pretrain / README.md
suku9's picture
Model save
338c55b verified
metadata
library_name: transformers
tags:
  - generated_from_trainer
model-index:
  - name: pretrain
    results: []

pretrain

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

  • Loss: 0.5260

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: 1024
  • eval_batch_size: 1024
  • seed: 42
  • optimizer: Use adamw_torch with betas=(0.9,0.95) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 156250
  • num_epochs: 25
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.4587 0.3774 500 1.7455
0.3676 0.7547 1000 1.3984
0.3343 1.1321 1500 1.2729
0.3118 1.5094 2000 1.1772
0.2953 1.8868 2500 1.0904
0.2771 2.2642 3000 1.0169
0.2605 2.6415 3500 0.9581
0.2501 3.0189 4000 0.8991
0.2351 3.3962 4500 0.8535
0.2245 3.7736 5000 0.8164
0.2168 4.1509 5500 0.7843
0.2121 4.5283 6000 0.7684
0.205 4.9057 6500 0.7447
0.1999 5.2830 7000 0.7284
0.196 5.6604 7500 0.7089
0.1894 6.0377 8000 0.7045
0.188 6.4151 8500 0.6867
0.1826 6.7925 9000 0.6750
0.1821 7.1698 9500 0.6672
0.1753 7.5472 10000 0.6650
0.1746 7.9245 10500 0.6485
0.1714 8.3019 11000 0.6420
0.1726 8.6792 11500 0.6365
0.169 9.0566 12000 0.6300
0.1659 9.4340 12500 0.6244
0.1653 9.8113 13000 0.6164
0.1646 10.1887 13500 0.6122
0.1623 10.5660 14000 0.6070
0.1629 10.9434 14500 0.6045
0.1603 11.3208 15000 0.5999
0.16 11.6981 15500 0.5948
0.1582 12.0755 16000 0.5898
0.1565 12.4528 16500 0.5868
0.1541 12.8302 17000 0.5844
0.1553 13.2075 17500 0.5798
0.152 13.5849 18000 0.5791
0.1536 13.9623 18500 0.5745
0.1525 14.3396 19000 0.5722
0.1516 14.7170 19500 0.5718
0.151 15.0943 20000 0.5675
0.1502 15.4717 20500 0.5672
0.1505 15.8491 21000 0.5639
0.1497 16.2264 21500 0.5607
0.1495 16.6038 22000 0.5583
0.1463 16.9811 22500 0.5547
0.1478 17.3585 23000 0.5556
0.1468 17.7358 23500 0.5534
0.1468 18.1132 24000 0.5509
0.1447 18.4906 24500 0.5480
0.1451 18.8679 25000 0.5479
0.1449 19.2453 25500 0.5453
0.1433 19.6226 26000 0.5449
0.1434 20.0 26500 0.5423
0.1434 20.3774 27000 0.5404
0.1428 20.7547 27500 0.5393
0.1435 21.1321 28000 0.5391
0.142 21.5094 28500 0.5371
0.142 21.8868 29000 0.5342
0.1418 22.2642 29500 0.5340
0.1417 22.6415 30000 0.5322
0.1405 23.0189 30500 0.5309
0.1412 23.3962 31000 0.5300
0.1395 23.7736 31500 0.5295
0.1383 24.1509 32000 0.5289
0.1373 24.5283 32500 0.5272
0.139 24.9057 33000 0.5260

Framework versions

  • Transformers 4.51.1
  • Pytorch 2.6.0+cu124
  • Datasets 3.5.0
  • Tokenizers 0.21.1