Commits · 83d969e3975d340ef980db59b07954a67d08ce6f · chenpangpang / ComfyUI

21 May, 2024 1 commit
- Disable xformers when tracing model. · 83d969e3
  comfyanonymous authored May 21, 2024
  
  83d969e3
20 May, 2024 1 commit
- Fix potential issue. · 1900e511
  comfyanonymous authored May 20, 2024
  
  1900e511
18 May, 2024 2 commits
- Cleanup. · 0bdc2b15
  comfyanonymous authored May 18, 2024
  
  0bdc2b15
- Remove unnecessary code. · 98f828fa
  comfyanonymous authored May 18, 2024
  
  98f828fa
16 May, 2024 1 commit
- Add debug options to force on and off attention upcasting. · 46daf0a9
  comfyanonymous authored May 16, 2024
  
  46daf0a9
14 May, 2024 3 commits
- Fix SAG. · ec6f16ad
  comfyanonymous authored May 14, 2024
  
  ec6f16ad
- Only enable attention upcasting on models that actually need it. · bb4940d8
  comfyanonymous authored May 14, 2024
  
  bb4940d8
- Refactor attention upcasting code part 1. · b0ab31d0
  comfyanonymous authored May 14, 2024
  
  b0ab31d0
01 May, 2024 1 commit
- Workaround xformers bug. · 2aed53c4
  comfyanonymous authored Apr 30, 2024
  
  2aed53c4
16 Mar, 2024 1 commit
- Move cascade scale factor from stage_a to latent_formats.py · d7897fff
  comfyanonymous authored Mar 16, 2024
  
  d7897fff
11 Mar, 2024 1 commit
- Switch some more prints to logging. · 2a813c3b
  comfyanonymous authored Mar 11, 2024
  
  2a813c3b
07 Mar, 2024 1 commit
- Support loading the sr cascade controlnet. · 5f60ee24
  comfyanonymous authored Mar 07, 2024
  
  5f60ee24
06 Mar, 2024 2 commits
- Set upscale algorithm to bilinear for stable cascade controlnet. · 03e6e816
  comfyanonymous authored Mar 06, 2024
  
  03e6e816
- Support stable cascade canny controlnet. · 03e83bb5
  comfyanonymous authored Mar 06, 2024
  
  03e83bb5
29 Feb, 2024 1 commit
- Allow image_only_indicator to be None. · cb7c3a29
  comfyanonymous authored Feb 29, 2024
  
  cb7c3a29
28 Feb, 2024 1 commit
- Koala 700M and 1B support. · b3e97fc7
  comfyanonymous authored Feb 28, 2024
```
Use the UNET Loader node to load the unet file to use them.
```
  b3e97fc7
19 Feb, 2024 2 commits
- Remove print. · e93cdd0a
  comfyanonymous authored Feb 19, 2024
  
  e93cdd0a
- Forgot to commit this. · a7b5eaa7
  comfyanonymous authored Feb 19, 2024
  
  a7b5eaa7
17 Feb, 2024 5 commits
- Fix attention masks properly for multiple batches. · 6bcf57ff
  comfyanonymous authored Feb 17, 2024
  
  6bcf57ff
- fp8 weight support for Stable Cascade. · 11e3221f
  comfyanonymous authored Feb 17, 2024
  
  11e3221f
- Fix attention mask batch size in some attention functions. · f8706546
  comfyanonymous authored Feb 17, 2024
  
  f8706546
- Properly fix attention masks in CLIP with batches. · 3b9969c1
  comfyanonymous authored Feb 17, 2024
  
  3b9969c1
- Make Stable Cascade work on old pytorch 2.0 · 805c36ac
  comfyanonymous authored Feb 17, 2024
  
  805c36ac
16 Feb, 2024 3 commits
- Stable Cascade Stage B. · 667c9281
  comfyanonymous authored Feb 16, 2024
  
  667c9281
- Stable Cascade Stage C. · f83109f0
  comfyanonymous authored Feb 16, 2024
  
  f83109f0
- Stable Cascade Stage A. · 5e06baf1
  comfyanonymous authored Feb 16, 2024
  
  5e06baf1
07 Feb, 2024 1 commit
- Don't use numpy for calculating sigmas. · c661a8b1
  comfyanonymous authored Feb 07, 2024
  
  c661a8b1
26 Jan, 2024 1 commit
- Remove some unused imports. · 89507f8a
  comfyanonymous authored Jan 25, 2024
  
  89507f8a
14 Jan, 2024 1 commit
- Make unclip more deterministic. · 2395ae74
  comfyanonymous authored Jan 14, 2024
```
Pass a seed argument note that this might make old unclip images different.
```
  2395ae74
09 Jan, 2024 1 commit
- Use basic attention implementation for small inputs on old pytorch. · 6a7bc35d
  comfyanonymous authored Jan 09, 2024
  
  6a7bc35d
07 Jan, 2024 2 commits
- Update optimized_attention_for_device function for new functions that · c6951548
  comfyanonymous authored Jan 07, 2024
```
support masked attention.
```
  c6951548
- Add attention mask support to sub quad attention. · aaa90173
  comfyanonymous authored Jan 07, 2024
  
  aaa90173
06 Jan, 2024 2 commits
- Support attention mask in split attention. · 0c2c9fbd
  comfyanonymous authored Jan 06, 2024
  
  0c2c9fbd
- Implement attention mask on xformers. · 3ad0191b
  comfyanonymous authored Jan 06, 2024
  
  3ad0191b
03 Jan, 2024 1 commit
- Implement noise augmentation for SD 4X upscale model. · 8c649357
  comfyanonymous authored Jan 03, 2024
  
  8c649357
02 Jan, 2024 1 commit
- Remove useless code. · 79f73a4b
  comfyanonymous authored Jan 02, 2024
  
  79f73a4b
26 Dec, 2023 1 commit
- Fix lowvram mode not working with unCLIP and Revision code. · 61b3f15f
  comfyanonymous authored Dec 26, 2023
  
  61b3f15f
24 Dec, 2023 1 commit
- Fix SVD lowvram mode. · d0165d81
  comfyanonymous authored Dec 24, 2023
  
  d0165d81
22 Dec, 2023 1 commit
- A few missing comfy ops in the VAE. · 261bcbb0
  comfyanonymous authored Dec 22, 2023
  
  261bcbb0
15 Dec, 2023 1 commit
- Remove useless code. · a5056cfb
  comfyanonymous authored Dec 15, 2023
  
  a5056cfb