Commits

Branch selector

 master

User selector

abattery

Datepicker

All time

Commit History

Commits on Oct 19, 2019

• Find node and registration from the correct subgraph while extracting profile events

  Show description for 294a822
  

  abattery

  authored and

  tensorflower-gardener

  committed
  294a822

  Copy full SHA for 294a822

• GL delegate in label image should be upgraded to v2.

  Show description for de16a4f
  

  abattery

  authored and

  tensorflower-gardener

  committed
  de16a4f

  Copy full SHA for de16a4f

• Fix memory access violation when finding unresolved flex ops in java api

  Show description for 5c99135
  

  abattery

  authored and

  tensorflower-gardener

  committed
  5c99135

  Copy full SHA for 5c99135

Commits on Oct 16, 2019

• Implement subgraph aware profiling tool

  Show description for b366763
  

  abattery

  authored and

  tensorflower-gardener

  committed
  b366763

  Copy full SHA for b366763

Commits on Oct 2, 2019

• Fix AVX2 build

  Show description for 496cc4a
  

  abattery

  authored and

  tensorflower-gardener

  committed
  496cc4a

  Copy full SHA for 496cc4a

Commits on Sep 20, 2019

• Remove eigen's shuffle-based transpose impl

  Show description for 41c5e37
  

  abattery

  authored and

  tensorflower-gardener

  committed
  41c5e37

  Copy full SHA for 41c5e37

• Fix Windows build breakage

  Show description for 85b0981
  

  abattery

  authored and

  tensorflower-gardener

  committed
  85b0981

  Copy full SHA for 85b0981

Commits on Sep 19, 2019

• Add optimized transpose 2d method, written in a template form.

  Show description for 966db41
  

  abattery

  authored and

  tensorflower-gardener

  committed
  966db41

  Copy full SHA for 966db41

Commits on Sep 17, 2019

• Transpose op optimization

  Show description for 376e283
  

  abattery

  authored and

  tensorflower-gardener

  committed
  376e283

  Copy full SHA for 376e283

• Automated rollback of commit 22c2799c5e2ace5b55150f068cc6a9f903b38bf4

  Show description for e95490d
  

  abattery

  authored and

  tensorflower-gardener

  committed
  e95490d

  Copy full SHA for e95490d

Commits on Sep 10, 2019

• Transpose op optimization

  Show description for 2a7d2c4
  

  abattery

  authored and

  tensorflower-gardener

  committed
  2a7d2c4

  Copy full SHA for 2a7d2c4

Commits on Sep 6, 2019

• Unloop more items from tanh and sigmoid for fixed point calculations

  Show description for d17a77b
  

  abattery

  authored and

  tensorflower-gardener

  committed
  d17a77b

  Copy full SHA for d17a77b

Commits on Sep 5, 2019

• Fix neon version of ApplyLayerNorm method in tensor utils

  Show description for 24d9191
  

  abattery

  authored and

  tensorflower-gardener

  committed
  24d9191

  Copy full SHA for 24d9191

Commits on Aug 26, 2019

• Add neon optimized methods for fixed quantized lstm except MatrixBatchVectoryMultiplyAccumulate method

  Show description for 41caf17
  

  abattery

  authored and

  tensorflower-gardener

  committed
  41caf17

  Copy full SHA for 41caf17

Commits on Aug 13, 2019

• Get the fixed point calculation based Tanh and Logistic method with int16 SIMD operations back

  Show description for 98ff991
  

  abattery

  authored and

  tensorflower-gardener

  committed
  98ff991

  Copy full SHA for 98ff991

Commits on Aug 9, 2019

• Add optimized int8 mean implementation.

  Show description for 90be336
  

  abattery

  authored and

  tensorflower-gardener

  committed
  90be336

  Copy full SHA for 90be336

Commits on Aug 8, 2019

• Splits mean op into two modes, reference op and optimized op

  Show description for eb05585
  

  abattery

  authored and

  tensorflower-gardener

  committed
  eb05585

  Copy full SHA for eb05585

Commits on Aug 7, 2019

• Split quantization/dequantization ops into reference and optimized implementations

  Show description for 8c62830
  

  abattery

  authored and

  tensorflower-gardener

  committed
  8c62830

  Copy full SHA for 8c62830

Commits on Aug 5, 2019

• Optimize requantize methods for int8 to int8 and uint8 to uint8

  Show description for e920181
  

  abattery

  authored and

  tensorflower-gardener

  committed
  e920181

  Copy full SHA for e920181

Commits on Aug 2, 2019

• Add optimized dequantize methods for int8 and int16

  Show description for eb4504d
  

  abattery

  authored and

  tensorflower-gardener

  committed
  eb4504d

  Copy full SHA for eb4504d

• Add NEON optimized AffineQuantize methods for int8, uint8 and int16 types

  Show description for 48f2007
  

  abattery

  authored and

  tensorflower-gardener

  committed
  48f2007

  Copy full SHA for 48f2007

• Add NEON optimized Dequantize method

  Show description for 1a30df5
  

  abattery

  authored and

  tensorflower-gardener

  committed
  1a30df5

  Copy full SHA for 1a30df5

Commits on Aug 1, 2019

• Treat the case where the LSTM builtin options is missing as an error while parsing op data

  Show description for 1e91bd2
  

  abattery

  authored and

  tensorflower-gardener

  committed
  1e91bd2

  Copy full SHA for 1e91bd2