metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - xtreme_s
metrics:
  - wer
base_model: facebook/wav2vec2-large-xlsr-53
model-index:
  - name: wav2vec2-XLS-R-Fleurs-demo-google-colab-Ezra_William_Prod7
    results:
      - task:
          type: automatic-speech-recognition
          name: Automatic Speech Recognition
        dataset:
          name: xtreme_s
          type: xtreme_s
          config: fleurs.id_id
          split: test
          args: fleurs.id_id
        metrics:
          - type: wer
            value: 0.5133213590779824
            name: Wer

wav2vec2-XLS-R-Fleurs-demo-google-colab-Ezra_William_Prod7

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the xtreme_s dataset. It achieves the following results on the evaluation set:

Loss: 1.1411
Wer: 0.5133

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: 0.001
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_steps: 600
num_epochs: 180
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.9813	18.18	300	2.8480	1.0
1.5729	36.36	600	0.8808	0.7159
0.219	54.55	900	0.9209	0.5983
0.1213	72.73	1200	0.9869	0.6005
0.0898	90.91	1500	1.0485	0.5840
0.0668	109.09	1800	1.0746	0.5514
0.0499	127.27	2100	1.0648	0.5341
0.0372	145.45	2400	1.1656	0.5280
0.0292	163.64	2700	1.1411	0.5133

Framework versions

Transformers 4.37.2
Pytorch 2.2.0+cu121
Datasets 2.16.1
Tokenizers 0.15.1