wav2vec2-pretraining-demo

This model is a fine-tuned version of on the librispeech_asr dataset. It achieves the following results on the evaluation set:

  • Loss: nan
  • Constrast Loss: 708.2256
  • Div Loss: nan

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • 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 Constrast Loss Div Loss
343.7458 0.3745 100 nan 706.8953 nan
360.8719 0.7491 200 nan 709.1356 nan
561.7222 1.1236 300 nan 725.8750 nan
450.0023 1.4981 400 nan 704.0518 nan
412.6041 1.8727 500 nan 702.4941 nan
955.4866 2.2472 600 nan 717.8548 nan
947.9007 2.6217 700 775.8956 763.6932 122.0238
487.9696 2.9963 800 nan 706.2196 nan
359.6353 3.3708 900 nan 727.8684 nan
299.9773 3.7453 1000 nan 689.3754 nan
557.0656 4.1199 1100 nan 741.6811 nan
625.427 4.4944 1200 nan 669.9859 nan
486.9457 4.8689 1300 nan 725.6080 nan
1067.3776 5.2434 1400 nan 703.2175 nan
958.6181 5.6180 1500 nan 742.9335 nan
482.8976 5.9925 1600 nan 724.2250 nan
350.0361 6.3670 1700 nan 715.3557 nan
350.6836 6.7416 1800 nan 711.0261 nan
522.1924 7.1161 1900 nan 702.5721 nan
483.4517 7.4906 2000 nan 741.1333 nan
590.2781 7.8652 2100 nan 728.1261 nan
1039.0322 8.2397 2200 nan 720.8858 nan
1146.0051 8.6142 2300 nan 717.9496 nan
626.2248 8.9888 2400 nan 712.2637 nan
371.0907 9.3633 2500 nan 727.3933 nan
362.5648 9.7378 2600 nan 708.2256 nan

Framework versions

  • Transformers 4.49.0.dev0
  • Pytorch 2.6.0+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0
Downloads last month
47
Safetensors
Model size
95M params
Tensor type
F32
·
Inference Providers NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API: The model has no pipeline_tag.

Dataset used to train ekunish/wav2vec2-pretraining-demo