Commits

Branch selector

 master

User selector

lrdxgm

Datepicker

All time

Commit History

Commits on Apr 17, 2020

• Nit: In C++, you need to call the constructor to start a lifetime of an object, or else it is undefined behavior (even though in practice, with many compilers you can simply static_cast raw memory …

  Show description for 907a8e1
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  907a8e1

  Copy full SHA for 907a8e1

Commits on Apr 16, 2020

• It is only necessary types to be trivially destructible in AllocatePOD. Also rename the function, as it's not allocating PODs any more.

  Show description for bb86c3c
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  bb86c3c

  Copy full SHA for bb86c3c

• Report allocation failures in SimpleMemoryAllocator.

  Show description for e4eb45e
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  e4eb45e

  Copy full SHA for e4eb45e

• Change SimpleMemoryAllocator to allow allocating at head of the area.

  Show description for 1aa4d61
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  1aa4d61

  Copy full SHA for 1aa4d61

Commits on Apr 15, 2020

• Add checks that the Allocate function returned successfully.

  Show description for 4a8d851
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  4a8d851

  Copy full SHA for 4a8d851

• Uncomment cstub dependency.

  Show description for 5ccc30b
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  5ccc30b

  Copy full SHA for 5ccc30b

• Clean up ParseOpData:

  Show description for e77122f
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  e77122f

  Copy full SHA for e77122f

Commits on Apr 14, 2020

• Check and report allocation failures in InitializeRuntimeTensor.

  Show description for e2cec54
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  e2cec54

  Copy full SHA for e2cec54

• Do not use TfLitePtrUnion.uint8, it's deprecated; use .data instead.

  Show description for 60140d9
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  60140d9

  Copy full SHA for 60140d9

• Do not use TfLitePtrUnion.raw, it's deprecated; use .data instead.

  Show description for aee4412
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  aee4412

  Copy full SHA for aee4412

Commits on Apr 13, 2020

• Add override specifier to the destructor.

  Show description for 7a895c7
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  7a895c7

  Copy full SHA for 7a895c7

Commits on Apr 10, 2020

• Update hifi_mini cstub location.

  Show description for 6e840b7
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  6e840b7

  Copy full SHA for 6e840b7

Commits on Apr 9, 2020

• HiFi intrinsics: Inline the confusingly named SaturatingMultiply function.

  Show description for 1f9d744
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  1f9d744

  Copy full SHA for 1f9d744

• HiFi intrinsics: Use multiply instead of multiply-accumulate with the accumulator set to zero.

  Show description for bc2c9c6
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  bc2c9c6

  Copy full SHA for bc2c9c6

• HiFi instrinsics: Use AE_MOVPA24 instead of AE_MOVPA24X2 where it makes sense.

  Show description for ab6083f
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  ab6083f

  Copy full SHA for ab6083f

Commits on Apr 8, 2020

• Refactor SoftmaxQuantized (for HiFi mini) so we have a nice error message when the output type is wrong.

  Show description for 7589306
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  7589306

  Copy full SHA for 7589306

Commits on Apr 6, 2020

• Log the human-readable name of current types in "not supported" messages.

  Show description for 76ac3a4
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  76ac3a4

  Copy full SHA for 76ac3a4

Commits on Apr 3, 2020

• Fix -Werror=shadow: local variable inputs hiding this->inputs();

  Show description for 1dca969
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  1dca969

  Copy full SHA for 1dca969

• Fix some compilation warnings:

  Show description for 459b60e
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  459b60e

  Copy full SHA for 459b60e

Commits on Apr 2, 2020

• Change Softmax to use tflite::reference_ops::Softmax instead of xtensa::hifimini::Softmax.

  Show description for 1e9b684
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  1e9b684

  Copy full SHA for 1e9b684

• Silence -Werror,-Wdouble-promotion error.

  Show description for b7e7573
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  b7e7573

  Copy full SHA for b7e7573

Commits on Apr 1, 2020

• Remove empty Init, Free, and Prepare functions.

  Show description for a0422e4
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  a0422e4

  Copy full SHA for a0422e4

• Initialize static TfLiteRegistration objects at compilation time.

  Show description for e9f0ae1
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  e9f0ae1

  Copy full SHA for e9f0ae1

Commits on Mar 26, 2020

• Add int16 output support for Softmax.

  Show description for 3b33da5
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  3b33da5

  Copy full SHA for 3b33da5

• Small Softmax cleanups:

  Show description for 3dc8f81
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  3dc8f81

  Copy full SHA for 3dc8f81

• Remove unused function

  Show description for d264daf
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  d264daf

  Copy full SHA for d264daf

• Nit: Create op_params in the smallest scope.

  Show description for 44430e7
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  44430e7

  Copy full SHA for 44430e7

Commits on Mar 24, 2020

• Fix quant_basic_lstm_test.cc: The previous output and state tensors need to be variable tensors.

  Show description for 605ebe7
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  605ebe7

  Copy full SHA for 605ebe7

• Small cleanup: Use GetInput/GetOutput helper functions.

  Show description for 4dce625
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  4dce625

  Copy full SHA for 4dce625

• Remove Softmax dimension checks: all implementations work on last dimension, other dimensions are just used as batches.

  Show description for cd661aa
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  cd661aa

  Copy full SHA for cd661aa

• Small cleanup: Use GetInput/GetOutput helper functions.

  Show description for d25dbae
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  d25dbae

  Copy full SHA for d25dbae

Commits on Mar 21, 2020

• Small Softmax cleanups:

  Show description for ca16bf3
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  ca16bf3

  Copy full SHA for ca16bf3

Commits on Mar 20, 2020

• Move the templatized integer implementation of Softmax to reference/softmax.h, make it work with uint8_t to replace the uint8 version there.

  Show description for a794c69
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  a794c69

  Copy full SHA for a794c69

• Remove unused function.

  Show description for 2528d18
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  2528d18

  Copy full SHA for 2528d18

Commits on Mar 19, 2020

• Remove obsolete comment.

  Show description for 66e8fea
  

  lrdxgm

  authored and

  tensorflower-gardener

  committed
  66e8fea

  Copy full SHA for 66e8fea

Pagination

Previous

Next