ManyaGupta's picture
End of training
380fe83 verified
metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - f1
  - precision
  - recall
model-index:
  - name: wav2vec2-base-ft-children-with-sli
    results: []

wav2vec2-base-ft-children-with-sli

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

  • Loss: 0.1079
  • Accuracy: 0.9722
  • F1: 0.9712
  • Precision: 0.9670
  • Recall: 0.9771

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: 5e-05
  • train_batch_size: 32
  • eval_batch_size: 32
  • 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: 10
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy F1 Precision Recall
0.6402 1.0 32 0.6563 0.6071 0.3778 0.3036 0.5
0.446 2.0 64 0.6949 0.5357 0.5048 0.7292 0.6176
0.3454 3.0 96 0.2953 0.8690 0.8677 0.8708 0.8886
0.0703 4.0 128 0.1439 0.9563 0.9548 0.9507 0.9605
0.0458 5.0 160 0.0759 0.9841 0.9834 0.9834 0.9834
0.0844 6.0 192 1.0983 0.7421 0.7416 0.8018 0.7876
0.1241 7.0 224 0.1313 0.9683 0.9672 0.9626 0.9739
0.0454 8.0 256 0.0824 0.9802 0.9794 0.9760 0.9837
0.0403 9.0 288 0.0715 0.9802 0.9794 0.9760 0.9837
0.0241 10.0 320 0.1079 0.9722 0.9712 0.9670 0.9771

Framework versions

  • Transformers 4.42.0.dev0
  • Pytorch 2.3.0+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1