Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Jun 3, 2019

• [Breaking] PTQ: Removed special handling of clipping overrides

  Show description for 3cde6c5
  

  levzlotnik

  authored and

  guyjacob

  committed
  3cde6c5

  Copy full SHA for 3cde6c5

Commits on May 30, 2019

• MNIST support

  Show description for f8085cf
  nzmora

  committed
  f8085cf

  Copy full SHA for f8085cf

Commits on May 27, 2019

• Bug fix for shared module (IntelLabs#268)

  Show description for d6efbe4
  levzlotnik

  authored
  d6efbe4

  Copy full SHA for d6efbe4

Commits on May 20, 2019

• Thinning: added support for group-wise convolutions

  Show description for 6b83202
  nzmora

  committed
  6b83202

  Copy full SHA for 6b83202

Commits on May 19, 2019

• Bugfix in bias handling in quant-aware training (fixes issue IntelLabs#248)

  guyjacob

  committed
  4c16369

  Copy full SHA for 4c16369

Commits on May 16, 2019

• Refactoring: utils.get_dummy_input()

  Show description for bf1e6a0
  nzmora

  committed
  bf1e6a0

  Copy full SHA for bf1e6a0

• Fix test_onnx.py

  Show description for af5c721
  nzmora

  committed
  af5c721

  Copy full SHA for af5c721

• Refactor export to ONNX functionality (IntelLabs#258)

  Show description for 5430481
  

  barrh

  authored and

  nzmora

  committed
  5430481

  Copy full SHA for 5430481

Commits on May 15, 2019

• SummaryGraph: fix ‘weights_vol’ attribute for conv and linear layers

  Show description for 08b5cd9
  nzmora

  committed
  08b5cd9

  Copy full SHA for 08b5cd9

• Revert "SummaryGraph: fix ‘weights_vol’ attribute for conv and linear layers"

  Show description for a0ebeb7
  nzmora

  committed
  a0ebeb7

  Copy full SHA for a0ebeb7

• SummaryGraph: fix ‘weights_vol’ attribute for conv and linear layers

  Show description for a3f2ce2
  nzmora

  committed
  a3f2ce2

  Copy full SHA for a3f2ce2

Commits on May 14, 2019

• Fix failing test after last post-train quant change

  guyjacob

  committed
  f06662d

  Copy full SHA for f06662d

Commits on May 6, 2019

• Fix broken load test (IntelLabs#245)

  Show description for 1f48fa6
  

  barrh

  authored and

  nzmora

  committed
  1f48fa6

  Copy full SHA for 1f48fa6

Commits on May 2, 2019

• Quantizer: Proper handling of modules that point to same object (IntelLabs#239)

  

  levzlotnik

  authored and

  guyjacob

  committed
  a69dd5d

  Copy full SHA for a69dd5d

Commits on Apr 30, 2019

• Added PackedSequence functionality (IntelLabs#236)

  Show description for 92fd001
  levzlotnik

  authored
  92fd001

  Copy full SHA for 92fd001

Commits on Apr 18, 2019

• Remove single worker limitation in deterministic mode (IntelLabs#227)

  Show description for 8c5de42
  

  barrh

  authored and

  guyjacob

  committed
  8c5de42

  Copy full SHA for 8c5de42

Commits on Apr 16, 2019

• LSTM: Modular implementation + Post-Train Quantization Sample (IntelLabs#196)

  Show description for a3c8d86
  

  levzlotnik

  authored and

  guyjacob

  committed
  a3c8d86

  Copy full SHA for a3c8d86

Commits on Apr 14, 2019

• Post-train quant: Extend acts clipping functionality (IntelLabs#225)

  Show description for 437e270
  guyjacob

  authored
  437e270

  Copy full SHA for 437e270

Commits on Apr 9, 2019

• SummaryGraph - fix MACs calculation for grouped-convolutions

  Show description for 1a8c6bb
  nzmora

  committed
  1a8c6bb

  Copy full SHA for 1a8c6bb

• Relaxation of the SummaryGraph API (IntelLabs#212)

  Show description for d08c373
  

  barrh

  authored and

  nzmora

  committed
  d08c373

  Copy full SHA for d08c373

Commits on Apr 8, 2019

• Removed sys.path modifications when importing distiller. (IntelLabs#224)

  levzlotnik

  authored
  72ef916

  Copy full SHA for 72ef916

• Fix issue IntelLabs#213 (IntelLabs#221)

  Show description for 73b3b3c
  nzmora

  authored
  73b3b3c

  Copy full SHA for 73b3b3c

Commits on Apr 1, 2019

• Quantizer: Specify # bias bits + custom overrides (BREAKING) (IntelLabs#178)

  Show description for 5271625
  

  levzlotnik

  authored and

  guyjacob

  committed
  5271625

  Copy full SHA for 5271625

• Load optimizer from checkpoint (BREAKING - see details) (IntelLabs#182)

  Show description for 992291c
  

  barrh

  authored and

  guyjacob

  committed
  992291c

  Copy full SHA for 992291c

Commits on Mar 27, 2019

• Fix argument typo in a couple of YAMLs (fix IntelLabs#191)

  guyjacob

  committed
  5800f35

  Copy full SHA for 5800f35

Commits on Mar 11, 2019

• Integrate Cadene pretrained PyTorch models [fixes IntelLabs#142] (IntelLabs#184)

  Show description for 321abb6
  

  barrh

  authored and

  nzmora

  committed
  321abb6

  Copy full SHA for 321abb6

Commits on Feb 26, 2019

• PyTorch 1.0.0 support + Proper Packaging (Release 0.3) (IntelLabs#144)

  Show description for 62862a0
  levzlotnik

  authored
  62862a0

  Copy full SHA for 62862a0

Commits on Feb 12, 2019

• CPU support: fix the case of loading a thinned GPU-model on the CPU

  Show description for ba05f6c
  nzmora

  committed
  ba05f6c

  Copy full SHA for ba05f6c

• Fix issue IntelLabs#148 + refactor load_checkpoint.py (IntelLabs#153)

  Show description for 1210f41
  nzmora

  authored
  1210f41

  Copy full SHA for 1210f41

Commits on Feb 11, 2019

• Post-train quant based on stats + additional modules quantized (IntelLabs#136)

  Show description for 28a8ee1
  guyjacob

  authored
  28a8ee1

  Copy full SHA for 28a8ee1

Commits on Feb 10, 2019

• Load different random subset of dataset on each epoch (IntelLabs#149)

  Show description for 4b1d0c8
  guyjacob

  authored
  4b1d0c8

  Copy full SHA for 4b1d0c8

Commits on Jan 24, 2019

• Bugfix in test_quantizer

  guyjacob

  committed
  8d694a0

  Copy full SHA for 8d694a0

Commits on Jan 23, 2019

• Quant-aware training: Quantize bias to 32 bits (Hard-coded for now)

  guyjacob

  committed
  c98df54

  Copy full SHA for c98df54

• Fix dtypes in test_post_train_quant.py

  guyjacob

  committed
  0dac3e0

  Copy full SHA for 0dac3e0

Commits on Jan 21, 2019

• Updates to quantization parameters calculation:

  Show description for 10ce938
  guyjacob

  committed
  10ce938

  Copy full SHA for 10ce938

Pagination

Previous

Next