Commits · 5058cc446b970aa63b54193d80127c8d6f235e8c · school / Capstone Design / 01 / AudioLM

Browse files

This project is mirrored from https://github.com/lucidrains/audiolm-pytorch.git. Pull mirroring failed May 17, 2023.
Repository mirroring has been paused due to too many failed attempts. It can be resumed by a project maintainer or owner.
Last successful update May 15, 2023.

Jan 12, 2023
- address two bugs, thanks to @apoorv2904 in issue... · 5058cc44
  Phil Wang authored Jan 11, 2023
  View commits for tag 0.3.6 0.3.6
  
  5058cc44
- need to access unwrapped soundstream in trainer · 4c3c3534
  Phil Wang authored Jan 11, 2023
  
  View commits for tag 0.3.5 0.3.5
  
  4c3c3534
Jan 11, 2023
- 0.3.4 · 36dc9e0d
  Phil Wang authored Jan 11, 2023
  
  View commits for tag 0.3.4 0.3.4
  
  36dc9e0d
- Merge pull request #46 from djqualia/patch-9 · 6ea8cf01
  Phil Wang authored Jan 11, 2023
  6ea8cf01
- Fix logging of recon_loss and small typo · b181beac
  djqualia authored Jan 10, 2023
  b181beac
- 0.3.3 · 8b70431f
  Phil Wang authored Jan 10, 2023
  
  View commits for tag 0.3.3 0.3.3
  
  8b70431f
- Merge pull request #44 from djqualia/patch-6 · c1d008c6
  Phil Wang authored Jan 10, 2023
  c1d008c6
- Add Accelerate init_tracker and log losses · c63b1362
  djqualia authored Jan 10, 2023
  c63b1362
- fix stft discriminator · 920a8a46
  Phil Wang authored Jan 10, 2023
  
  View commits for tag 0.3.2 0.3.2
  
  920a8a46
- fix stft discriminator loss, after fixing the distributed issue · c5ea4618
  Phil Wang authored Jan 10, 2023
  
  View commits for tag 0.3.1 0.3.1
  
  c5ea4618
Jan 10, 2023
- just do what encodec does by concatting real and imag after the stft transform... · 8259b0d0
  Phil Wang authored Jan 09, 2023
  View commits for tag 0.3.0 0.3.0
  
  8259b0d0
Jan 07, 2023
- update · 86b5e4a0
  Phil Wang authored Jan 06, 2023
  
  86b5e4a0
Jan 05, 2023
- fix issue with accelerator wrapped models when saving and loading · bcf77f91
  Phil Wang authored Jan 04, 2023
  
  View commits for tag 0.2.3 0.2.3
  
  bcf77f91
- actually save optim state for soundstream · e4d31ce4
  Phil Wang authored Jan 04, 2023
  
  View commits for tag 0.2.2 0.2.2
  
  e4d31ce4
- add helper save load methods to all the trainers, for saving optimizer states,... · 89a35002
  Phil Wang authored Jan 04, 2023
  View commits for tag 0.2.1 0.2.1
  
  89a35002
Dec 28, 2022
- bring the feedforward design from localvit to audiolm · deceb7a0
  Phil Wang authored Dec 27, 2022
  
  View commits for tag 0.2.0 0.2.0
  
  deceb7a0
Dec 16, 2022
- 0.1.22 · 94fe05d1
  Phil Wang authored Dec 15, 2022
  
  View commits for tag 0.1.22 0.1.22
  
  94fe05d1
- Merge pull request #37 from djqualia/patch-3 · 6178b7d6
  Phil Wang authored Dec 15, 2022
  6178b7d6
- Make the training sample output filename unique · 7f3b2d68
  djqualia authored Dec 15, 2022
  7f3b2d68
- 0.1.21 · d4e278b6
  Phil Wang authored Dec 15, 2022
  
  View commits for tag 0.1.21 0.1.21
  
  d4e278b6
- Merge pull request #34 from djqualia/patch-2 · b6c24794
  Phil Wang authored Dec 15, 2022
  b6c24794
Dec 15, 2022
- fix soundstream training · b0384e83
  Phil Wang authored Dec 14, 2022
  
  View commits for tag 0.1.20 0.1.20
  
  b0384e83
- fix bug with collation · 8ebbb2e9
  Phil Wang authored Dec 14, 2022
  
  View commits for tag 0.1.19 0.1.19
  
  8ebbb2e9
- Pad right side of audio vs. left · 4b893421
  djqualia authored Dec 14, 2022
  4b893421
Dec 14, 2022
- not all the transformer may need conditioning, perhaps only semantic is... · 60e4cd84
  Phil Wang authored Dec 13, 2022
  View commits for tag 0.1.18 0.1.18
  
  60e4cd84
- product management · 2a785158
  Phil Wang authored Dec 13, 2022
  2a785158
- get a working example of text conditioned semantic transformer training · ca13fbae
  Phil Wang authored Dec 13, 2022
  
  View commits for tag 0.1.17 0.1.17
  
  ca13fbae
Dec 13, 2022
- proper credit assignment · 228a2298
  Phil Wang authored Dec 12, 2022
  
  228a2298
- switch to modrelu, seems to be recommended by a lot of torch complex packages · 4d081c7a
  Phil Wang authored Dec 12, 2022
  
  4d081c7a
Dec 12, 2022
- allow fine transformer to be trained on pre-encoded soundstream codebook indices · aa053305
  Phil Wang authored Dec 11, 2022
  
  View commits for tag 0.1.15 0.1.15
  
  aa053305
Dec 10, 2022
- set things up for eventual text conditioned audio synthesis · 8ad550a5
  Phil Wang authored Dec 09, 2022
  
  View commits for tag 0.1.14 0.1.14
  
  8ad550a5
Dec 09, 2022
- expose the ema decay for the residual vq in soundstream · a1cf75ca
  Phil Wang authored Dec 08, 2022
  
  View commits for tag 0.1.12 0.1.12
  
  a1cf75ca
- 0.1.11 · 03b21607
  Phil Wang authored Dec 08, 2022
  
  View commits for tag 0.1.11 0.1.11
  
  03b21607
- add local attention at the innermost layer of soundstream, flanking the... · b1f4c6fd
  Phil Wang authored Dec 08, 2022
  b1f4c6fd
Dec 08, 2022
- 0.1.10 · 1a6cfc30
  Phil Wang authored Dec 07, 2022
  
  View commits for tag 0.1.10 0.1.10
  
  1a6cfc30
- Merge pull request #30 from erogol/stft_fix · b0e10403
  Phil Wang authored Dec 07, 2022
  b0e10403
- Fix stft · 554ffa16
  Eren Gölge authored Dec 08, 2022
  
  554ffa16
- some tweaks to soundstream, include commitment loss · d0151be9
  Phil Wang authored Dec 07, 2022
  
  View commits for tag 0.1.9 0.1.9
  
  d0151be9
- make sure dropout on post-softmax attention is present · 894d8925
  Phil Wang authored Dec 07, 2022
  
  894d8925
- validate transformers being passed into audiolm · 98723add
  Phil Wang authored Dec 07, 2022
  
  View commits for tag 0.1.7 0.1.7
  
  98723add