Commits · aac8fc99d6a06e9e3b4c0689c1fff3d379dd0672 · chenpangpang / ComfyUI

28 Oct, 2023 1 commit
- Fix checkpoint loader with config. · 2a134bfa
  comfyanonymous authored Oct 27, 2023
  
  2a134bfa
27 Oct, 2023 3 commits
- SD1 and SD2 clip and tokenizer code is now more similar to the SDXL one. · e60ca692
  comfyanonymous authored Oct 27, 2023
  
  e60ca692
- Support SSD1B model and make it easier to support asymmetric unets. · 6ec3f12c
  comfyanonymous authored Oct 27, 2023
  
  6ec3f12c
- Restrict loading embeddings from embedding folders. · 434ce25e
  comfyanonymous authored Oct 27, 2023
  
  434ce25e
26 Oct, 2023 2 commits
- Faster clip image processing. · 723847f6
  comfyanonymous authored Oct 26, 2023
  
  723847f6
- Fix some OOM issues with split and sub quad attention. · a373367b
  comfyanonymous authored Oct 25, 2023
  
  a373367b
25 Oct, 2023 4 commits
- Fix uni_pc returning noisy image when steps <= 3 · 7fbb217d
  comfyanonymous authored Oct 25, 2023
  
  7fbb217d
- change 'c_adm' to 'y' in ControlNet.get_control · 3783cb8b
  Jedrzej Kosinski authored Oct 25, 2023
  
  3783cb8b
- Pass extra conds directly to unet. · d1d2fea8
  comfyanonymous authored Oct 25, 2023
  
  d1d2fea8
- Refactor to make it easier to add custom conds to models. · 036f88c6
  comfyanonymous authored Oct 24, 2023
  
  036f88c6
24 Oct, 2023 1 commit
- Sampling code refactor to make it easier to add more conds. · 3fce8881
  comfyanonymous authored Oct 24, 2023
  
  3fce8881
22 Oct, 2023 4 commits
- Empty the cache when torch cache is more than 25% free mem. · 8594c8be
  comfyanonymous authored Oct 22, 2023
  
  8594c8be
- attention_basic now works with hypertile. · 8b65f5de
  comfyanonymous authored Oct 22, 2023
  
  8b65f5de
- Make sub_quad and split work with hypertile. · e6bc42df
  comfyanonymous authored Oct 22, 2023
  
  e6bc42df
- Fix t2i adapter issue. · a0690f9d
  comfyanonymous authored Oct 21, 2023
  
  a0690f9d
21 Oct, 2023 1 commit
- Make xformers work with hypertile. · 9906e3ef
  comfyanonymous authored Oct 21, 2023
  
  9906e3ef
20 Oct, 2023 1 commit
- Fix uni_pc sampler math. This changes the images this sampler produces. · 4185324a
  comfyanonymous authored Oct 20, 2023
  
  4185324a
19 Oct, 2023 2 commits
- Make sure cond_concat is on the right device. · e6962120
  comfyanonymous authored Oct 19, 2023
  
  e6962120
- Refactor cond_concat into conditioning. · 45c972ab
  comfyanonymous authored Oct 18, 2023
  
  45c972ab
18 Oct, 2023 3 commits
- Fix some potential issues. · 430a8334
  comfyanonymous authored Oct 18, 2023
  
  430a8334
- Refactor cond_concat into model object. · 782a24fc
  comfyanonymous authored Oct 18, 2023
  
  782a24fc
- Fix memory issue related to control loras. · 0d45a565
  comfyanonymous authored Oct 18, 2023
```
The cleanup function was not getting called.
```
  0d45a565
17 Oct, 2023 2 commits
- Make VAE code closer to sgm. · d44a2de4
  comfyanonymous authored Oct 17, 2023
  
  d44a2de4
- Refactor the attention stuff in the VAE. · 23680a91
  comfyanonymous authored Oct 17, 2023
  
  23680a91
16 Oct, 2023 2 commits
- Add some Quadro cards to the list of cards with broken fp16. · c8013f73
  comfyanonymous authored Oct 16, 2023
  
  c8013f73
- Add a separate optimized_attention_masked function. · bb064c97
  comfyanonymous authored Oct 16, 2023
  
  bb064c97
13 Oct, 2023 2 commits
- Add a --bf16-unet to test running the unet in bf16. · fd4c5f07
  comfyanonymous authored Oct 13, 2023
  
  fd4c5f07
- Refactor code so model can be a dtype other than fp32 or fp16. · 9a55dadb
  comfyanonymous authored Oct 13, 2023
  
  9a55dadb
12 Oct, 2023 4 commits
- pytorch_attention_enabled can now return True when xformers is enabled. · 88733c99
  comfyanonymous authored Oct 11, 2023
  
  88733c99
- Pull some small changes from the other repo. · 20d3852a
  comfyanonymous authored Oct 11, 2023
  
  20d3852a
- Allow attn_mask in attention_pytorch. · ac7d8cfa
  comfyanonymous authored Oct 11, 2023
  
  ac7d8cfa
- Refactor the attention functions. · 1a4bd9e9
  comfyanonymous authored Oct 11, 2023
```
There's no reason for the whole CrossAttention object to be repeated when
only the operation in the middle changes.
```
  1a4bd9e9
11 Oct, 2023 2 commits
- Let unet wrapper functions have .to attributes. · 8cc75c64
  comfyanonymous authored Oct 11, 2023
  
  8cc75c64
- Cleanup. · 5e885bd9
  comfyanonymous authored Oct 10, 2023
  
  5e885bd9
10 Oct, 2023 1 commit
- Supports TAESD models in safetensors format · 9eb621c9
  Yukimasa Funaoka authored Oct 10, 2023
  
  9eb621c9
06 Oct, 2023 1 commit
- load_checkpoint_guess_config can now optionally output the model. · 72188dff
  comfyanonymous authored Oct 06, 2023
  
  72188dff
04 Oct, 2023 2 commits
- Option to input directory · 63e5fd17
  Jairo Correa authored Oct 04, 2023
  
  63e5fd17
- Fix quality loss due to low precision · 9bfec2bd
  City authored Oct 04, 2023
  
  9bfec2bd
29 Sep, 2023 2 commits
- fix: typo in extra sampler · 0f17993d
  badayvedat authored Sep 29, 2023
  
  0f17993d
- Add SamplerDPMPP_2M_SDE node. · 66756de1
  comfyanonymous authored Sep 28, 2023
  
  66756de1