Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 1, 2021

• Merge pull request dblalock#5 from newalter/master

  Show description for 45454e6
  dblalock

  authored
  45454e6

  Copy full SHA for 45454e6

• Merge pull request dblalock#7 from calvinmccarter/master

  Show description for a2c4d72
  dblalock

  authored
  a2c4d72

  Copy full SHA for a2c4d72

Commits on Aug 5, 2021

• bazel runs, all tests passed - 666312 assertions in 8 test cases

  calvinmccarter

  committed
  bf1d528

  Copy full SHA for bf1d528

Commits on Jun 22, 2021

• update readme to mention MADDNESS

  

  anon-paper-submissions-1982

  committed
  faeb551

  Copy full SHA for faeb551

• fix type1 fonts in figs for camera-ready

  

  anon-paper-submissions-1982

  committed
  d30f14a

  Copy full SHA for d30f14a

Commits on Aug 21, 2020

• add scalar quantize results + figs

  dblalock

  committed
  9d19741

  Copy full SHA for 9d19741

Commits on Feb 22, 2020

• actually check in final figure creation code

  dblalock

  committed
  f033286

  Copy full SHA for f033286

Commits on Feb 4, 2020

• make scan + encode speed figs slightly shorter

  dblalock

  committed
  3c4898f

  Copy full SHA for 3c4898f

• update figs to use paper method name + be shorter

  dblalock

  committed
  4e108dc

  Copy full SHA for 4e108dc

Commits on Jan 29, 2020

• switch mithral scan to upcast every 16 elems instead of 128

  dblalock

  committed
  764902b

  Copy full SHA for 764902b

• figs + results fully working

  dblalock

  committed
  5ffc2a2

  Copy full SHA for 5ffc2a2

Commits on Jan 28, 2020

• have profile scan all also profile 4B codes

  dblalock

  committed
  4217210

  Copy full SHA for 4217210

• add ucr softmax option + have mithral use sklearn ridge for stability + have caltech use 1 sample/class to train

  dblalock

  committed
  8bbe42f

  Copy full SHA for 8bbe42f

Commits on Jan 27, 2020

• fix Mithral too large UpcastEvery, try_ndims=1, MithralPQ != Mithral with L=1

  dblalock

  committed
  3e86ac6

  Copy full SHA for 3e86ac6

Commits on Jan 14, 2020

• fix / clean up encode speed fig and make scan speed fig not bold mithral

  dblalock

  committed
  d8ba569

  Copy full SHA for d8ba569

Commits on Jan 13, 2020

• nicer ucr fig

  dblalock

  committed
  1441b2c

  Copy full SHA for 1441b2c

Commits on Jan 10, 2020

• get ucr fig showing plots from different k values

  dblalock

  committed
  0e9886e

  Copy full SHA for 0e9886e

• get caltech fig with both filters working + get ucr with three k values results workingd

  dblalock

  committed
  58f572a

  Copy full SHA for 58f572a

Commits on Jan 5, 2020

• bolt not crashing on difference of gaussians, spca aborting under stricter conditions; expanded caltech + ucr experiments

  dblalock

  committed
  21f1826

  Copy full SHA for 21f1826

Commits on Jan 3, 2020

• add support for caltech difference of gaussians + {DoG,Sobel} output metrics

  dblalock

  committed
  b5d8fd7

  Copy full SHA for b5d8fd7

• have ucr experiments measure rbf acc + also write out raw 1nn acc without snc

  dblalock

  committed
  c40bfd9

  Copy full SHA for c40bfd9

• get ucr fig working

  dblalock

  committed
  7e69564

  Copy full SHA for 7e69564

• get caltech fig working

  dblalock

  committed
  2eb243d

  Copy full SHA for 2eb243d

Commits on Jan 1, 2020

• FIX broken caltech experiments from always using first A seen in amm classes

  dblalock

  committed
  8c04632

  Copy full SHA for 8c04632

Commits on Dec 31, 2019

• get cifar figs looking better

  dblalock

  committed
  1a773e3

  Copy full SHA for 1a773e3

• fix osnap + hashjl amm

  dblalock

  committed
  9b214b2

  Copy full SHA for 9b214b2

Commits on Dec 30, 2019

• get caltech stuff running on more than 2 imgs + with better task names

  dblalock

  committed
  6f698a2

  Copy full SHA for 6f698a2

• get cifar plots with new methods working

  dblalock

  committed
  92162b4

  Copy full SHA for 92162b4

• clean up lut speed + enc speed figs

  dblalock

  committed
  8e468f8

  Copy full SHA for 8e468f8

Commits on Dec 28, 2019

• FIX mithral producing garbage for sufficiently large training sets

  dblalock

  committed
  a1a52c6

  Copy full SHA for a1a52c6

• misc changes

  dblalock

  committed
  f8d8a06

  Copy full SHA for f8d8a06

• add joblib caching to MithralPQ

  dblalock

  committed
  2a5421b

  Copy full SHA for 2a5421b

• correct r_sq computation + add r computation to metrics

  dblalock

  committed
  0380fb9

  Copy full SHA for 0380fb9

• have caltech sliding window extraction use chw layout

  dblalock

  committed
  d90d17f

  Copy full SHA for d90d17f

Commits on Dec 23, 2019

• add random / simple sketching baselines + make sure trainedpca actually is just sucking on caltech, not bugged

  dblalock

  committed
  2d52198

  Copy full SHA for 2d52198

Pagination

Previous

Next