Commit Graph
98 Commits
Author SHA1 Message Date
City 4a11543beb Add switch for new xformers
#86 - patch provided by @SaiZyca
2024-12-11 22:48:51 +01:00
City ee982c5d73 Remove timm as a dependency 2024-12-11 22:44:09 +01:00
City ca2bc8bc65 Make mlp args match the one from timm 2024-12-11 22:37:13 +01:00
City d1f751677a Patch size logic from common dit 2024-12-11 22:34:31 +01:00
City 2beff5a4aa Faster model initialization
don't initialize base Sana model for SanaMS
2024-12-11 22:33:57 +01:00
City 7e19ac5e43 Use native ops for sana 2024-12-11 22:01:11 +01:00
City 5505ab4f40 Sana loader logic 2024-12-11 18:55:17 +01:00
City 06a936813f Consolidate loader logic 2024-12-11 17:32:57 +01:00
City 6a88140002 Unused 2024-12-11 17:24:18 +01:00
City d5a47fa3e5 Use native ops for PixArt 2024-12-11 00:52:48 +01:00
City 45423d9673 Remove HyDiT - supported in main 2024-12-10 22:24:37 +01:00
City 3f844cac42 PixArt initial rewrite 2024-12-10 22:19:02 +01:00
City 7dc1403cd5 Update README.md 2024-12-09 23:38:52 +01:00
City 78e10dab45 Update README.md
#84
2024-12-05 19:19:37 +01:00
City e9cc31c9cb Add empty latent node for Sana
Also remove unused display name mappings
2024-12-05 18:45:54 +01:00
City ef5a322efe Merge pull request #84 from lawrence-cj/main
Add Sana as an ExtraModel
2024-12-05 18:37:45 +01:00
City c6527a45f1 Set t5_attention_mask for new comfy versions
#79
2024-10-24 21:00:57 +02:00
City 193c2fd9d3 Remove unused arg
#80
2024-08-07 13:42:31 +02:00
City 7e96bf676c Add force offload node 2024-08-02 15:27:46 +02:00
City 6fb8649168 PixArt: fix OneTrainer LoRA 2024-07-27 17:52:02 +02:00
City d89de6e98e Add protobuf to requirements
#74
2024-07-17 00:25:46 +02:00
City 1db046de78 PixArt: Update import for SD3 clip
#76
2024-07-17 00:22:24 +02:00
City 40d2a76310 HunYuanDiT: Add 1.2 support 2024-07-04 18:01:53 +02:00
City 62b84531f4 Add ostris/vae-kl-f8-d16
https://huggingface.co/ostris/vae-kl-f8-d16
2024-07-02 23:02:31 +02:00
City faf6979f34 Merge pull request #65 from GavChap/get-depth-programatically
Get depth programatically
2024-06-23 22:13:59 +02:00
City de52d3aa45 PixArt auto checkpoint loader
Needs better logic for diffusers weights
2024-06-19 03:03:43 +02:00
City 2ad0b09cb9 PixArt allow on-the-fly PE interpolation 2024-06-19 02:26:01 +02:00
City 5101719ca8 HunYuan config fix 2024-06-16 17:07:18 +02:00
City 9a4629aba7 PixArt SD3 node prompt weight fix
#61
2024-06-16 02:51:21 +02:00
City c174662c5b Allow PixArt to use SD3 clip 2024-06-14 16:32:05 +02:00
City 9da9978f6c PixArt LoRA changes
#60
2024-06-12 01:04:22 +02:00
City 0f83907911 HunYuanDiT fix double text encoder input
#58
2024-05-31 21:56:37 +02:00
City 58ecf66b7a Update README.md 2024-05-29 00:09:58 +02:00
City 7e2db6ee78 HunYuanDiT fix image size cond logic (again)
#54
2024-05-28 12:19:38 +02:00
City b773f348e6 HunYuanDiT fix image size cond logic
#54
2024-05-28 00:18:44 +02:00
City 014e067483 Update README.md 2024-05-25 18:38:43 +02:00
City 6867aacb26 PixArt fix float64 conversion error
#48
2024-05-25 18:09:56 +02:00
City 4946b5038e HunYuanDiT switch to local tokenizer 2024-05-25 17:50:31 +02:00
City 904e5bcacf HunYuanDiT Size cond wrong 2024-05-25 17:44:28 +02:00
City e5c15ba479 HunYuanDiT Fix text encoder device select
Fixes #47 and #50
2024-05-25 16:44:58 +02:00
City 1953a3bc99 Update README.md 2024-05-21 23:48:31 +02:00
City 2021226406 Merge pull request #41 from zmwv823/main
Add MiaoBi
2024-05-21 23:44:41 +02:00
City 13dc46fffc HunYuanDiT missing import 2024-05-21 13:59:42 +02:00
City ea705d06e8 Update README.md 2024-05-20 11:52:48 +02:00
City bad1b65c09 HunYuanDiT Initial proof of concept
#35
2024-05-20 11:41:58 +02:00
City 296411cc93 Merge pull request #36 from GavChap/fix-sdp-attention-mask
* Fix SDP attention mask by creating a diagonal block mask in the right shape.
2024-05-16 15:46:31 +02:00
City 1e5dc8e10d Merge pull request #33 from GavChap/sdp-attn-fix
Update PixArt_blocks.py
2024-05-12 22:07:52 +02:00
City c2e2f6e810 Update README.md 2024-05-08 13:21:07 +02:00
City 77d92abf1c Allow changing T5 in extra_model_paths 2024-05-06 16:36:38 +02:00
City 47758b848b Update requirements.txt 2024-04-29 15:45:01 +02:00
City bb4b4401d5 Update README.md 2024-04-26 22:42:49 +02:00
City d52d732677 PixArt 2K config
#23
2024-04-23 18:31:40 +02:00
City 2f2797e502 PixArt Allow missing state dict keys
Fixes #21
2024-04-21 21:12:15 +02:00
City f9e1178de7 T5 offload w/ bnb
Mentioned in #22
2024-04-21 20:41:25 +02:00
City 92323c62e7 T5 second GPU offload 2024-04-10 01:54:08 +02:00
City 5e3886c094 PixArt Sigma 1024 2024-04-09 23:48:34 +02:00
City c325dec0d3 PixArt sigma
Mentioned in #19
Also removes old example sampler since updating it would be too much overhead.
Pytorch attention is still wonky.
2024-04-07 17:55:33 +02:00
City 059c87ebb5 Add movq3 VAE 2024-04-06 22:13:32 +02:00
City 18fb3f5674 Quick fix for new comfy version 2024-04-06 12:05:04 +02:00
City 6f80adfce5 Fix for deprecated op 2024-03-16 18:18:56 +01:00
City 8befcdd9d6 PixArt controlnet test 2024-02-28 19:05:08 +01:00
City d2488d73f6 Quick FP8 fix. 2024-01-28 17:47:02 +01:00
City 23e59bbffe PixArt LoRA support
Supports models from the example training code.
Weight loading will probably have to be changed for native formats.
2023-12-22 16:16:14 +01:00
City 451c7588c7 dtype cleanup 2023-12-22 15:54:16 +01:00
City 1a968d7bc8 Add LCM instructions 2023-12-15 22:02:21 +01:00
City a457da3f12 Add support for PixArt diffusers weights 2023-12-15 20:43:32 +01:00
City 3782b16606 Re add BF16 for VAE 2023-12-15 00:33:25 +01:00
City 169a9aea03 Update pip command
Add `-s` flag to python command as per issue #5
2023-12-12 07:14:00 +01:00
City b261c66f29 Fix DiT sampling
Apply the PixArt fix here as well.

Move models to checkpoints folder as that makes more sense in this case.
2023-12-08 18:40:25 +01:00
City 80d5d9299b Temp fix for mask shape 2023-12-08 17:23:51 +01:00
City a634b1ea09 Update README.md 2023-12-08 16:47:09 +01:00
City af62f98b4f PixArt KSampler support
Thanks to @lawrence-cj for his help on this!
2023-12-08 16:25:01 +01:00
City b31d8c517b PixArt small fix
Issue #3 + some code changes from upstream.
2023-12-06 21:31:07 +01:00
City 389c16f2f5 Add dtype selection to T5 2023-11-28 00:49:06 +01:00
City 33e5db52a3 Never used 2023-11-22 20:58:21 +01:00
City 4c711dddc4 Fix decoder init 2023-11-22 20:46:34 +01:00
City 06fdd1f2d6 SDV VideoDecoder 2023-11-22 20:20:32 +01:00
City a3e1804def Consistency decoder full
With model definitions from https://gist.github.com/madebyollin/865fa6a18d9099351ddbdfbe7299ccbf
2023-11-07 21:26:54 +01:00
City 4fb1292c50 Update README.md 2023-11-06 23:25:49 +01:00
City b7a293cc27 Consistency decoder initial
Quick and lazy, just ported the example code. Can't do much more without having the model arch.

https://github.com/openai/consistencydecoder/issues/1
2023-11-06 23:19:42 +01:00
City 5a27b5b1b8 Update README.md 2023-11-06 18:43:26 +01:00
City aee031b692 Fix DiT forward pass cond 2023-10-29 12:51:20 +01:00
City 1d5f745bbf Merge pull request #1 from deinferno/main
Fix PixArtT5TextEncode error
2023-10-29 12:24:12 +01:00
City 67ab30aa18 Instructions for ComfyUI standalone. 2023-10-28 04:19:35 +02:00
City 5b42a090c4 xformers 2023-10-28 02:58:13 +02:00
City 7e067d2ac5 Requirements for T5 2023-10-28 00:35:45 +02:00
City 7a37319826 Formatting / order
Models should be at the top
2023-10-28 00:05:55 +02:00
City dabf25a076 Update README.md 2023-10-27 21:47:35 +02:00
City 814fe50852 PixArt support 2023-10-27 21:42:00 +02:00
City 519da29c81 T5v1.1 support 2023-10-27 21:37:10 +02:00
City 6f0c07f247 Update README.md 2023-10-21 22:33:56 +02:00
City dcae5c25b0 VQModel VAE 2023-10-21 22:22:28 +02:00
City 1c72bd4ae7 Custom VAE 2023-10-21 21:26:23 +02:00
City bca8487272 Update README.md 2023-10-21 02:35:18 +02:00
City 91e2405932 Update README.md 2023-10-21 01:45:23 +02:00
City 72a178ed63 Update README.md 2023-10-21 00:04:26 +02:00
City bad33b51ab DiT Support 2023-10-20 23:48:22 +02:00
City 2c5a0e0fe3 Initial commit 2023-10-20 23:19:45 +02:00