flozi00's picture
update model card README.md
fdc04c0
metadata
language:
  - de
tags:
  - automatic-speech-recognition
  - mozilla-foundation/common_voice_10_0
  - generated_from_trainer
model-index:
  - name: wav2vec2-xls-r-1b-german
    results: []

wav2vec2-xls-r-1b-german

This model is a fine-tuned version of wav2vec2-xls-r-1b-german on the MOZILLA-FOUNDATION/COMMON_VOICE_10_0 - DE dataset. It achieves the following results on the evaluation set:

  • Loss: 0.3513
  • Wer: 0.2931

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-07
  • train_batch_size: 48
  • eval_batch_size: 48
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 96
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 2.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
0.2009 1.0 4816 0.3486 0.2940
0.1637 2.0 9632 0.3459 0.2915

Framework versions

  • Transformers 4.21.3
  • Pytorch 1.12.1
  • Datasets 2.4.0
  • Tokenizers 0.12.1