metadata

base_model: facebook/wav2vec2-xls-r-300m
datasets:
  - fleurs
library_name: transformers
license: apache-2.0
metrics:
  - wer
tags:
  - generated_from_trainer
model-index:
  - name: wav2vec2-xls-r-Wolof-20-hours-alffa-plus-fleurs-dataset
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: fleurs
          type: fleurs
          config: wo_sn
          split: None
          args: wo_sn
        metrics:
          - type: wer
            value: 0.423307335820052
            name: Wer

wav2vec2-xls-r-Wolof-20-hours-alffa-plus-fleurs-dataset

This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the fleurs dataset. It achieves the following results on the evaluation set:

Loss: 1.2175
Wer: 0.4233
Cer: 0.1476

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
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_ratio: 0.1
num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.7783	1.2365	400	1.5400	0.8840	0.3321
0.9198	2.4730	800	0.8296	0.6037	0.2088
0.6527	3.7094	1200	0.8048	0.5326	0.1865
0.5793	4.9459	1600	0.7323	0.5155	0.1849
0.5234	6.1824	2000	0.6832	0.5016	0.1782
0.4583	7.4189	2400	0.7613	0.4925	0.1770
0.4251	8.6553	2800	0.6324	0.4951	0.1750
0.3763	9.8918	3200	0.7067	0.4810	0.1697
0.3503	11.1283	3600	0.7494	0.5030	0.1767
0.325	12.3648	4000	0.6757	0.4997	0.1734
0.288	13.6012	4400	0.6880	0.4851	0.1679
0.2673	14.8377	4800	0.6832	0.5268	0.1930
0.2485	16.0742	5200	0.6879	0.4730	0.1726
0.2229	17.3107	5600	0.7245	0.4907	0.1727
0.2088	18.5471	6000	0.7458	0.4642	0.1626
0.1903	19.7836	6400	0.7750	0.4572	0.1610
0.1767	21.0201	6800	0.8254	0.4550	0.1606
0.1613	22.2566	7200	0.7481	0.4613	0.1600
0.1554	23.4930	7600	0.8343	0.4425	0.1573
0.1429	24.7295	8000	0.8984	0.4465	0.1590
0.1384	25.9660	8400	0.7643	0.4624	0.1649
0.1295	27.2025	8800	0.8534	0.4489	0.1568
0.1178	28.4389	9200	0.9115	0.4481	0.1587
0.1146	29.6754	9600	0.8835	0.4403	0.1565
0.1033	30.9119	10000	0.9292	0.4525	0.1575
0.0957	32.1484	10400	1.0917	0.4407	0.1526
0.0944	33.3849	10800	0.9554	0.4474	0.1573
0.0878	34.6213	11200	1.0647	0.4385	0.1531
0.0828	35.8578	11600	1.0365	0.4356	0.1512
0.0821	37.0943	12000	1.0613	0.4461	0.1571
0.0736	38.3308	12400	1.0852	0.4333	0.1521
0.0701	39.5672	12800	1.0551	0.4334	0.1518
0.0677	40.8037	13200	1.1009	0.4327	0.1500
0.0645	42.0402	13600	1.1279	0.4277	0.1490
0.0607	43.2767	14000	1.1581	0.4214	0.1483
0.0593	44.5131	14400	1.2038	0.4231	0.1473
0.0557	45.7496	14800	1.1791	0.4249	0.1525
0.0554	46.9861	15200	1.2012	0.4248	0.1484
0.0543	48.2226	15600	1.2087	0.4232	0.1471
0.0525	49.4590	16000	1.2175	0.4233	0.1476

Framework versions

Transformers 4.44.1
Pytorch 2.1.0+cu118
Datasets 2.17.0
Tokenizers 0.19.1