Commits · a13c0ce526d1486238e398bcbae741010b402c1f · OpenDAS / OpenFold

10 Oct, 2023 3 commits
- Renamed `preembedding_embedder` to `input_embedder` · 9b114f28
  sachinkadyan7 authored Sep 11, 2023
  
  9b114f28
- Added documentation for some sequence embedding model changes. · 29962990
  Sachin Kadyan authored Oct 28, 2022
  
  29962990
- Added switch for using the single sequence embedder when using the model in `seqemb` mode. · 75889e9a
  Sachin Kadyan authored Oct 18, 2022
```
- `seqemb_mode_enabled` added as a configuration option.
- `model.py` switches to using the `PreembeddingEmbedder` when the flag is `True`.
```
  75889e9a
05 Oct, 2023 1 commit
- Added kernel to template pair stack and updated tests · a3de9cb9
  Christina Floristean authored Oct 05, 2023
  
  a3de9cb9
07 Sep, 2023 1 commit
- Integrated deepspeed attention kernel and added initial tests. · f0a320e0
  Christina Floristean authored Sep 07, 2023
  
  f0a320e0
19 Aug, 2022 2 commits
- Fix new template bug · 349fdbd9
  Gustaf Ahdritz authored Aug 19, 2022
  
  349fdbd9
- Squeeze templates · 29864369
  Gustaf Ahdritz authored Aug 19, 2022
  
  29864369
17 Aug, 2022 1 commit
- Fix template masking bug · e56b5976
  Gustaf Ahdritz authored Aug 17, 2022
  
  e56b5976
08 Jul, 2022 1 commit
- Add tracing, TensorFloat32 utilization, and FlashAttention · 4f53624d
  Gustaf Ahdritz authored Jul 08, 2022
  
  4f53624d
03 Jul, 2022 1 commit
- Add conditions to model constructor · 4c37c386
  Gustaf Ahdritz authored Jul 03, 2022
  
  4c37c386
21 Jun, 2022 2 commits
- Refactor inplace operations, fix training · 143ba486
  Gustaf Ahdritz authored Jun 21, 2022
  
  143ba486
- Move chunking code, add memory optimizations · 5aa21eae
  Gustaf Ahdritz authored Jun 21, 2022
  
  5aa21eae
14 Jun, 2022 1 commit
- Remove removed function call · ceb95303
  Gustaf Ahdritz authored Jun 14, 2022
  
  ceb95303
12 Jun, 2022 2 commits
- Remove redundant checkpoint functions · 8036a213
  Gustaf Ahdritz authored Jun 11, 2022
  
  8036a213
- Speed up inference · bebbe782
  Gustaf Ahdritz authored Jun 11, 2022
  
  bebbe782
10 Jun, 2022 1 commit
- Vastly lower peak inference memory usage · 6e66b218
  Gustaf Ahdritz authored Jun 10, 2022
  
  6e66b218
13 May, 2022 1 commit
- Improve ease of use of LMA · 722a5e01
  Gustaf Ahdritz authored May 12, 2022
  
  722a5e01
19 Apr, 2022 1 commit
- Fix template bugs · 591d10d2
  Gustaf Ahdritz authored Apr 19, 2022
  
  591d10d2
13 Apr, 2022 1 commit
- Remove exception for num_recycle=0 · 9a8910f9
  Gustaf Ahdritz authored Apr 13, 2022
  
  9a8910f9
03 Feb, 2022 2 commits
- DeepSpeed + PL bfloat16 working · 1df4991d
  Gustaf Ahdritz authored Jan 13, 2022
  
  1df4991d
- Fix mask casting · 53bb9c10
  Gustaf Ahdritz authored Jan 10, 2022
  
  53bb9c10
06 Jan, 2022 1 commit
- Add chunking experiment · df89bb28
  Gustaf Ahdritz authored Jan 05, 2022
  
  df89bb28
30 Dec, 2021 1 commit
- Slightly refactor recycling code to improve FP16 stability · bb6f6145
  Gustaf Ahdritz authored Dec 30, 2021
  
  bb6f6145
28 Dec, 2021 1 commit
- Fix typo in model.py · e1142cf3
  Gustaf Ahdritz authored Dec 28, 2021
  
  e1142cf3
18 Nov, 2021 1 commit
- Free used recycling tensors · 326e08de
  Gustaf Ahdritz authored Nov 18, 2021
  
  326e08de
28 Oct, 2021 2 commits
- Make template code use less memory during inference · 2faff451
  Gustaf Ahdritz authored Oct 28, 2021
  
  2faff451
- Fix empty template feature bug · d71d37ff
  Gustaf Ahdritz authored Oct 27, 2021
  
  d71d37ff
21 Oct, 2021 2 commits
- Fix typo · 13290527
  Gustaf Ahdritz authored Oct 20, 2021
  
  13290527
- Remove dependency on training-time feature · bf3eed6e
  Gustaf Ahdritz authored Oct 20, 2021
  
  bf3eed6e
20 Oct, 2021 2 commits
- Move template pair stack out of single-template block (for DeepSpeed) · aea444ed
  Gustaf Ahdritz authored Oct 20, 2021
  
  aea444ed
- Bring tests up to speed · 49767099
  Gustaf Ahdritz authored Oct 19, 2021
  
  49767099
18 Oct, 2021 1 commit
- Make chunk size more flexible, reduce verbosity · b026de28
  Gustaf Ahdritz authored Oct 17, 2021
  
  b026de28
16 Oct, 2021 1 commit
- Standardize code style · 07e64267
  Gustaf Ahdritz authored Oct 16, 2021
  
  07e64267
15 Oct, 2021 1 commit
- Add training parsers · d48c052c
  Gustaf Ahdritz authored Oct 15, 2021
  
  d48c052c
05 Oct, 2021 1 commit
- Get FP16 training working · 893fe372
  Gustaf Ahdritz authored Oct 05, 2021
  
  893fe372
02 Oct, 2021 1 commit
- Sidestep AMP bug · f34bef8e
  Gustaf Ahdritz authored Oct 02, 2021
  
  f34bef8e
30 Sep, 2021 2 commits
- Add pTM config options, spruce up pretrained script, add test script · 54317fe4
  Gustaf Ahdritz authored Sep 30, 2021
  
  54317fe4
- Finish purge of in-place ops, get grads working, add TM · eb49136d
  Gustaf Ahdritz authored Sep 30, 2021
  
  eb49136d
28 Sep, 2021 1 commit
- Finish accommodating FP16. FAPE does not decrease · 1d47c1e7
  Gustaf Ahdritz authored Sep 28, 2021
  
  1d47c1e7
27 Sep, 2021 1 commit
- Fix tensor casting, improve TorchScript compatibility · de0fa7b1
  Gustaf Ahdritz authored Sep 27, 2021
  
  de0fa7b1