albert-base-ours-run-4

This model is a fine-tuned version of albert-base-v2 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 1.9565
  • Accuracy: 0.72
  • Precision: 0.6790
  • Recall: 0.6770
  • F1: 0.6766

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: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1
1.0253 1.0 200 0.8974 0.605 0.7186 0.5341 0.4555
0.8121 2.0 400 0.8260 0.675 0.6792 0.6308 0.6112
0.6153 3.0 600 0.8504 0.66 0.6180 0.6026 0.6073
0.441 4.0 800 0.8917 0.685 0.6463 0.6385 0.6403
0.3273 5.0 1000 0.9384 0.69 0.6534 0.6602 0.6561
0.2138 6.0 1200 1.3501 0.705 0.6573 0.6374 0.6388
0.1435 7.0 1400 1.4614 0.71 0.6693 0.6553 0.6601
0.1202 8.0 1600 1.5825 0.7 0.6648 0.6592 0.6530
0.0587 9.0 1800 1.7755 0.72 0.6839 0.6849 0.6840
0.0237 10.0 2000 1.7240 0.735 0.6960 0.6924 0.6940
0.018 11.0 2200 1.7230 0.745 0.7105 0.7003 0.7026
0.0096 12.0 2400 1.7812 0.75 0.7225 0.7142 0.7158
0.006 13.0 2600 1.8223 0.75 0.7265 0.7082 0.7147
0.0033 14.0 2800 1.9872 0.76 0.7434 0.7107 0.7188
0.003 15.0 3000 1.8818 0.72 0.6778 0.6766 0.6765
0.0027 16.0 3200 1.9816 0.75 0.7125 0.6990 0.7043
0.002 17.0 3400 1.9268 0.725 0.6832 0.6834 0.6825
0.0023 18.0 3600 1.9456 0.73 0.6913 0.6898 0.6898
0.0025 19.0 3800 1.9543 0.72 0.6790 0.6770 0.6766
0.0016 20.0 4000 1.9565 0.72 0.6790 0.6770 0.6766

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.13.0+cu116
  • Tokenizers 0.13.2
Downloads last month
28
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for SkyR/albert-base-ours-run-4

Finetuned
(170)
this model