Edit model card

BioMedNLP_DeBERTa_all_updates

This model is a fine-tuned version of hongpingjun98/BioMedNLP_DeBERTa on the sem_eval_2024_task_2 dataset. It achieves the following results on the evaluation set:

  • Loss: 2.1863
  • Accuracy: 0.705
  • Precision: 0.7238
  • Recall: 0.7050
  • F1: 0.6987

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: 16
  • eval_batch_size: 16
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 20
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1
0.4238 1.0 116 0.6639 0.665 0.6678 0.665 0.6636
0.4316 2.0 232 0.6644 0.68 0.6875 0.6800 0.6768
0.3819 3.0 348 0.7328 0.71 0.7188 0.71 0.7071
0.3243 4.0 464 0.9162 0.7 0.7083 0.7 0.6970
0.4053 5.0 580 0.7145 0.715 0.7214 0.7150 0.7129
0.2548 6.0 696 1.0598 0.69 0.7016 0.69 0.6855
0.3455 7.0 812 0.7782 0.72 0.7232 0.72 0.7190
0.2177 8.0 928 1.1182 0.69 0.6950 0.69 0.6880
0.2304 9.0 1044 1.4332 0.695 0.708 0.695 0.6902
0.2103 10.0 1160 1.2736 0.7 0.7198 0.7 0.6931
0.1748 11.0 1276 1.2654 0.675 0.6816 0.675 0.6720
0.1608 12.0 1392 1.8885 0.63 0.6689 0.63 0.6074
0.1082 13.0 1508 1.7004 0.68 0.7005 0.6800 0.6716
0.1074 14.0 1624 1.8145 0.67 0.6804 0.67 0.6652
0.0238 15.0 1740 1.7608 0.68 0.6931 0.68 0.6745
0.038 16.0 1856 1.9937 0.67 0.6953 0.6700 0.6589
0.0365 17.0 1972 2.1871 0.675 0.6964 0.675 0.6659
0.0144 18.0 2088 2.1093 0.695 0.7059 0.6950 0.6909
0.0014 19.0 2204 2.1559 0.695 0.7103 0.6950 0.6893
0.0324 20.0 2320 2.1863 0.705 0.7238 0.7050 0.6987

Framework versions

  • Transformers 4.35.2
  • Pytorch 2.1.0+cu121
  • Datasets 2.16.1
  • Tokenizers 0.15.0
Downloads last month
13
Safetensors
Model size
108M params
Tensor type
F32
·

Finetuned from

Evaluation results