bdpc's picture
Saving best model of SciBERT_TwoWayLoss_25K_bs64_P10_N5 to hub
6fb0f12 verified
metadata
base_model: allenai/scibert_scivocab_uncased
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - precision
  - recall
  - f1
model-index:
  - name: SciBERT_TwoWayLoss_25K_bs64_P10_N5
    results: []

SciBERT_TwoWayLoss_25K_bs64_P10_N5

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

  • Loss: 15.1250
  • Accuracy: 0.7066
  • Precision: 0.0321
  • Recall: 0.9982
  • F1: 0.0622
  • Hamming: 0.2934

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: 64
  • eval_batch_size: 64
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • training_steps: 25000

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1 Hamming
28.5732 0.16 5000 26.4288 0.6945 0.0307 0.9910 0.0595 0.3055
19.8755 0.32 10000 18.9620 0.7010 0.0315 0.9959 0.0610 0.2990
17.1294 0.47 15000 16.5587 0.7021 0.0316 0.9970 0.0613 0.2979
15.8209 0.63 20000 15.4919 0.7053 0.0320 0.9982 0.0620 0.2947
15.4304 0.79 25000 15.1250 0.7066 0.0321 0.9982 0.0622 0.2934

Framework versions

  • Transformers 4.33.3
  • Pytorch 2.2.0.dev20231002
  • Datasets 2.7.1
  • Tokenizers 0.13.3