testc8-1 / README.md
shafin's picture
update model card README.md
17f6d0d
metadata
tags:
  - generated_from_trainer
model-index:
  - name: testc8-1
    results: []

testc8-1

This model is a fine-tuned version of shafin/chemical-bert-uncased-finetuned-cust-c2 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1490

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
  • num_epochs: 30
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.0415 1.0 16 0.1392
0.0443 2.0 32 0.1289
0.0471 3.0 48 0.1363
0.042 4.0 64 0.1598
0.0452 5.0 80 0.1571
0.0446 6.0 96 0.1733
0.0466 7.0 112 0.1301
0.0391 8.0 128 0.1359
0.0425 9.0 144 0.1324
0.0436 10.0 160 0.0939
0.0406 11.0 176 0.1495
0.0387 12.0 192 0.1592
0.0335 13.0 208 0.1118
0.0413 14.0 224 0.1508
0.0363 15.0 240 0.1471
0.0428 16.0 256 0.1721
0.0384 17.0 272 0.1853
0.0381 18.0 288 0.1578
0.0373 19.0 304 0.1707
0.0351 20.0 320 0.1241
0.0346 21.0 336 0.1602
0.0386 22.0 352 0.1207
0.0274 23.0 368 0.1642
0.0338 24.0 384 0.1169
0.0327 25.0 400 0.1461
0.026 26.0 416 0.1323
0.0315 27.0 432 0.1403
0.042 28.0 448 0.1056
0.0346 29.0 464 0.1186
0.0294 30.0 480 0.1490

Framework versions

  • Transformers 4.24.0
  • Pytorch 1.12.1+cu113
  • Datasets 2.7.1
  • Tokenizers 0.13.2