metadata

license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: wav2vec_arabic_mdd
    results: []

wav2vec_arabic_mdd

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

Loss: 0.3192
Wer: 0.0429
Per: 0.0326

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
train_batch_size: 2
eval_batch_size: 2
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Per
8.6256	1.0	1617	3.2158	1.0	1.0
1.3467	2.0	3234	0.3775	0.0753	0.0605
0.1996	3.0	4851	0.3054	0.0561	0.0421
0.1311	4.0	6468	0.3120	0.0507	0.0387
0.1044	5.0	8085	0.3169	0.0507	0.0386
0.0891	6.0	9702	0.3030	0.0496	0.0372
0.0788	7.0	11319	0.3174	0.0520	0.0391
0.0619	8.0	12936	0.3312	0.0546	0.0417
0.0552	9.0	14553	0.3353	0.0480	0.0358
0.0493	10.0	16170	0.2742	0.0452	0.0336
0.0417	11.0	17787	0.2894	0.0459	0.0355
0.0382	12.0	19404	0.2949	0.0463	0.0350
0.0306	13.0	21021	0.3041	0.0472	0.0348
0.03	14.0	22638	0.3109	0.0453	0.0339
0.0244	15.0	24255	0.3234	0.0448	0.0335
0.0215	16.0	25872	0.3321	0.0436	0.0332
0.0182	17.0	27489	0.3367	0.0420	0.0315
0.0171	18.0	29106	0.3317	0.0430	0.0328
0.0155	19.0	30723	0.3249	0.0423	0.0321
0.0132	20.0	32340	0.3192	0.0429	0.0326

Framework versions

Transformers 4.34.1
Pytorch 2.1.0+cu118
Datasets 2.14.6
Tokenizers 0.14.1