metadata

language:
  - 'no'
license: apache-2.0
base_model: NbAiLab/nb-whisper-medium-v0.8-vad3
tags:
  - audio
  - asr
  - automatic-speech-recognition
  - hf-asr-leaderboard
model-index:
  - name: nb-whisper-medium-v0.8-vad3-verbatim
    results: []

nb-whisper-medium-v0.8-vad3-verbatim

This model is a fine-tuned version of NbAiLab/nb-whisper-medium-v0.8-vad3 on the NbAiLab/NPSC dataset. It achieves the following results on the evaluation set:

step: 249
validation_loss: 0.6296
train_loss: 0.4324
validation_wer: 8.2769
validation_cer: 2.8193
validation_exact_wer: 8.4048
validation_exact_cer: 2.8363

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 2.5e-05
lr_scheduler_type: linear
per_device_train_batch_size: 32
total_train_batch_size_per_node: 128
total_train_batch_size: 1024
total_optimization_steps: 250
starting_optimization_step: None
finishing_optimization_step: 250
num_train_dataset_workers: 32
num_hosts: 8
total_num_training_examples: 256,000
steps_per_epoch: 45
num_beams: None
weight_decay: 0.01
adam_beta1: 0.9
adam_beta2: 0.98
adam_epsilon: 1e-06
dropout: True
bpe_dropout_probability: 0.2
activation_dropout_probability: 0.1

Training results

step	validation_loss	train_loss	validation_wer	validation_cer	validation_exact_wer	validation_exact_cer
0	1.5895	1.4606	17.5605	10.5650	33.0099	13.8415
40	0.6409	0.5035	9.1662	3.0250	9.3637	3.0542
80	0.6309	0.4790	8.7132	2.9755	8.8730	2.9952
120	0.6250	0.4480	8.4503	2.8812	8.6079	2.9019
160	0.6294	0.4423	8.4000	2.8641	8.5345	2.8810
200	0.6276	0.4467	8.3161	2.8345	8.4668	2.8534
240	0.6287	0.4376	8.2266	2.7917	8.3597	2.8087
249	0.6296	0.4324	8.2769	2.8193	8.4048	2.8363

Framework versions

Transformers 4.34.1
Datasets 2.16.1
Tokenizers 0.14.1