Commits

Branch selector

 master

User selector

All users

All time

Commit History

Commits on Nov 25, 2019

• Remove project metadata files from docs/ directory.

  Show description for f78f974
  

  lamberta

  authored and

  tensorflower-gardener

  committed
  f78f974

  Copy full SHA for f78f974

Commits on Nov 20, 2019

• Fix scipy import issue.

  Show description for f317240
  

  qlzh727

  authored and

  tensorflower-gardener

  committed
  f317240

  Copy full SHA for f317240

Commits on Nov 19, 2019

• Disable FakeQuants for Softmax.

  Show description for 275cbe0
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  275cbe0

  Copy full SHA for 275cbe0

Commits on Nov 11, 2019

• Enable/Disable DepthwiseConv2D tests for new quant scheme.

  Show description for 73ae4dd
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  73ae4dd

  Copy full SHA for 73ae4dd

• Fix broken pointer to PrunableLayer (to handle custom Keras layers) and block sparsity configuration.

  Show description for 3a278e5
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  3a278e5

  Copy full SHA for 3a278e5

Commits on Nov 8, 2019

• Use new quant scheme in folded Conv/BatchNorm layers.

  Show description for 3b820c8
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  3b820c8

  Copy full SHA for 3b820c8

Commits on Nov 7, 2019

• Only set weights returned by Transform.

  Show description for f251f71
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  f251f71

  Copy full SHA for f251f71

• Create Quantizers specific to Conv/DConv for new quant scheme.

  Show description for 0b4eb1c
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  0b4eb1c

  Copy full SHA for 0b4eb1c

Commits on Nov 6, 2019

• Change pruning_step type to int64 from int32.

  Show description for 2adce8a
  

  Xhark

  authored and

  tensorflower-gardener

  committed
  2adce8a

  Copy full SHA for 2adce8a

• Update Quantizer params to match new scheme.

  Show description for 4a96efe
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  4a96efe

  Copy full SHA for 4a96efe

Commits on Nov 4, 2019

• Move weight building logic to quantizer.

  Show description for 062500a
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  062500a

  Copy full SHA for 062500a

• Pull out narrow_range as a parameter from Quantizers.

  Show description for 4fc81fc
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  4fc81fc

  Copy full SHA for 4fc81fc

Commits on Oct 31, 2019

• Cleanup. Delete old comments, and remove obsolete file.

  Show description for 870681a
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  870681a

  Copy full SHA for 870681a

• Update QuantizeAwareActivation for activations.

  Show description for 4f2dd5b
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  4f2dd5b

  Copy full SHA for 4f2dd5b

Commits on Oct 29, 2019

• Fix build breakage due to keras utils shuffling.

  Show description for d54f3de
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  d54f3de

  Copy full SHA for d54f3de

• Preserve weights in TFLite batch norm fusing transforms.

  Show description for 63718a6
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  63718a6

  Copy full SHA for 63718a6

• Link to contributor guidelines from main Github README.

  Show description for 2bc28ad
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  2bc28ad

  Copy full SHA for 2bc28ad

• Initial contributor guidelines. Work-in-progress.

  Show description for 19f83c9
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  19f83c9

  Copy full SHA for 19f83c9

Commits on Oct 24, 2019

• Implement output quantization of tensors.

  Show description for 681c284
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  681c284

  Copy full SHA for 681c284

• Refactor and update TFLiteQuantizeRegistry.

  Show description for 452b898
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  452b898

  Copy full SHA for 452b898

• Quantization support for post call (result) in QuantizeProvider.

  Show description for 3e65026
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  3e65026

  Copy full SHA for 3e65026

Commits on Oct 22, 2019

• Successful conversion of quantized Mobilenet v1.

  Show description for 1eb9f71
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  1eb9f71

  Copy full SHA for 1eb9f71

Commits on Oct 18, 2019

• Apply model transformation before quantization.

  Show description for eca5344
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  eca5344

  Copy full SHA for eca5344

• TFLite layout transform implementation.

  Show description for a9ae2ca
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  a9ae2ca

  Copy full SHA for a9ae2ca

• Ensure ModelTransformer only accepts Functional models.

  Show description for 837cfa4
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  837cfa4

  Copy full SHA for 837cfa4

Commits on Oct 17, 2019

• Extend QuantizeLayoutTransform to accept quantize map.

  Show description for 9a2a818
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  9a2a818

  Copy full SHA for 9a2a818

• TFLite transformations for batchnorm folding.

  Show description for a9eb9ea
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  a9eb9ea

  Copy full SHA for a9eb9ea

• ModelTransformer now has the ability to receive metadata

  Show description for fdde0c7
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  fdde0c7

  Copy full SHA for fdde0c7

• 1. Quantized DepthwiseConv with folded batchnorm layer.

  Show description for 5bf8c90
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  5bf8c90

  Copy full SHA for 5bf8c90

Commits on Oct 16, 2019

• Remove old quantization API implementation. All functionality has been migrated to new one.

  Show description for 463f020
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  463f020

  Copy full SHA for 463f020

• Only match patterns if layers in candidate layers.

  Show description for d45523a
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  d45523a

  Copy full SHA for d45523a

• Prevent infinite loops when pattern replaces match.

  Show description for 6ec1e45
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  6ec1e45

  Copy full SHA for 6ec1e45

Commits on Oct 7, 2019

• Remove the __init__.py content for keras/utils.

  Show description for 2d3faaa
  

  qlzh727

  authored and

  tensorflower-gardener

  committed
  2d3faaa

  Copy full SHA for 2d3faaa

Commits on Oct 4, 2019

• Quantized form of Conv with folded Batchnorm layer.

  Show description for 479c4c9
  

  alanchiao

  authored and

  tensorflower-gardener

  committed
  479c4c9

  Copy full SHA for 479c4c9

Commits on Oct 3, 2019

• Add support for weights in model transformation.

  Show description for b66edaf
  

  nutsiepully

  authored and

  tensorflower-gardener

  committed
  b66edaf

  Copy full SHA for b66edaf

Pagination

Previous

Next