Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on May 21, 2017

• switch back to mnist

  PeterMitrano

  committedMay 21, 2017
  cd2c7aa

Commits on May 20, 2017

• FUCKKKKKK

  PeterMitrano

  committedMay 20, 2017
  2d26cea

• even more data

  PeterMitrano

  committedMay 20, 2017
  851363c

• more data

  PeterMitrano

  committedMay 20, 2017
  f6ffd28

• nonoverlaping blocks

  PeterMitrano

  committedMay 20, 2017
  3bac52f

• change block size

  PeterMitrano

  committedMay 20, 2017
  942f6fb

• change block size

  PeterMitrano

  committedMay 20, 2017
  86d84e7

• changes for clarity only

  PeterMitrano

  committedMay 20, 2017
  440ec9b

• gonna try running again

  PeterMitrano

  committedMay 20, 2017
  491498b

• tried to fix segment ids

  PeterMitrano

  committedMay 20, 2017
  d2956b1

Commits on May 19, 2017

• trying big dataset

  PeterMitrano

  committedMay 19, 2017
  17309ba

• C=1e-7

  PeterMitrano

  committedMay 19, 2017
  d550da9

• C=1e-2

  PeterMitrano

  committedMay 19, 2017
  b7e0bac

• C=1e-1

  PeterMitrano

  committedMay 19, 2017
  5fe0a31

• ok so now the images match what they should be

  PeterMitrano

  committedMay 19, 2017
  e24d616

• still does really bad

  PeterMitrano

  committedMay 19, 2017
  09b4ec1

• holy shit it worked

  PeterMitrano

  committedMay 19, 2017
  98d6d6f

• the image summary in the class doesn't update

  PeterMitrano

  committedMay 19, 2017
  3e3ed1b

• test dataset shows the tfrecord is probably fine

  PeterMitrano

  committedMay 19, 2017
  a7ec15b

• upgrade to tf 1.1.0-rc2

  PeterMitrano

  committedMay 19, 2017
  cb4d33d

• make it deterministic

  PeterMitrano

  committedMay 19, 2017
  589ce6c

• turns out test labels are just wrong

  PeterMitrano

  committedMay 19, 2017
  df1d561

• checking labels

  PeterMitrano

  committedMay 19, 2017
  d201d35

• turns out I can't have image_batch None to start

  PeterMitrano

  committedMay 19, 2017
  b937470

• fixed a bug input source (I think)

  PeterMitrano

  committedMay 19, 2017
  0e83b1f

• messing with summaries still

  PeterMitrano

  committedMay 19, 2017
  b987715

• messing with summaries

  PeterMitrano

  committedMay 19, 2017
  33870ca

• train on test and test on train somehow worked???

  PeterMitrano

  committedMay 19, 2017
  5d8c55a

• new batches do work for train data

  PeterMitrano

  committedMay 19, 2017
  cf15ea3

• not sure batches are working

  PeterMitrano

  committedMay 19, 2017
  7bbfc16

• run with train instead of test to eval

  PeterMitrano

  committedMay 19, 2017
  1d738e3

• running with flipped sign eigenvectors

  PeterMitrano

  committedMay 19, 2017
  03e3d80

• it seems my eigenvectors are opposite sign as OG

  PeterMitrano

  committedMay 19, 2017
  d49af05

• re-ran with larger batch to check covariance

  PeterMitrano

  committedMay 19, 2017
  551c56f

• stage 1 covariance matrices are similar when normalized

  PeterMitrano

  committedMay 19, 2017
  0769c53

Pagination

PreviousNext