:19 -07:00
• 4b4ebab7c3 Merge pull request #83 from lukovnikov/fix_ddim Phil Wang 2022-08-31 07:02:26 -07:00
• e4a4e4acaa Revert "Revert "fix ddim sampling"" lukovnikov 2022-08-31 15:28:35 +02:00
• cd8329cdd7 Revert "fix ddim sampling" lukovnikov 2022-08-31 15:25:51 +02:00
• aec2a26984 fix ddim sampling lukovnikov 2022-08-31 15:24:59 +02:00
• c78709f887 0.27.6 0.27.6 Phil Wang 2022-08-31 06:20:59 -07:00
• 4436128a0b Merge pull request #82 from TheDudeFromCI/patch-1 Phil Wang 2022-08-31 06:20:31 -07:00
• e46a89e2bc Update step before training checkpoint TheDudeFromCI 2022-08-31 02:19:12 -07:00
• 42158d6248 fix ddim, for issue https://github.com/lucidrains/denoising-diffusion-pytorch/issues/81 0.27.5 Phil Wang 2022-08-30 20:30:29 -07:00
• 44f95e2e9d readme Phil Wang 2022-08-22 09:13:32 -07:00
• d9275a744c add weight standardization prior to groupnorm, lessen cosine sim attention scale to 10 for fp16 0.27.4 Phil Wang 2022-08-17 11:42:54 -07:00
• 5389c1a1a1 add weight standardization prior to groupnorm, lessen cosine sim attention scale to 10 for fp16 0.27.3 Phil Wang 2022-08-17 11:40:41 -07:00
• beb2f2d8dd add self conditioning for elucidated ddpm 0.27.2 Phil Wang 2022-08-10 13:15:45 -07:00
• f0d59acdfd fix sampling ddpm tqdm 0.27.1 Phil Wang 2022-08-10 12:02:56 -07:00
• 689593a579 add the new self conditioning technique from hintons group from bit diffusion paper 0.27.0 Phil Wang 2022-08-10 10:52:07 -07:00
• eba44498d1 higher epsilon for fp16 in layernorm 0.26.5 Phil Wang 2022-07-29 13:29:52 -07:00
• 12f95b33d8 rescale values to prevent linear attention from overflowing in fp16 setting 0.26.4 Phil Wang 2022-07-27 12:25:40 -07:00
• 6b504c4ae9 fix accelerator prepare bug for dataloader 0.26.3 Phil Wang 2022-07-25 08:13:46 -07:00
• 37334ae824 fix a bug with ddim and predict x0 objective v0.26.1 Phil Wang 2022-07-18 19:04:57 -07:00
• 6eba6cdd50 refresh images Phil Wang 2022-07-18 11:30:34 -07:00
• 555566c188 take a gamble on cosine sim attention v0.26.0 Phil Wang 2022-07-18 11:29:20 -07:00
• 2b742dd2cc move accelerator backward outside of autocast context, also calculate total loss correctly across gradient accumulated steps v0.25.3 Phil Wang 2022-07-11 21:02:06 -07:00
• 1345a8a41d do not noise at the last timestep for ddim v0.25.2 Phil Wang 2022-07-09 18:36:45 -07:00
• 931a5af2c3 bring in ddim sampling v0.25.1 Phil Wang 2022-07-09 16:10:23 -07:00
• cb4d57149e bring in ddim sampling v0.25.0 Phil Wang 2022-07-09 15:22:53 -07:00
• a0c3443eaa optimizer should be saved and loaded v0.24.4 Phil Wang 2022-07-08 17:43:41 -07:00
• cee30ce304 optimizer should be saved and loaded v0.24.3 Phil Wang 2022-07-08 17:07:32 -07:00
• 662172851b add convert_image_to keyword argument, for forcing images being loaded to be converted to some format, greyscale, rgb, rgba, whatever v0.24.2 Phil Wang 2022-07-08 09:22:59 -07:00
• 0248b5e4d3 also make sure grad scaler actually exists in the saved pt file v0.24.1 Phil Wang 2022-07-06 11:48:04 -07:00
• 6b56af08a2 support multi-gpu training using huggingface accelerate, addressing https://github.com/lucidrains/denoising-diffusion-pytorch/pull/54 v0.24.0 Phil Wang 2022-07-06 11:45:56 -07:00
• d4420248f1 tqdm auto instead v0.23.4 Phil Wang 2022-06-29 19:48:49 -07:00
• a536e5bee9 reorganize elucidating code Phil Wang 2022-06-29 08:49:38 -07:00
• 1b85379d3a add clamping option to elucidated diffusion v0.23.3 Phil Wang 2022-06-29 08:32:10 -07:00
• 8859864f63 patch v0.23.2 Phil Wang 2022-06-29 07:55:04 -07:00
• 32657f035f Merge pull request #52 from AryaAftab/patch-1 Phil Wang 2022-06-29 07:54:48 -07:00
• d97bc0278c Solve issue #26 Arya Aftab 2022-06-29 11:42:54 +04:30
• 8408775cfc fix bug in elucidating sampling v0.23.1 Phil Wang 2022-06-28 17:52:02 -07:00
• 86fcb6785b release elucidating diffusion v0.23.0 Phil Wang 2022-06-28 17:39:48 -07:00
• c535d31fc5 Merge pull request #51 from lucidrains/pw/elucidating-ddpm Phil Wang 2022-06-28 17:29:28 -07:00
• 5db64fec4b refactor sigmas and gamma generation Phil Wang 2022-06-28 17:26:58 -07:00
• b87ea27781 fix off by one Phil Wang 2022-06-28 16:05:21 -07:00
• a8403b83fe no clamping when training from sigmas drawn from log normal distribution, clamp final images being sampled Phil Wang 2022-06-28 16:00:16 -07:00
• f4b1d7a67c complete a first pass of elucidated ddpm Phil Wang 2022-06-28 15:15:21 -07:00
• 618493714f clamp the sigma coming out of the log normal distribution Phil Wang 2022-06-28 14:34:39 -07:00
• 76b79aa847 take care of equation 7 in the paper Phil Wang 2022-06-28 14:16:05 -07:00
• be2bd8d320 cleanup again Phil Wang 2022-06-28 13:42:41 -07:00
• c3d1607019 cleanup Phil Wang 2022-06-28 13:38:37 -07:00
• 06b2e52645 get training working Phil Wang 2022-06-28 13:35:06 -07:00
• 09b8a1c805 some basic scaffold for elucidating diffusion and derived values Phil Wang 2022-06-28 13:11:56 -07:00
• d26acbcae6 more skip connections, as in guided diffusion v0.22.0 Phil Wang 2022-06-27 13:23:32 -07:00
• 9939a48139 make sure all versions of torch supported v0.21.2 Phil Wang 2022-06-23 12:28:09 -07:00
• 75ea49a7ef pass parameter for Trainer to EMA properly v0.21.1 Phil Wang 2022-06-21 07:37:43 -07:00
• 8c3609a6e3 move EMA logic out of the repository for clarity v0.21.0 Phil Wang 2022-06-20 13:17:51 -07:00
• 1586d1a8a0 just pluck the image size off the gaussian diffusion class v0.20.2a Phil Wang 2022-06-17 13:54:41 -07:00
• b4fb8804d2 conditioning on final resnet block v0.20.2 v0.20.1 Phil Wang 2022-06-17 10:38:17 -07:00
• 9fd05f1b1f switch to learned sinsuoidal pos emb for the continuous case v0.20.0 Phil Wang 2022-06-17 09:24:51 -07:00
• ec2397f0ba add one more residual v0.19.2 Phil Wang 2022-06-16 11:08:42 -07:00
• 844e557dfb fix a missing residual needed at the top most resolution in the unet v0.19.1 Phil Wang 2022-06-15 19:10:05 -07:00
• b7c359fb49 fix a missing residual needed at the top most resolution in the unet v0.19.0 Phil Wang 2022-06-15 17:55:53 -07:00
• 8b30be8042 add p2 loss reweighting for default ddpm as an option v0.18.4 Phil Wang 2022-06-14 10:49:13 -07:00
• f2f3994b92 link to Letitia Phil Wang 2022-06-12 14:57:44 -07:00
• 8ec4ea56a5 link to yannic Phil Wang 2022-06-12 14:56:55 -07:00
• 99cf9b5b96 link to ai educator Phil Wang 2022-06-12 14:54:43 -07:00
• ecc6f30901 for https://github.com/lucidrains/denoising-diffusion-pytorch/issues/36 v0.18.3 Phil Wang 2022-06-11 10:51:13 -07:00
• f900f40f14 allow for turning off horizontal flip augmentation v0.18.2 Phil Wang 2022-06-09 20:59:59 -07:00
• 79c5f045e1 allow for turning off horizontal flip augmentation v0.18.1 Phil Wang 2022-06-09 20:52:44 -07:00
• 479f60c178 add p2 loss weighting to SNR version of denoising diffusion, brought up by @Mut1nyJD, paper is https://arxiv.org/abs/2204.00227 v0.18.0 Phil Wang 2022-06-09 08:25:05 -07:00
• 96bb2ff310 alpha cosine noise schedule is now working for continuous time gaussian diffusion v0.17.7 Phil Wang 2022-06-08 23:07:01 -07:00
• 582bfe275b successfully did some basic math and clipped the predicted x0 intermediate for the continuous time case v0.17.6 Phil Wang 2022-06-08 17:59:41 -07:00
• 339fc024ba successfully did some basic math and clipped the predicted x0 intermediate for the continuous time case v0.17.5 Phil Wang 2022-06-08 17:53:26 -07:00