metadata
language:
- 'no'
license: apache-2.0
base_model: NbAiLabBeta/nb-whisper-base
tags:
- audio
- asr
- automatic-speech-recognition
- hf-asr-leaderboard
model-index:
- name: nb-whisper-base-v0.7-semantic
results: []
nb-whisper-base-v0.7-semantic
This model is a fine-tuned version of NbAiLabBeta/nb-whisper-base on the NbAiLab/ncc_speech_styling_v4 dataset. It achieves the following results on the evaluation set:
- step: 249
- validation_nst_loss: 0.7883
- train_loss: 0.8748
- validation_nst_wer: 5.3569
- validation_nst_cer: 1.6919
- validation_nst_exact_wer: 6.1244
- validation_nst_exact_cer: 1.8077
- validation_clean_stortinget_no_loss: 0.6909
- validation_clean_stortinget_no_wer: 13.2615
- validation_clean_stortinget_no_cer: 7.8992
- validation_clean_stortinget_no_exact_wer: 16.9470
- validation_clean_stortinget_no_exact_cer: 8.5431
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: 0.0001
- 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: To be computed after first epoch
- 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_nst_loss | train_loss | validation_nst_wer | validation_nst_cer | validation_nst_exact_wer | validation_nst_exact_cer | validation_clean_stortinget_no_loss | validation_clean_stortinget_no_wer | validation_clean_stortinget_no_cer | validation_clean_stortinget_no_exact_wer | validation_clean_stortinget_no_exact_cer |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0.4888 | 1.1200 | 5.6399 | 1.7618 | 6.4075 | 1.8792 | 0.6481 | 13.1075 | 7.4191 | 16.6481 | 8.0230 |
| 40 | 0.8222 | 1.0012 | 5.5202 | 1.7665 | 6.3422 | 1.8856 | 0.7037 | 13.3444 | 7.8759 | 16.8426 | 8.4830 |
| 80 | 0.8323 | 0.8735 | 5.6563 | 1.8085 | 6.4565 | 1.9277 | 0.7085 | 13.5505 | 8.1181 | 17.1629 | 8.7365 |
| 120 | 0.8210 | 0.8613 | 5.4331 | 1.6873 | 6.2388 | 1.8123 | 0.7040 | 13.3515 | 7.9877 | 17.0703 | 8.6261 |
| 160 | 0.8069 | 0.8800 | 5.4058 | 1.7516 | 6.1898 | 1.8673 | 0.6981 | 13.3325 | 8.0031 | 17.0395 | 8.6495 |
| 200 | 0.7985 | 0.8615 | 5.3569 | 1.6863 | 6.1571 | 1.8050 | 0.6937 | 13.2402 | 7.9655 | 16.9327 | 8.6074 |
| 240 | 0.7868 | 0.8575 | 5.3187 | 1.6705 | 6.1190 | 1.7913 | 0.6916 | 13.2473 | 7.9068 | 16.9399 | 8.5517 |
| 250 | 0.7883 | 0.8748 | 5.3569 | 1.6919 | 6.1244 | 1.8077 | 0.6909 | 13.2615 | 7.8992 | 16.9470 | 8.5431 |
Framework versions
- Transformers 4.34.1
- Datasets 2.15.0
- Tokenizers 0.14.1