Code

Streaming ww dev#168

Code

Merged

Peter-Chang merged 183 commits intomlcommons:mastermlcommons/tiny:masterfrom jeremy-syn:streaming_ww_devjeremy-syn/tiny:streaming_ww_devCopy head branch name to clipboardJan 20, 2025

+209,143-5

Commits

Commits on Oct 9, 2023

• moved training_torch to experimental and added a README

  jeremy-syn

  committed
  b16b8d9

  Copy full SHA for b16b8d9

Commits on Jan 22, 2024

• starting to move code into here

  jeremy-syn

  committed
  31dd4be

  Copy full SHA for 31dd4be

Commits on Jan 27, 2024

• some updates to streaming wakeword

  jeremy-syn

  committed
  0626c97

  Copy full SHA for 0626c97

• updated streaming wakeword model to be the actual candidate DS-TCN model w/ no residual layers

  jeremy-syn

  committed
  03c9d8b

  Copy full SHA for 03c9d8b

• set default features to 40-D LFBEs

  jeremy-syn

  committed
  725cb69

  Copy full SHA for 725cb69

Commits on Jan 29, 2024

• changed num_classes to 3 in train.py

  jeremy-syn

  committed
  8e533a3

  Copy full SHA for 8e533a3

• demo notebook (in progress) added

  jeremy-syn

  committed
  9c15b56

  Copy full SHA for 9c15b56

• demo notebook runs through small training run

  jeremy-syn

  committed
  5475c7b

  Copy full SHA for 5475c7b

• added count_labels and is_batched()

  jeremy-syn

  committed
  c7138c4

  Copy full SHA for c7138c4

• demo now adds some silence waveforms (which then have noise added) to training dataset

  jeremy-syn

  committed
  b089ce0

  Copy full SHA for b089ce0

• updated get_dataset (mostly copied from demo.ipynb) and removed use_sam part from train.py

  jeremy-syn

  committed
  6c8f06d

  Copy full SHA for 6c8f06d

• fixed default model architecture flag

  jeremy-syn

  committed
  e7922f7

  Copy full SHA for e7922f7

• fixed some issues with building model

  jeremy-syn

  committed
  5a9a235

  Copy full SHA for 5a9a235

Commits on Jan 31, 2024

• added from_logits argument to model.compile

  jeremy-syn

  committed
  7193979

  Copy full SHA for 7193979

Commits on Feb 2, 2024

• cleanup changes to get_dataset and demo notebook

  jeremy-syn

  committed
  5e74320

  Copy full SHA for 5e74320

• catching up on edits

  jeremy-syn

  committed
  4e02106

  Copy full SHA for 4e02106

• keras_model does not need tf datasets module

  jeremy-syn

  committed
  5d9932c

  Copy full SHA for 5d9932c

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  4edc86e

  Copy full SHA for 4edc86e

Commits on Feb 5, 2024

• cleaning up demo notebook

  jeremy-syn

  committed
  b72d850

  Copy full SHA for b72d850

• cleaning up demo notebook

  jeremy-syn

  committed
  fc7a4d8

  Copy full SHA for fc7a4d8

• made path to speech commands dataset easier to config per location/user without upsetting git

  jeremy-syn

  committed
  e9de194

  Copy full SHA for e9de194

• beginning of code to test long waveform in python

  jeremy-syn

  committed
  73e3d63

  Copy full SHA for 73e3d63

Commits on Feb 7, 2024

• some updates

  jeremy-syn

  committed
  2d69501

  Copy full SHA for 2d69501

• added option to read in model config file

  jeremy-syn

  committed
  6916250

  Copy full SHA for 6916250

• set validation set to incorporate background noise. also fixed issue where background_volume option was being ignored

  jeremy-syn

  committed
  e91a3ed

  Copy full SHA for e91a3ed

• moved code to add silent (or white noise) frames to dataset into its function. Applies to val set now, but still not test set

  jeremy-syn

  committed
  eb4d120

  Copy full SHA for eb4d120

• fixed argument error

  jeremy-syn

  committed
  29da207

  Copy full SHA for 29da207

Commits on Feb 10, 2024

• added post-training quantization

  jeremy-syn

  committed
  c6f23f1

  Copy full SHA for c6f23f1

Commits on Feb 11, 2024

• several changes in order to use QAT and evaluate on long waveforms:

  Show description for f478259
  jeremy-syn

  committed
  f478259

  Copy full SHA for f478259

• notebook updated to work with last commits on get_data, keras_model

  jeremy-syn

  committed
  b46d797

  Copy full SHA for b46d797

• changed default LR schedule to reduce_on_plateau so it scales better with more epochs

  jeremy-syn

  committed
  c6185a7

  Copy full SHA for c6185a7

• some more edits to get QAT working

  jeremy-syn

  committed
  9dd2f1b

  Copy full SHA for 9dd2f1b

Commits on Feb 13, 2024

• changed labels to one-hot to work with precision/recall metrics. Also changed feature extractor code in dataset preparation to optionally work in a standalone model.

  jeremy-syn

  committed
  19470b0

  Copy full SHA for 19470b0

• added notebook to develop tflite model for feature extraction

  jeremy-syn

  committed
  772e6cc

  Copy full SHA for 772e6cc

• removed some prints from get_dataset. added an evaluation to train

  jeremy-syn

  committed
  a6ac856

  Copy full SHA for a6ac856

• adjusted reduce lr on plateau settings

  jeremy-syn

  committed
  7302a8d

  Copy full SHA for 7302a8d

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  20abafc

  Copy full SHA for 20abafc

• fixed plotting error

  jeremy-syn

  committed
  8ef587c

  Copy full SHA for 8ef587c

Commits on Feb 14, 2024

• working on different options to run the feature extractor on MCU

  jeremy-syn

  committed
  deec7c9

  Copy full SHA for deec7c9

Commits on Feb 17, 2024

• small changes to notebook

  jeremy-syn

  committed
  468a344

  Copy full SHA for 468a344

• removed old commented-out code that loaded pre-built dataset

  jeremy-syn

  committed
  2964369

  Copy full SHA for 2964369

• tflite_feature_extractor.ipynb very much a work in progress

  jeremy-syn

  committed
  ee12fe4

  Copy full SHA for ee12fe4

Commits on Mar 6, 2024

• added setup instructions and a to the streaming wakeword benchmark (mlcommons#155)

  AlexMontgomerie

  authored
  ac40098

  Copy full SHA for ac40098

Commits on Apr 15, 2024

• Merge branch 'master' of gh-syn:mlcommons/tiny

  jeremy-syn

  committed
  380f117

  Copy full SHA for 380f117

Commits on May 21, 2024

• cache datasets after spectrogram computation to avoid recomputing them at every epoch

  jeremy-syn

  committed
  9798093

  Copy full SHA for 9798093

• fixed data_dir default to point to speech_commands_v0.02

  jeremy-syn

  committed
  c584f23

  Copy full SHA for c584f23

• fixed data_dir default to point to speech_commands_v0.02

  jeremy-syn

  committed
  12d727f

  Copy full SHA for 12d727f

• added BooleanOptionalAction to correctly parse boolean Flags

  jeremy-syn

  committed
  0e122cf

  Copy full SHA for 0e122cf

• fixed parsing of bool args (use_qat, run_test_set) to work with python 3.8

  jeremy-syn

  committed
  7bfefc4

  Copy full SHA for 7bfefc4

Commits on May 26, 2024

• changed so parse_command raises exception on unrecognized flags

  jeremy-syn

  committed
  49330a9

  Copy full SHA for 49330a9

• changed so parse_command raises exception on unrecognized flags

  jeremy-syn

  committed
  6dc9314

  Copy full SHA for 6dc9314

Commits on May 28, 2024

• added foreground scaling args foreground_volume_min, _max to train on quiter wakewords. Also changed defaults for data/background paths (to HOME/speech_commands_v0.02) to align with default filename

  jeremy-syn

  committed
  252cafc

  Copy full SHA for 252cafc

Commits on Jun 3, 2024

• set is_training true for ds_val so it gets noise added

  jeremy-syn

  committed
  2d80a04

  Copy full SHA for 2d80a04

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  7270a9a

  Copy full SHA for 7270a9a

• edits to str ww model

  jeremy-syn

  committed
  5cf7657

  Copy full SHA for 5cf7657

• edits to data set building

  jeremy-syn

  committed
  eca617c

  Copy full SHA for eca617c

• saved training history along with plot

  jeremy-syn

  committed
  0e7ac0f

  Copy full SHA for 0e7ac0f

• Merge branch 'streaming_ww_dev' of gh-syn:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  8bedf74

  Copy full SHA for 8bedf74

• removed average pooling, increased initial feature stride

  jeremy-syn

  committed
  7ac2852

  Copy full SHA for 7ac2852

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  01470d7

  Copy full SHA for 01470d7

Commits on Jun 4, 2024

• Fixed bug where np.random is only evaluated at graph creation, so all foregrounds are scaled by the same amount. Also added condition so empty frames are not added to calibration set.

  jeremy-syn

  committed
  3b0fb65

  Copy full SHA for 3b0fb65

Commits on Jun 5, 2024

• fixed several places where np.random was used in a tf graph, resulting in the same value being used for the whole dataset

  jeremy-syn

  committed
  3f3dabf

  Copy full SHA for 3f3dabf

• widened filters in 2nd,3rd layers to 128

  jeremy-syn

  committed
  4fd29c1

  Copy full SHA for 4fd29c1

• changed back from 32 LFBEs to 40

  jeremy-syn

  committed
  6f8a272

  Copy full SHA for 6f8a272

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  ac85b48

  Copy full SHA for ac85b48

Commits on Jun 6, 2024

• minor cleanup -- whitespace, removing old commented out lines, etc.

  jeremy-syn

  committed
  d7c607c

  Copy full SHA for d7c607c

• fixed error - val set was using target words from training set

  jeremy-syn

  committed
  870ac09

  Copy full SHA for 870ac09

• minor cleanup -- whitespace, removing old commented out lines, etc.

  jeremy-syn

  committed
  8e9a0ee

  Copy full SHA for 8e9a0ee

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  264eae7

  Copy full SHA for 264eae7

• changed ordering in data prep, now shuffle before batching

  jeremy-syn

  committed
  6a3b15e

  Copy full SHA for 6a3b15e

• adding current version of trained and quantized streaming ww model

  jeremy-syn

  committed
  76413de

  Copy full SHA for 76413de

Commits on Jun 7, 2024

• minor edits/cleanup

  jeremy-syn

  committed
  dec0abb

  Copy full SHA for dec0abb

• changed Flags.num_train_samples to num_samples_training. same for test, validation. refactoring get_dataset code

  jeremy-syn

  committed
  b11d123

  Copy full SHA for b11d123

Commits on Jun 8, 2024

• added 1st pass at get_data_config(), refactoring dataset build

  jeremy-syn

  committed
  b1b1204

  Copy full SHA for b1b1204

Commits on Jun 9, 2024

• refactored dataset building. train.py runs now, have not tested performance

  jeremy-syn

  committed
  d313d65

  Copy full SHA for d313d65

Commits on Jun 10, 2024

• setup_example is work in progress, just capturing progress

  jeremy-syn

  committed
  c4c654b

  Copy full SHA for c4c654b

• train.py runs but gives random-level validation accuracy. demo notebook fails

  jeremy-syn

  committed
  a6028cf

  Copy full SHA for a6028cf

Commits on Jun 14, 2024

• flag parsing used 'train' instead of 'training' and therefore was not shuffling the training set

  jeremy-syn

  committed
  d6e23bb

  Copy full SHA for d6e23bb

Commits on Jun 15, 2024

• updated demo to match changes in data

  jeremy-syn

  committed
  de7a198

  Copy full SHA for de7a198

Commits on Jun 22, 2024

• minor updates

  jeremy-syn

  committed
  af713d8

  Copy full SHA for af713d8

• dumps options as json into plot_dir

  jeremy-syn

  committed
  b50d8fc

  Copy full SHA for b50d8fc

• fixed demo to work with new get_data code. moved take after shuffle so subsets are correctly mixed. but makes even runs on a small subset of training data slow, because shuffle runs on everything

  jeremy-syn

  committed
  7168f50

  Copy full SHA for 7168f50

Commits on Jun 24, 2024

• moved softmax to inside the model; adjusted loss function accordingly

  jeremy-syn

  committed
  859b3b3

  Copy full SHA for 859b3b3

Commits on Jun 29, 2024

• moved softmax calculation into the model

  jeremy-syn

  committed
  36cd5b8

  Copy full SHA for 36cd5b8

• working on true pos/false pos computation

  jeremy-syn

  committed
  ea3eb2e

  Copy full SHA for ea3eb2e

• resolved merge

  jeremy-syn

  committed
  c7d2f1c

  Copy full SHA for c7d2f1c

Commits on Jun 30, 2024

• fixed error, post-wakeword extension was being added twice

  jeremy-syn

  committed
  35d4394

  Copy full SHA for 35d4394

Commits on Jul 1, 2024

• fixing notebook counting of true/false positives

  jeremy-syn

  committed
  9184f66

  Copy full SHA for 9184f66

• removed commented-out code; added zero2nan()

  jeremy-syn

  committed
  8e671d0

  Copy full SHA for 8e671d0

• added multiple background noise paths, can split long bg files into smaller chunks, added clip_len_samples arg to prepare_bg_data

  jeremy-syn

  committed
  a1b5673

  Copy full SHA for a1b5673

• change QAT initial LR to Flags.lr, LR is too small after float pre-training

  jeremy-syn

  committed
  37ca39d

  Copy full SHA for 37ca39d

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  7f14c47

  Copy full SHA for 7f14c47

• fixed cmd line arg processing to accomodate multile bg noise paths

  jeremy-syn

  committed
  811db67

  Copy full SHA for 811db67

Commits on Jul 2, 2024

• removed commented out code from demo notebook

  jeremy-syn

  committed
  53bb590

  Copy full SHA for 53bb590

Commits on Jul 7, 2024

• convert only-target dataset to numpy array and back so cardinality() works

  jeremy-syn

  committed
  e1d75f9

  Copy full SHA for e1d75f9

Commits on Jul 9, 2024

• refactored num_silent, num_repeats in to fraction_silent and fraction_target to make varied-length experiments work better

  jeremy-syn

  committed
  3eb0329

  Copy full SHA for 3eb0329

• fixed cmd line arg processing to accomodate multile bg noise paths

  jeremy-syn

  committed
  ec0d024

  Copy full SHA for ec0d024

Commits on Jul 16, 2024

• fixing code for smaller datasets

  jeremy-syn

  committed
  5ecfeec

  Copy full SHA for 5ecfeec

• catching up on demo edits

  jeremy-syn

  committed
  6a2d715

  Copy full SHA for 6a2d715

• added code to run quantized model on long waveform

  jeremy-syn

  committed
  a2a5f8b

  Copy full SHA for a2a5f8b

• working on long wav file creation; added poisson process to place wake words

  jeremy-syn

  committed
  45d0b72

  Copy full SHA for 45d0b72

• updated long wave creation, need to move it to a separate file soon. moved get_true_and_false_detections() to util

  jeremy-syn

  committed
  609f901

  Copy full SHA for 609f901

• increased number of background files from 50 to 100

  jeremy-syn

  committed
  d059f4a

  Copy full SHA for d059f4a

Commits on Jul 18, 2024

• added code to illustrate false detects/rejects

  jeremy-syn

  committed
  9d11002

  Copy full SHA for 9d11002

• updating background noise creation to avoid train/val duplicates

  jeremy-syn

  committed
  95dda3e

  Copy full SHA for 95dda3e

• added exclude_background_files.txt

  jeremy-syn

  committed
  3473dbb

  Copy full SHA for 3473dbb

Commits on Jul 19, 2024

• put code to build the long test wav into its own (two) files

  jeremy-syn

  committed
  5114246

  Copy full SHA for 5114246

Commits on Jul 20, 2024

• added eval_long_wav.py to test fpr, fnr on a long wav

  jeremy-syn

  committed
  ccec9f4

  Copy full SHA for ccec9f4

• made build_long_wav work with the musan_path from streaming_config.json

  jeremy-syn

  committed
  6910b1c

  Copy full SHA for 6910b1c

• made build_long_wav work with the musan_path from streaming_config.json

  jeremy-syn

  committed
  04b0608

  Copy full SHA for 04b0608

• fixed a typo

  jeremy-syn

  committed
  c599598

  Copy full SHA for c599598

• fixed issue with path construction in long wav spec

  jeremy-syn

  committed
  0dced76

  Copy full SHA for 0dced76

Commits on Jul 21, 2024

• added l2 reg to conv layers

  jeremy-syn

  committed
  223f315

  Copy full SHA for 223f315

• added L2 reg to conv layers

  jeremy-syn

  committed
  76de629

  Copy full SHA for 76de629

• removed some old commented-out code

  jeremy-syn

  committed
  2c77d4e

  Copy full SHA for 2c77d4e

• eval_long_wav can now test either h5 models or tflite models

  jeremy-syn

  committed
  1d1a112

  Copy full SHA for 1d1a112

• Merge branch 'streaming_ww_dev' of gh-syn:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  c6332f2

  Copy full SHA for c6332f2

• added script to create indices into the val set for calibration

  jeremy-syn

  committed
  7724e98

  Copy full SHA for 7724e98

Commits on Jul 22, 2024

• code to create calibration set is working

  jeremy-syn

  committed
  05cf6d3

  Copy full SHA for 05cf6d3

• fixed quantize.py to work with extracted npz calibration set

  jeremy-syn

  committed
  7cbcd9f

  Copy full SHA for 7cbcd9f

• adjusted volume of foreground and background for testing

  jeremy-syn

  committed
  df1d3e8

  Copy full SHA for df1d3e8

• added code to save spectrogram in build_long_wav.py

  jeremy-syn

  committed
  d341993

  Copy full SHA for d341993

Commits on Jul 23, 2024

• demo notebook should now work with current code

  jeremy-syn

  committed
  e30cbd3

  Copy full SHA for e30cbd3

Commits on Jul 25, 2024

• separated augmentation (built by get_augment_wavs_func()) and feature extractor (from get_lfbe_func())

  jeremy-syn

  committed
  0fba129

  Copy full SHA for 0fba129

• made l2 reg parameter a commmand line flag

  jeremy-syn

  committed
  40cbecf

  Copy full SHA for 40cbecf

• fixed eval_long_wav to work with feature extractor changes

  jeremy-syn

  committed
  6c6c330

  Copy full SHA for 6c6c330

• added validation set measurments to eval_long_wav.py

  jeremy-syn

  committed
  f480533

  Copy full SHA for f480533

• moved eval_long_wav to evaluate.py

  jeremy-syn

  committed
  8cfc991

  Copy full SHA for 8cfc991

• added threshold=0.95 to precision/recall metrics to match evaluate.py

  jeremy-syn

  committed
  7801dd2

  Copy full SHA for 7801dd2

Commits on Jul 26, 2024

• added a list of 'bad' marvin wav files. modified build_long_wav_spec and get_dataset to exclude them

  jeremy-syn

  committed
  23e4434

  Copy full SHA for 23e4434

• edited comment on saved_model_path to reflect evaluate.py

  jeremy-syn

  committed
  2672489

  Copy full SHA for 2672489

• added bad_marvin_files.txt

  jeremy-syn

  committed
  d0f30f8

  Copy full SHA for d0f30f8

Commits on Jul 27, 2024

• fixed error in number of unknown samples for reduced runs

  jeremy-syn

  committed
  6d8cb29

  Copy full SHA for 6d8cb29

• renamed build_long_wav_spec.py -> build_long_wav_def.py to avoid ambiguity of spec (specification vs spectrogram)

  jeremy-syn

  committed
  ba80cc1

  Copy full SHA for ba80cc1

Commits on Jul 28, 2024

• renamed features back to audio to allow easy skipping of feature extraction

  jeremy-syn

  committed
  d626e61

  Copy full SHA for d626e61

Commits on Jul 29, 2024

• minor edits

  jeremy-syn

  committed
  8eb1f6c

  Copy full SHA for 8eb1f6c

• removed debug print statement

  jeremy-syn

  committed
  5960e11

  Copy full SHA for 5960e11

• fixed code for tflite models

  jeremy-syn

  committed
  b0c424a

  Copy full SHA for b0c424a

• adjusted some default training params

  jeremy-syn

  committed
  0ba2b18

  Copy full SHA for 0ba2b18

• catching notebook up to other code

  jeremy-syn

  committed
  547569b

  Copy full SHA for 547569b

• clearing out some debug prints

  jeremy-syn

  committed
  6341ee8

  Copy full SHA for 6341ee8

• added trained model

  jeremy-syn

  committed
  5a16bb9

  Copy full SHA for 5a16bb9

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  fad10d4

  Copy full SHA for fad10d4

• moved label_count out of model_settings into a flag

  jeremy-syn

  committed
  0c0c89f

  Copy full SHA for 0c0c89f

• Merge branch 'streaming_ww_dev' of gh-syn:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  a348974

  Copy full SHA for a348974

• minor edits

  jeremy-syn

  committed
  a707678

  Copy full SHA for a707678

Commits on Aug 4, 2024

• added random timeshift

  jeremy-syn

  committed
  6836643

  Copy full SHA for 6836643

• added a couple more bad marvins to exclude

  jeremy-syn

  committed
  143dd8e

  Copy full SHA for 143dd8e

• added random timeshifting to augmentation function

  jeremy-syn

  committed
  2461453

  Copy full SHA for 2461453

Commits on Aug 5, 2024

• added flag to enforce a minimum SNR

  jeremy-syn

  committed
  9c89eb7

  Copy full SHA for 9c89eb7

• centralized data paths in streaming_config.json (no command line arguments needed or observed)

  jeremy-syn

  committed
  c91b792

  Copy full SHA for c91b792

• removed some obsolete cmd line args and modified get_dataset to respect time_shift_ms

  jeremy-syn

  committed
  c02b502

  Copy full SHA for c02b502

• fixed evaulate.py to work with changes on speech_commands_path

  jeremy-syn

  committed
  0a88769

  Copy full SHA for 0a88769

• changed evaluate and quantize to use model_init_path, so by default the reference model will still be used

  jeremy-syn

  committed
  f0cd836

  Copy full SHA for f0cd836

• adjusted trainign params

  jeremy-syn

  committed
  61153a4

  Copy full SHA for 61153a4

• adjusted trainign params

  jeremy-syn

  committed
  42d51a0

  Copy full SHA for 42d51a0

• updated long wav info

  jeremy-syn

  committed
  9e9e2ad

  Copy full SHA for 9e9e2ad

• updated README

  jeremy-syn

  committed
  2258d55

  Copy full SHA for 2258d55

• updated reference model

  jeremy-syn

  committed
  15a2d89

  Copy full SHA for 15a2d89

Commits on Aug 6, 2024

• removed some info messages

  jeremy-syn

  committed
  b49eb7d

  Copy full SHA for b49eb7d

Commits on Aug 9, 2024

• add line to create plot_dir if it does not exist

  jeremy-syn

  committed
  9c655d2

  Copy full SHA for 9c655d2

• reduced noise level in long wav

  jeremy-syn

  committed
  b9cd142

  Copy full SHA for b9cd142

Commits on Aug 11, 2024

• refactored command line argument parsing

  jeremy-syn

  committed
  54de3e9

  Copy full SHA for 54de3e9

Commits on Aug 12, 2024

• refactored command line argument parsing

  jeremy-syn

  committed
  c354ac6

  Copy full SHA for c354ac6

• Merge branch 'streaming_ww_dev' of github.com:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  7e69344

  Copy full SHA for 7e69344

• fixed some errors in README

  jeremy-syn

  committed
  478f6ae

  Copy full SHA for 478f6ae

• fixed quantize to use saved_model_path instead of model_init_path

  jeremy-syn

  committed
  7d97168

  Copy full SHA for 7d97168

• added calibration_samples.npz

  jeremy-syn

  committed
  bd96237

  Copy full SHA for bd96237

• fixing argument processing for evaluate.py to work with either tflite or h5 model

  jeremy-syn

  committed
  2bc1175

  Copy full SHA for 2bc1175

• fixed typo in evaluate.py

  jeremy-syn

  committed
  8e7b718

  Copy full SHA for 8e7b718

• fixed typo in evaluate

  jeremy-syn

  committed
  8c63415

  Copy full SHA for 8c63415

• fixing merge

  jeremy-syn

  committed
  7608316

  Copy full SHA for 7608316

• updated tflite model

  jeremy-syn

  committed
  30308c9

  Copy full SHA for 30308c9

• fixed issue with plot_dir

  jeremy-syn

  committed
  1fea3a0

  Copy full SHA for 1fea3a0

• Merge branch 'streaming_ww_dev' of gh-syn:mlcommons/tiny into streaming_ww_dev

  jeremy-syn

  committed
  0c73cf8

  Copy full SHA for 0c73cf8

• ignoring trained models other than reference model

  jeremy-syn

  committed
  fb4e6a2

  Copy full SHA for fb4e6a2

• updated readme

  jeremy-syn

  committed
  68f53b8

  Copy full SHA for 68f53b8

• updated demo notebook

  jeremy-syn

  committed
  02953c9

  Copy full SHA for 02953c9

• added note about the demo notebook to readme

  jeremy-syn

  committed
  c280e5d

  Copy full SHA for c280e5d

• merged gitignore from master branch

  jeremy-syn

  committed
  900e126

  Copy full SHA for 900e126

Commits on Sep 9, 2024

• Merge branch 'master' into streaming_ww_dev

  Peter-Chang

  authored
  f9ec337

  Copy full SHA for f9ec337

Commits on Jan 19, 2025

• merged work from runner_dev_jeremy

  jeremy-syn

  committed
  5cfabb6

  Copy full SHA for 5cfabb6

• Merge branch 'master' into streaming_ww_dev

  jeremy-syn

  authored
  50104d0

  Copy full SHA for 50104d0