metadata

language:
  - lv
license: apache-2.0
base_model: arturslogins/whisper-medium-lv
tags:
  - hf-asr-leaderboard
  - generated_from_trainer
datasets:
  - ta-dataset/training
metrics:
  - wer
model-index:
  - name: Whisper medium LV - Arturs Logins
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Recorded Voice
          type: ta-dataset/training
          config: lv
          split: test
          args: 'config: lv, split: test'
        metrics:
          - name: Wer
            type: wer
            value: 43.705463182897866

Whisper medium LV - Arturs Logins

This model is a fine-tuned version of arturslogins/whisper-medium-lv on the Recorded Voice dataset. It achieves the following results on the evaluation set:

Loss: 1.0843
Wer: 43.7055

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: 1e-05
train_batch_size: 16
eval_batch_size: 12
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
training_steps: 5000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0097	6.3898	1000	0.9780	45.8432
0.0023	12.7796	2000	1.0158	44.9525
0.0002	19.1693	3000	1.0549	46.1758
0.0001	25.5591	4000	1.0765	44.9881
0.0001	31.9489	5000	1.0843	43.7055

Framework versions

Transformers 4.41.0.dev0
Pytorch 2.0.1
Datasets 2.19.0
Tokenizers 0.19.1