Code

Deepspeech workload variants#627

Code

Closed

pomonam wants to merge 74 commits intomainmlcommons/algorithmic-efficiency:mainfrom deepspeech_workload_variantsmlcommons/algorithmic-efficiency:deepspeech_workload_variantsCopy head branch name to clipboard

+870-154

Commits

Commits on Dec 20, 2023

• modify jax model to add variants

  priyakasimbeg

  committed
  cae33e2

• add jax deepspeech workload variants

  priyakasimbeg

  committed
  b17fc26

• add tanh variant

  priyakasimbeg

  committed
  b042167

• add deepspeech pytorch variants

  priyakasimbeg

  committed
  83f079b

• fix deepspeech batchnorm layer

  priyakasimbeg

  committed
  7e2ca31

• add deepspeech workoad variant names to docker script

  priyakasimbeg

  committed
  50ff71f

Commits on Jan 13, 2024

• Add pass/fail thresholds to traindiffs test

  runame

  committed
  a8187b8

• Add traindiffs_test option to docker startup script

  runame

  committed
  2373e15

• Rename PytWorkload to PyTorchWorkload

  runame

  committed
  d1da9c7

• Add traindiffs tests to workflows (self-hosted)

  runame

  committed
  6a5d63a

Commits on Jan 17, 2024

• Merge branch 'dev' into traindiffs

  Show description for 047475c
  runame

  committed
  047475c

Commits on Jan 18, 2024

• add variant scoring conditions

  priyakasimbeg

  committed
  1683ba3

• add flag for self-tuning rulset

  priyakasimbeg

  committed
  370687d

• score group of submissions

  priyakasimbeg

  committed
  2128ce8

Commits on Jan 19, 2024

• Merge branch 'dev' into scoring_fixes

  priyakasimbeg

  committed
  c65794d

• correct max number of steps

  priyakasimbeg

  committed
  d43ccf4

• add heldout workloads"

  priyakasimbeg

  committed
  fb81436

Commits on Jan 23, 2024

• add trial args to docker startup.sh"

  priyakasimbeg

  committed
  1ea2282

Commits on Jan 24, 2024

• add script for sampling held out workloads

  priyakasimbeg

  committed
  0bcb969

Commits on Jan 25, 2024

• add code for run workloads

  priyakasimbeg

  committed
  ce5f202

• add workload sampling

  priyakasimbeg

  committed
  f431eef

• formatting

  priyakasimbeg

  committed
  f260497

• imports

  priyakasimbeg

  committed
  1a41f8b

• make seed splitting parallelizable

  priyakasimbeg

  committed
  87df162

• fix

  priyakasimbeg

  committed
  9d9cdb9

• formatting

  priyakasimbeg

  committed
  1775307

• Merge pull request #613 from runame/traindiffs

  Show description for 8108c00
  priyakasimbeg

  authored
  8108c00

• held out workloads example

  priyakasimbeg

  committed
  2a11708

• add docker for run_workloads.py

  priyakasimbeg

  committed
  a8385a2

• fix run_workloads.py

  priyakasimbeg

  committed
  ffddbdc

• fix

  priyakasimbeg

  committed
  91cdf34

• add rng seed to startup.sh docker script

  priyakasimbeg

  committed
  95572ad

• fix

  priyakasimbeg

  committed
  d577d5c

• fix

  priyakasimbeg

  committed
  91ff705

Commits on Jan 26, 2024

• fix

  priyakasimbeg

  committed
  296dc1e

• fix

  priyakasimbeg

  committed
  a5b1154

• fix

  priyakasimbeg

  committed
  226544d

• fix

  priyakasimbeg

  committed
  6faad04

• fix log message

  priyakasimbeg

  committed
  9e7def9

• fix

  priyakasimbeg

  committed
  9b410b7

• debug

  priyakasimbeg

  committed
  7634a0b

• debugging

  priyakasimbeg

  committed
  235bc69

• debugging

  priyakasimbeg

  committed
  a8d04cc

• fix

  priyakasimbeg

  committed
  b2571b2

• remove debugging statemetns

  priyakasimbeg

  committed
  18bc347

• fix

  priyakasimbeg

  committed
  4a98698

• formatting

  priyakasimbeg

  committed
  4d38e55

Commits on Jan 27, 2024

• take into account median of studies for scoring

  priyakasimbeg

  committed
  4d413f4

• remove debugging

  priyakasimbeg

  committed
  84c87b9

• formatting

  priyakasimbeg

  committed
  d6e2a36

• documentation

  priyakasimbeg

  committed
  f34838a

Commits on Jan 29, 2024

• Merge branch 'dev' into scoring_fixes

  priyakasimbeg

  authored
  be263a2

Commits on Jan 30, 2024

• fix

  priyakasimbeg

  committed
  84dbb07

• Merge branch 'scoring_fixes' of github.com:mlcommons/algorithmic-efficiency into scoring_fixes

  priyakasimbeg

  committed
  af0b608

Commits on Jan 31, 2024

• remove indexing for rng_subkeys

  priyakasimbeg

  committed
  6d3b0ae

• add documentation

  priyakasimbeg

  committed
  7b23443

• fix documentation

  priyakasimbeg

  committed
  4b77ddd

• add warning

  priyakasimbeg

  committed
  2f48009

• typo

  priyakasimbeg

  committed
  d39eb24

• fix documentation

  priyakasimbeg

  committed
  aecb37f

• remove prng import from generate_held_out_workloads.py

  priyakasimbeg

  committed
  5135cc8

Commits on Feb 1, 2024

• fix technical documentation

  priyakasimbeg

  committed
  6d4f82e

• formatting

  priyakasimbeg

  committed
  aaa1014

• add default for workload metadata config file

  priyakasimbeg

  committed
  761a877

• yapf fix

  priyakasimbeg

  committed
  6b3827a

• import order

  priyakasimbeg

  committed
  c0e1aad

• Merge pull request #618 from mlcommons/scoring_fixes

  Show description for ff3c9b0
  priyakasimbeg

  authored
  ff3c9b0

Commits on Feb 6, 2024

• minor

  pomonam

  committed
  34996e9

• Revert batch size

  pomonam

  committed
  72be296

• minor

  pomonam

  committed
  daef103

• minor

  pomonam

  committed
  3a78765

• minor

  pomonam

  committed
  d718663

• minor

  pomonam

  committed
  4bc8fec

• Merge conflicts

  pomonam

  committed
  0dced90