Code

LM Toolkit Refactor#381

Code

Merged

tab-cmd merged 57 commits into2.0.0CAMBI-tech/BciPy:2.0.0from

lm-toolkitCAMBI-tech/BciPy:lm-toolkitCopy head branch name to clipboard

Apr 29, 2025

Commits

Commits on Jan 8, 2025

• add script for simulating lm change with different phrases, small updates

  tab-cmd

  committed
  737a8e2

  Copy full SHA for 737a8e2

• add processing script, add NullDAQ

  tab-cmd

  committed
  a18a35c

  Copy full SHA for a18a35c

• Update to custom typing parameters

  Show description for 8d9fb88
  tab-cmd

  committed
  8d9fb88

  Copy full SHA for 8d9fb88

• Add progress bar and WIP update to average phrases across LM

  tab-cmd

  committed
  c68803c

  Copy full SHA for c68803c

• update the process script to the new output format

  tab-cmd

  committed
  c284227

  Copy full SHA for c284227

• Add final phrases

  tab-cmd

  committed
  0bdcfc7

  Copy full SHA for 0bdcfc7

• add plotting and stats to the processing script

  tab-cmd

  committed
  ab9244a

  Copy full SHA for ab9244a

• add missing params command

  tab-cmd

  committed
  1b12409

  Copy full SHA for 1b12409

Commits on Jan 9, 2025

• add more logging and custom metrics

  tab-cmd

  committed
  205d929

  Copy full SHA for 205d929

• Integration of causal model, add phrases, update processing scripts

  tab-cmd

  committed
  ff6e9e2

  Copy full SHA for ff6e9e2

Commits on Jan 15, 2025

• matrix processing

  tab-cmd

  committed
  a58b8cf

  Copy full SHA for a58b8cf

Commits on Jan 24, 2025

• update figure

  tab-cmd

  committed
  0dc02fe

  Copy full SHA for 0dc02fe

Commits on Feb 27, 2025

• move script to a group demo

  tab-cmd

  committed
  2ac395e

  Copy full SHA for 2ac395e

• Merge remote-tracking branch 'origin/2.0.0' into BANFF_lm

  tab-cmd

  committed
  a262d40

  Copy full SHA for a262d40

• reset devices default

  tab-cmd

  committed
  9b29a51

  Copy full SHA for 9b29a51

• remove retry logic from language model init

  tab-cmd

  committed
  32b82f6

  Copy full SHA for 32b82f6

• reset static defaults

  tab-cmd

  committed
  8cf25e1

  Copy full SHA for 8cf25e1

• update parameters

  tab-cmd

  committed
  a43ba68

  Copy full SHA for a43ba68

• remove bad test

  tab-cmd

  committed
  0fb843a

  Copy full SHA for 0fb843a

• lint

  tab-cmd

  committed
  7e5994b

  Copy full SHA for 7e5994b

• remove integration tests (for now) and add some info to sim README + linting

  tab-cmd

  committed
  3ee8b9b

  Copy full SHA for 3ee8b9b

• drop support for 3.8

  tab-cmd

  committed
  a9ebcd4

  Copy full SHA for a9ebcd4

Commits on Mar 5, 2025

• Added textpredict dependency, removed LM dependencies that are included in that package now

  dcgaines

  committed
  7b7e560

  Copy full SHA for 7b7e560

• Refactored main language model classes into adapters that use the new textpredict package

  dcgaines

  committed
  4f827d9

  Copy full SHA for 4f827d9

• Renamed ngram model

  dcgaines

  committed
  b408af6

  Copy full SHA for b408af6

• Updated imports

  dcgaines

  committed
  3461308

  Copy full SHA for 3461308

• More ngram renaming

  dcgaines

  committed
  5f3016f

  Copy full SHA for 5f3016f

• More ngram renaming, adjusted mixture default params

  dcgaines

  committed
  956ca98

  Copy full SHA for 956ca98

• Updated textpredict version

  dcgaines

  committed
  7d96e9a

  Copy full SHA for 7d96e9a

• Converted mixture model to adapter

  dcgaines

  committed
  0c4a167

  Copy full SHA for 0c4a167

Commits on Mar 6, 2025

• Conveted oracle model into adapter

  dcgaines

  committed
  7401c97

  Copy full SHA for 7401c97

• Deprecated InvalidLanguageModelException from bcipy in favor of aactextpredict's exception

  dcgaines

  committed
  63f8cd2

  Copy full SHA for 63f8cd2

• Upgraded transformers version to address pytorch deprecation warnings

  dcgaines

  committed
  c75ad28

  Copy full SHA for c75ad28

• Adjusted max bump to 1, all mass on target

  dcgaines

  committed
  a937b72

  Copy full SHA for a937b72

• Updated test cases to use adapters

  dcgaines

  committed
  8384e34

  Copy full SHA for 8384e34

• Store bcipy symbol set alongside toolkit model symbol set

  dcgaines

  committed
  911676b

  Copy full SHA for 911676b

Commits on Mar 7, 2025

• Updated demos to use new adapter classes

  dcgaines

  committed
  6e1c4b5

  Copy full SHA for 6e1c4b5

• Updated test class names

  dcgaines

  committed
  2e2b97e

  Copy full SHA for 2e2b97e

• Fixed LM class references

  dcgaines

  committed
  204134f

  Copy full SHA for 204134f

• Updated LM documentation

  dcgaines

  committed
  1106682

  Copy full SHA for 1106682

• Fixed misc LM class references

  dcgaines

  committed
  1f337f0

  Copy full SHA for 1f337f0

• Updated LM class references

  dcgaines

  committed
  0a421f5

  Copy full SHA for 0a421f5

• Merge pull request #379 from CAMBI-tech/2.0.0

  Show description for 3151958
  dcgaines

  authored
  3151958

  Copy full SHA for 3151958

• Merge branch 'lm-toolkit' into BANFF_lm

  dcgaines

  authored
  71bbe11

  Copy full SHA for 71bbe11

• Merge pull request #380 from CAMBI-tech/BANFF_lm

  Show description for b073149
  dcgaines

  authored
  b073149

  Copy full SHA for b073149

Commits on Mar 20, 2025

• Merge branch '2.0.0' into lm-toolkit

  tab-cmd

  committed
  fceca14

  Copy full SHA for fceca14

• lint

  tab-cmd

  committed
  f809adb

  Copy full SHA for f809adb

Commits on Mar 24, 2025

• Update textpredict dependency to fix 3.10

  dcgaines

  committed
  de49406

  Copy full SHA for de49406

Commits on Mar 28, 2025

• Converted LanguageModelAdapter base class to BciPyLanguageModel protocol. Adjusted subclasses and references

  dcgaines

  committed
  616d57f

  Copy full SHA for 616d57f

• Restored BciPy's own uniform model

  dcgaines

  committed
  348decd

  Copy full SHA for 348decd

Commits on Apr 2, 2025

• Renamed base LM class back to LanguageModel

  dcgaines

  committed
  a52209f

  Copy full SHA for a52209f

Commits on Apr 15, 2025

• Removed language module exclusion from mypy

  dcgaines

  committed
  9831833

  Copy full SHA for 9831833

• Deprecated response type, added separate protocols for char and word LMs

  dcgaines

  committed
  7e3cd92

  Copy full SHA for 7e3cd92

• Updated dependency to the new toolkit name

  dcgaines

  committed
  a22f375

  Copy full SHA for a22f375

• Loosened textslinger dependency to allow minor changes without updating requirement every time

  dcgaines

  committed
  3fe2cf9

  Copy full SHA for 3fe2cf9

Commits on Apr 22, 2025

• Refactored Language Model protocols; fixed linting issues and import sort order.

  lawhead

  committed
  777ba10

  Copy full SHA for 777ba10

Commits on Apr 23, 2025

• Merge pull request #390 from CAMBI-tech/lm-toolkit-refinements

  Show description for 6779982
  dcgaines

  authored
  6779982

  Copy full SHA for 6779982