tapt_reinit_only_4K_no_reinit_classifier_llrd0.9_LR-2e-05

This model is a fine-tuned version of microsoft/BiomedNLP-BiomedBERT-base-uncased-abstract-fulltext on the Mardiyyah/TAPT_data_V2_split dataset. It achieves the following results on the evaluation set:

  • Loss: 1.6722

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: 2e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 3407
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-06 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.06
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss
1.8199 1.0 609 1.6390
1.6303 2.0 1218 1.6489
1.5042 3.0 1827 1.5989
1.4113 4.0 2436 1.6950
1.3254 5.0 3045 1.6834
1.2811 6.0 3654 1.7477
1.2411 7.0 4263 1.7607
1.1898 8.0 4872 1.7285
1.146 9.0 5481 1.7246
1.1379 10.0 6090 1.7101
1.1208 11.0 6699 1.6693
1.0634 12.0 7308 1.7053
1.0582 13.0 7917 1.6786
1.0313 14.0 8526 1.7939
1.0062 15.0 9135 1.7675
0.9965 16.0 9744 1.7152
0.9845 17.0 10353 1.6762
0.9914 18.0 10962 1.7251
0.9665 19.0 11571 1.7469
0.9734 20.0 12180 1.7760

Framework versions

  • Transformers 4.48.2
  • Pytorch 2.4.1+cu121
  • Datasets 3.0.2
  • Tokenizers 0.21.0
Downloads last month
15
Safetensors
Model size
0.1B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for Mardiyyah/tapt_reinit_only_4K_no_reinit_classifier_llrd0.9_LR-2e-05