excalibur12's picture
End of training
ce05036 verified
|
raw
history blame
2.42 kB
metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
  - generated_from_trainer
model-index:
  - name: cmb-20s_asr-scr_w2v2-base_003
    results: []

cmb-20s_asr-scr_w2v2-base_003

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1978
  • Per: 0.1287
  • Pcc: 0.6493
  • Ctc Loss: 0.4014
  • Mse Loss: 0.9499

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: 1e-05
  • train_batch_size: 16
  • eval_batch_size: 1
  • seed: 3333
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 8928
  • training_steps: 89280
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Per Pcc Ctc Loss Mse Loss
11.3245 3.0 8928 4.4416 0.9956 0.6159 3.7640 0.8975
2.9636 6.0 17856 1.4930 0.1745 0.6638 0.6280 0.8352
1.0327 9.0 26784 1.3313 0.1461 0.6666 0.4797 0.8799
0.5448 12.0 35712 1.2880 0.1394 0.6530 0.4501 0.9425
0.1232 15.0 44640 1.0484 0.1354 0.6481 0.4289 0.8871
-0.3248 18.0 53568 1.3777 0.1330 0.6373 0.4163 1.1622
-0.7634 21.0 62496 1.0371 0.1312 0.6499 0.4094 1.0824
-1.2089 24.0 71424 0.4166 0.1298 0.6454 0.4060 0.9053
-1.613 27.0 80352 0.2751 0.1290 0.6473 0.4021 0.9426
-1.8704 30.0 89280 0.1978 0.1287 0.6493 0.4014 0.9499

Framework versions

  • Transformers 4.38.1
  • Pytorch 2.0.1
  • Datasets 2.18.0
  • Tokenizers 0.15.2