Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 18, 2018

• NOTE

  

  HaydenFaulkner

  committedSep 18, 2018
  32cbecc

Commits on Aug 27, 2018

• lil eval script clean

  

  HaydenFaulkner

  committedAug 27, 2018
  b596aeb

• little note in magnet loss about sigma squaring

  

  HaydenFaulkner

  committedAug 27, 2018
  f32ed03

• lil readme update

  

  HaydenFaulkner

  committedAug 27, 2018
  62c6833

• okay let's finalise these experiments and move on to detection capabilities

  

  HaydenFaulkner

  committedAug 27, 2018
  a587ad3

• added evaluation script

  

  HaydenFaulkner

  committedAug 27, 2018
  0bb1b3e

• myloss works better with euc squared dist

  

  HaydenFaulkner

  committedAug 27, 2018
  691f6da

• results coming soon

  

  HaydenFaulkner

  committedAug 27, 2018
  a3bead1

• readme table test3

  

  HaydenFaulkner

  committedAug 27, 2018
  1d3d8b9

• readme table test2

  

  HaydenFaulkner

  committedAug 27, 2018
  5c2d618

• readme table test

  

  HaydenFaulkner

  committedAug 27, 2018
  332004b

• change my loss to just euclid, rather than euclid squared

  

  HaydenFaulkner

  committedAug 27, 2018
  65426ff

• changed to not squared as it works better

  

  HaydenFaulkner

  committedAug 27, 2018
  74fa9b7

• debug batch acc by changin variable name

  

  HaydenFaulkner

  committedAug 27, 2018
  5fbd371

Commits on Aug 26, 2018

• repmet clean again

  

  HaydenFaulkner

  committedAug 26, 2018
  0fdaa30

• clean up and double check of the repmet losses

  

  HaydenFaulkner

  committedAug 26, 2018
  60892b3

• magnet loss debug, fixed denom masking based on clusters rather than cluster classes

  

  HaydenFaulkner

  committedAug 26, 2018
  70183c7

Commits on Aug 25, 2018

• slight update to my loss readme

  

  HaydenFaulkner

  committedAug 25, 2018
  326d511

• adding my own loss, not confident in it, but seems to work better than my repmet implementations

  

  HaydenFaulkner

  committedAug 25, 2018
  260633d

Commits on Aug 24, 2018

• exps

  

  HaydenFaulkner

  committedAug 24, 2018
  dd14ea4

• couple of debugs

  

  HaydenFaulkner

  committedAug 24, 2018
  ed337cf

Commits on Aug 22, 2018

• compare v1, v2, v3 losses exp

  

  HaydenFaulkner

  committedAug 22, 2018
  6e98fc8

• added notes and repmetv3 based on loss bounded by alpha

  

  HaydenFaulkner

  committedAug 22, 2018
  5acf4a2

Commits on Aug 21, 2018

• debug var name

  

  HaydenFaulkner

  committedAug 21, 2018
  a3fe266

• modified the defaults more and setup exp

  

  HaydenFaulkner

  committedAug 21, 2018
  8dd3be9

• modified the defaults to work better

  

  HaydenFaulkner

  committedAug 21, 2018
  b899f0a

Commits on Aug 20, 2018

• added some notes and more exp, oh and better normalisation of the centroids for the repmets

  

  HaydenFaulkner

  committedAug 20, 2018
  cb3d5b7

• more debugging and cleaning up experiments

  

  HaydenFaulkner

  committedAug 20, 2018
  01132f2

Commits on Aug 19, 2018

• debug and added inception backbone

  

  HaydenFaulkner

  committedAug 19, 2018
  d240f9d

Commits on Aug 18, 2018

• added oxford pet, yet to fully debug

  

  HaydenFaulkner

  committedAug 18, 2018
  d2fdba9

• added repmet v2 loss calcs... so simples

  

  HaydenFaulkner

  committedAug 18, 2018
  2468f9e

• nice readme update, yet to implement v2 loss, will be next push

  

  HaydenFaulkner

  committedAug 18, 2018
  a425a29

Commits on Aug 17, 2018

• reorganise, setup more exp, and prep for weekend ... it's workweekend :P

  

  HaydenFaulkner

  committedAug 17, 2018
  ac758ba

Commits on Aug 16, 2018

• debug for 2 sep optimizers for repmet

  

  HaydenFaulkner

  committedAug 16, 2018
  31d51e4

• fixed bug with repmet clusters not getting backpropped properly, todo test versions of nets on all data to determine best arch

  

  HaydenFaulkner

  committedAug 16, 2018
  37722f1

Pagination

PreviousNext