README.md

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: 0.1339808598771604

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: 0.2054
• Wer: 0.1340

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: 3e-05
• train_batch_size: 32
• 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: 80
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.0483	1.72	200	3.0438	1.0
2.6284	3.45	400	1.1501	0.9229
1.4359	5.17	600	0.5618	0.5329
1.1366	6.9	800	0.3899	0.3845
0.988	8.62	1000	0.3370	0.3302
0.8377	10.34	1200	0.2765	0.2834
0.8001	12.07	1400	0.2750	0.2438
0.8678	13.79	1600	0.2258	0.2160
0.7023	15.52	1800	0.2260	0.2072
0.8111	17.24	2000	0.2223	0.2070
0.658	18.97	2200	0.2121	0.1834
0.6574	20.69	2400	0.2136	0.1812
0.7521	22.41	2600	0.2175	0.1775
0.6515	24.14	2800	0.2018	0.1718
0.6955	25.86	3000	0.2121	0.1863
0.6605	27.59	3200	0.2003	0.1607
0.5403	29.31	3400	0.2042	0.1668
0.5064	31.03	3600	0.2021	0.1616
0.6811	32.76	3800	0.2026	0.1668
0.6787	34.48	4000	0.2122	0.1613
0.5595	36.21	4200	0.2001	0.1547
0.5225	37.93	4400	0.1992	0.1615
0.5522	39.66	4600	0.2023	0.1603
0.5364	41.38	4800	0.1992	0.1531
0.5157	43.1	5000	0.2060	0.1550
0.4382	44.83	5200	0.1985	0.1427
0.3658	46.55	5400	0.1964	0.1427
0.5336	48.28	5600	0.2143	0.1471
0.5479	50.0	5800	0.1962	0.1402
0.5203	51.72	6000	0.2022	0.1418
0.363	53.45	6200	0.2103	0.1429
0.3828	55.17	6400	0.2070	0.1417
0.3875	56.9	6600	0.2070	0.1411
0.3433	58.62	6800	0.2049	0.1418
0.2826	60.34	7000	0.2047	0.1417
0.294	62.07	7200	0.2022	0.1369
0.2776	63.79	7400	0.2115	0.1365
0.3178	65.52	7600	0.2005	0.1377
0.2913	67.24	7800	0.2047	0.1355
0.2642	68.97	8000	0.2069	0.1338
0.255	70.69	8200	0.2041	0.1336
0.2746	72.41	8400	0.2064	0.1331
0.2485	74.14	8600	0.2068	0.1327
0.2741	75.86	8800	0.2073	0.1331
0.2223	77.59	9000	0.2055	0.1338
0.2327	79.31	9200	0.2054	0.1340

Framework versions

• Transformers 4.28.1
• Pytorch 2.0.0+cu117
• Datasets 2.11.0
• Tokenizers 0.13.3