Commits

Branch selector

 develop-upstream

User selector

chococigar

Datepicker

All time

Commit History

Commits on May 7, 2024

• Adjust rtol value for confirming quantization.

  Show description for 62c4300
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  62c4300

  Copy full SHA for 62c4300

Commits on Apr 15, 2024

• (1/N) Remove StableHLO Quantizer quantize pass dependency to preset.

  Show description for 7aac924
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  7aac924

  Copy full SHA for 7aac924

Commits on Apr 12, 2024

• Create an OSS version of StableHLO Quantizer demonstration for ODML.

  Show description for 88310dd
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  88310dd

  Copy full SHA for 88310dd

Commits on Apr 9, 2024

• Add proper QuantizationSpec for WeightOnlyPtq.

  Show description for 7a20b31
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  7a20b31

  Copy full SHA for 7a20b31

Commits on Apr 8, 2024

• Fix conditions for tfl.batch_matmul legalization and e2e tests.

  Show description for e131656
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  e131656

  Copy full SHA for e131656

Commits on Apr 4, 2024

• Support lifting of same shape bias for stablehlo.convolution.

  Show description for 59d0af4
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  59d0af4

  Copy full SHA for 59d0af4

• Relax subgraph byte constraint to required_bytes <= bytes rather than exact match.

  Show description for 8d698c3
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  8d698c3

  Copy full SHA for 8d698c3

Commits on Apr 3, 2024

• Preserve stablehlo.gather requantize pattern to meet operand and result type constraint.

  Show description for 5568a4d
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  5568a4d

  Copy full SHA for 5568a4d

Commits on Mar 28, 2024

• Add quantization/legalization for stablehlo.add and respective pipeline changes.

  Show description for 9926a40
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  9926a40

  Copy full SHA for 9926a40

• Do not wrap lifted function in TF.CustomAggregatorOp with improper quantization_method.

  Show description for 2457ae2
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  2457ae2

  Copy full SHA for 2457ae2

Commits on Mar 25, 2024

• Refactor patterns in quantize_patterns.cc depending on compute-heaviness.

  Show description for 8fcb611
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  8fcb611

  Copy full SHA for 8fcb611

• Add a function to determine whether an op is inside a StableHLO op with Region.

  Show description for 2a8da79
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  2a8da79

  Copy full SHA for 2a8da79

Commits on Mar 21, 2024

• Migrate UniformSupport* and FakeQuantSupport* components from lite directory.

  Show description for 635a857
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  635a857

  Copy full SHA for 635a857

Commits on Mar 20, 2024

• Add check conditions in quantization_driver_test.cc.

  Show description for 3d281e8
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  3d281e8

  Copy full SHA for 3d281e8

Commits on Mar 19, 2024

• Fixes SAME padding calculation to produce float results for correct std::ceil input.

  Show description for 40ae4b6
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  40ae4b6

  Copy full SHA for 40ae4b6

Commits on Mar 15, 2024

• Fix forward: address tsan issue in prepare_quantize.cc.

  Show description for a3cd06b
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  a3cd06b

  Copy full SHA for a3cd06b

Commits on Mar 14, 2024

• Support per-channel quantization of dot_general layers.

  Show description for 6700669
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  6700669

  Copy full SHA for 6700669

Commits on Mar 13, 2024

• Remove stale bugs from quantization_lib.

  Show description for 004e9ee
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  004e9ee

  Copy full SHA for 004e9ee

Commits on Mar 11, 2024

• Support stablehlo.convolution legalization to tfl.transpose_conv op.

  Show description for 279fe90
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  279fe90

  Copy full SHA for 279fe90

• Add Dynamic Range Quantized op support for op_stat_pass.cc.

  Show description for c3fd2b6
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  c3fd2b6

  Copy full SHA for c3fd2b6

Commits on Mar 8, 2024

• Rewrite TODO's according to latest syntax.

  Show description for 690d3ab
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  690d3ab

  Copy full SHA for 690d3ab

Commits on Mar 7, 2024

• Support SAME padding legalization when stride > 1.

  Show description for 70097ad
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  70097ad

  Copy full SHA for 70097ad

• Support SAME padding legalization when stride > 1.

  Show description for 57de1c5
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  57de1c5

  Copy full SHA for 57de1c5

• Add tests to evaluate XlaCallModule op.

  Show description for 963268c
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  963268c

  Copy full SHA for 963268c

Commits on Mar 6, 2024

• TFLite Interpreter: Add operand / result tensor type list in _get_op_details.

  Show description for 98294ab
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  98294ab

  Copy full SHA for 98294ab

Commits on Mar 5, 2024

• Cleanup test files in mlir/quantization.

  Show description for f33fa68
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  f33fa68

  Copy full SHA for f33fa68

• (2/N) Apply c++ format to stablehlo quantization passes.

  Show description for 41becb5
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  41becb5

  Copy full SHA for 41becb5

• Migrate common quantization components from lite directory.

  Show description for 16f2b0b
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  16f2b0b

  Copy full SHA for 16f2b0b

Commits on Feb 29, 2024

• (2/N) Fix c++ formats in quantization_lib.

  Show description for 31481eb
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  31481eb

  Copy full SHA for 31481eb

Commits on Feb 28, 2024

• Apply c++ format to stablehlo quantization passess.

  Show description for 09f8b71
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  09f8b71

  Copy full SHA for 09f8b71

• Fix c++ formats in quantization_lib.

  Show description for f7f9936
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  f7f9936

  Copy full SHA for f7f9936

Commits on Feb 27, 2024

• Remove unused logging formats in quantization_driver.cc.

  Show description for a5a8519
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  a5a8519

  Copy full SHA for a5a8519

Commits on Feb 26, 2024

• Add tests for quantization_driver.

  Show description for 156a1bc
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  156a1bc

  Copy full SHA for 156a1bc

• Cleanup quantization_utils: basic formatting

  Show description for 8d0c35b
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  8d0c35b

  Copy full SHA for 8d0c35b

• Cleanup quantization_driver.cc: refactoring

  Show description for 492adc6
  

  chococigar

  authored and

  tensorflower-gardener

  committed
  492adc6

  Copy full SHA for 492adc6

Pagination

Previous

Next