Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Sep 4, 2021

• [ADD] comment about Keras return sequences = False, also we don't have stacked LSTMs in our pytorch version :thinking_face:

  ruohoruotsi

  committed
  2512e52

  Copy full SHA for 2512e52

Commits on Sep 3, 2021

• [FIX] name, since LSTM class, has a linear output layer and softmax activation

  ruohoruotsi

  committed
  3340255

  Copy full SHA for 3340255

Commits on Aug 24, 2021

• [ADD] minor change to specify that partition is a dataset partition, perhaps that's not obvious

  ruohoruotsi

  committed
  9f7e5b0

  Copy full SHA for 9f7e5b0

Commits on Aug 6, 2021

• [ADD] fixes to get validation loss to display in the progress bar per n steps. Shuffle dataloading batches

  ruohoruotsi

  committed
  40793a2

  Copy full SHA for 40793a2

• [CLEANUP] logging, attempts to debug validation progress bar issues & warnings, as well as proper per-step, per-epoch reporting

  ruohoruotsi

  committed
  e46e129

  Copy full SHA for e46e129

• [ADD] changes for first working training pytorch-lighting effort, :hands-raised:

  ruohoruotsi

  committed
  a5a24fb

  Copy full SHA for a5a24fb

Commits on Aug 5, 2021

• [FIX] but in choosing dev partition, tested by enumerating overall train and test batches. Looks good

  ruohoruotsi

  committed
  610ad55

  Copy full SHA for 610ad55

• [ADD] checkpoint for working GTZANDataset and use in a batching DataLoader. Nextup, the DataModule?

  ruohoruotsi

  committed
  c4e6f4e

  Copy full SHA for c4e6f4e

Commits on Aug 3, 2021

• [WIP] checkpoint, need to get DataModule working, unsure where the current input,target permutation goes. Need a good non-MNIST DataModule example :thinking:

  ruohoruotsi

  committed
  f520be3

  Copy full SHA for f520be3

• [ADD] a main function and move over data loading and training loop

  ruohoruotsi

  committed
  f871f54

  Copy full SHA for f871f54

• [ADD] a few cleanups to logging to stdout, as well as hidden -> hidden_state, to not have interscope shadowing

  ruohoruotsi

  committed
  c32344e

  Copy full SHA for c32344e

• [ADD] matplotlib req and cleanup continue_training bool variable to stateful for better readability. Ditto in WIP PTLighting version

  ruohoruotsi

  committed
  8fe3443

  Copy full SHA for 8fe3443

Commits on Jul 30, 2021

• [WIP] checkpoint for pt-lightning port

  ruohoruotsi

  committed
  d37386f

  Copy full SHA for d37386f

Commits on Jul 28, 2021

• [ADD] new PyTorch Lightning requirements

  ruohoruotsi

  committed
  e99e075

  Copy full SHA for e99e075

Commits on Jul 27, 2021

• Merge pull request #13 from Ch4ndelier/Ch4ndelier-patch-1

  Show description for 925a0df
  ruohoruotsi

  authored
  925a0df

  Copy full SHA for 925a0df

• Merge pull request #11 from abhishek546/master

  Show description for 4cebde1
  ruohoruotsi

  authored
  4cebde1

  Copy full SHA for 4cebde1

Commits on Dec 5, 2020

• Update lstm_genre_classifier_pytorch.py

  Show description for b157b64
  Ch4ndelier

  authored
  b157b64

  Copy full SHA for b157b64

Commits on Jul 29, 2020

• Update lstm_genre_classifier_keras.py

  abhishek-mvs

  authored
  c331836

  Copy full SHA for c331836

• Update lstm_genre_classifier_keras.py

  abhishek-mvs

  authored
  f2feaf5

  Copy full SHA for f2feaf5

Commits on Jul 2, 2020

• Update lstm_genre_classifier_keras.py

  abhishek-mvs

  authored
  4ca9ec8

  Copy full SHA for 4ca9ec8

• Update lstm_genre_classifier_keras.py

  abhishek-mvs

  authored
  a4ccaad

  Copy full SHA for a4ccaad

Commits on Oct 30, 2019

• [UPDATE] links to dependencies

  ruohoruotsi

  authored
  5f235f6

  Copy full SHA for 5f235f6

Commits on Jul 24, 2019

• Update Readme

  Show description for 82b2f1a
  ruohoruotsi

  authored
  82b2f1a

  Copy full SHA for 82b2f1a

Commits on Jul 19, 2019

• [FIX] english in the raison d'etre

  ruohoruotsi

  committed
  7517cc3

  Copy full SHA for 7517cc3

Commits on Jun 19, 2019

• [ADD] model saving to keras script, clean up comments that explain the architecture

  ruohoruotsi

  committed
  c6f50f6

  Copy full SHA for c6f50f6

Commits on Jun 6, 2019

• [UPDATE] raison d'etre ... clarify which toolkit we're discussing

  ruohoruotsi

  committed
  5be6360

  Copy full SHA for 5be6360

Commits on Jun 1, 2019

• [FIX] comment on dataloader work, to simplify this repo and canonicalize training steps

  ruohoruotsi

  committed
  176f897

  Copy full SHA for 176f897

• [FIX] a few more cosmetic bits with plots and number of epochs to wait before an evaluation

  ruohoruotsi

  committed
  be93582

  Copy full SHA for be93582

• [ADD] plotting of validation loss & accuracy, rearrange code a bit, add comments

  ruohoruotsi

  committed
  8dcccff

  Copy full SHA for 8dcccff

• [FIX] formatting in results

  ruohoruotsi

  committed
  1c141d6

  Copy full SHA for 1c141d6

• [ADD] validation step to training loop. The dynamics of validation acc/loss from epoch to epoch are different keras. Still some visualiation and examination of results to ensure there isn't a bug. …

  Show description for e866d57
  ruohoruotsi

  committed
  e866d57

  Copy full SHA for e866d57

• [CHANGE] name of audio directory from the <<audios>> to <<audio>>, update readme and predict script

  ruohoruotsi

  committed
  2f9eb91

  Copy full SHA for 2f9eb91

Commits on May 16, 2019

• [CLEANUP] add comments, remove debugging prints, clean up layout to better match the keras version of this code

  ruohoruotsi

  committed
  c0da3eb

  Copy full SHA for c0da3eb

• [ADD] rearrangements to model setup and hyper-param specification to better match the keras version of this code

  ruohoruotsi

  committed
  85d1828

  Copy full SHA for 85d1828

• [ADD] clarifications to sequence length computation, default fft size of 2048 needs to be figured into calculation

  ruohoruotsi

  committed
  8e943e3

  Copy full SHA for 8e943e3

Pagination

Previous

Next