metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - common_voice_8_0
metrics:
  - wer
model-index:
  - name: wav2vec2-large-xls-r-1b-frisian-cv-8
    results:
      - task:
          name: Automatic Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: common_voice_8_0
          type: common_voice_8_0
          config: fy-NL
          split: validation
          args: fy-NL
        metrics:
          - name: Wer
            type: wer
            value: 0.15637051849735753

wav2vec2-large-xls-r-1b-frisian-cv-8

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the common_voice_8_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.2219
Wer: 0.1564

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: 7e-05
train_batch_size: 30
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 50
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.9356	2.42	300	3.0022	1.0
1.7278	4.84	600	0.4414	0.4147
0.9407	7.26	900	0.3058	0.2955
0.943	9.68	1200	0.2678	0.2530
0.7468	12.1	1500	0.2443	0.2237
0.6009	14.52	1800	0.2381	0.2097
0.6101	16.94	2100	0.2339	0.2003
0.5646	19.35	2400	0.2357	0.2047
0.5875	21.77	2700	0.2219	0.1914
0.5245	24.19	3000	0.2525	0.1807
0.5971	26.61	3300	0.2432	0.1784
0.563	29.03	3600	0.2454	0.1753
0.4441	31.45	3900	0.2237	0.1776
0.5552	33.87	4200	0.2313	0.1629
0.5568	36.29	4500	0.2318	0.1602
0.4342	38.71	4800	0.2324	0.1556
0.4723	41.13	5100	0.2296	0.1602
0.3357	43.55	5400	0.2267	0.1575
0.4588	45.97	5700	0.2243	0.1558
0.4594	48.39	6000	0.2219	0.1564

Framework versions

Transformers 4.28.1
Pytorch 2.0.0+cu117
Datasets 2.11.0
Tokenizers 0.13.3