Edit model card

wav2vec2-large-xls-r-300m-Swedish

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

  • Loss: 0.3641
  • Wer: 0.2473
  • Cer: 0.0758

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 7.5e-05
  • train_batch_size: 64
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 1000
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer Cer
6.1097 5.49 500 3.1422 1.0 1.0
2.985 10.98 1000 1.7357 0.9876 0.4125
1.0363 16.48 1500 0.4773 0.3510 0.1047
0.6111 21.97 2000 0.3937 0.2998 0.0910
0.4942 27.47 2500 0.3779 0.2776 0.0844
0.4421 32.96 3000 0.3745 0.2630 0.0807
0.4018 38.46 3500 0.3685 0.2553 0.0781
0.3759 43.95 4000 0.3618 0.2488 0.0761
0.3646 49.45 4500 0.3641 0.2473 0.0758

Framework versions

  • Transformers 4.17.0.dev0
  • Pytorch 1.10.2+cu102
  • Datasets 1.18.2.dev0
  • Tokenizers 0.11.0
Downloads last month
2
Hosted inference API
or or
This model can be loaded on the Inference API on-demand.

Dataset used to train kingabzpro/wav2vec2-large-xls-r-300m-Swedish

Evaluation results