KasuleTrevor's picture
End of training
4c95b98 verified
|
raw
history blame
2.65 kB
metadata
license: apache-2.0
base_model: facebook/wav2vec2-large
tags:
  - generated_from_trainer
metrics:
  - wer
model-index:
  - name: wav2vec2-large-sw-cv-100hr-v3
    results: []

wav2vec2-large-sw-cv-100hr-v3

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

  • Loss: 2.8721
  • Model Preparation Time: 0.0042
  • Wer: 0.9997
  • Cer: 0.9176

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: 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: 120
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Model Preparation Time Wer Cer
4.0159 0.9998 2079 2.7894 0.0042 0.9999 0.9048
2.8122 2.0 4159 2.8360 0.0042 1.0 1.0
2.8312 2.9998 6238 2.8331 0.0042 1.0 1.0
2.8295 4.0 8318 2.8340 0.0042 1.0 1.0
2.8306 4.9998 10397 2.8309 0.0042 1.0 1.0
2.8301 6.0 12477 2.8451 0.0042 1.0 1.0
2.8305 6.9998 14556 2.8320 0.0042 1.0 1.0
2.8547 8.0 16636 2.8626 0.0042 1.0 1.0
2.8611 8.9998 18715 2.8569 0.0042 1.0 1.0
2.8596 10.0 20795 2.8573 0.0042 1.0 1.0
2.861 10.9998 22874 2.8570 0.0042 1.0 1.0

Framework versions

  • Transformers 4.44.0
  • Pytorch 2.1.0+cu118
  • Datasets 2.20.0
  • Tokenizers 0.19.1