Edit model card

roberta-large-finetuned-code-mixed-DS

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

  • Loss: 3.1340
  • Accuracy: 0.7203
  • Precision: 0.6584
  • Recall: 0.6548
  • F1: 0.6558

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: 16
  • eval_batch_size: 32
  • seed: 43
  • 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
0.9729 1.0 248 0.7491 0.6922 0.6434 0.6625 0.6358
0.7474 1.99 496 0.6947 0.7183 0.6712 0.6915 0.6760
0.5938 2.99 744 0.7370 0.7123 0.6624 0.6839 0.6642
0.4264 3.98 992 0.8820 0.7123 0.6540 0.6636 0.6492
0.2806 4.98 1240 1.2022 0.7404 0.6807 0.6694 0.6742
0.2239 5.98 1488 1.3933 0.7223 0.6593 0.6587 0.6568
0.1585 6.97 1736 1.8543 0.7304 0.6730 0.6763 0.6737
0.1302 7.97 1984 2.0783 0.7143 0.6495 0.6520 0.6504
0.1008 8.96 2232 2.3523 0.7183 0.6588 0.6561 0.6552
0.0793 9.96 2480 2.5260 0.7163 0.6516 0.6566 0.6538
0.0498 10.96 2728 2.6074 0.7425 0.6902 0.6817 0.6830
0.0484 11.95 2976 2.6758 0.7284 0.6687 0.6734 0.6709
0.0409 12.95 3224 2.8658 0.7425 0.6817 0.6756 0.6781
0.0239 13.94 3472 2.9484 0.7465 0.6980 0.6818 0.6870
0.025 14.94 3720 3.0827 0.7304 0.6778 0.6577 0.6641
0.0286 15.94 3968 3.0011 0.7183 0.6509 0.6475 0.6491
0.0264 16.93 4216 3.1581 0.7264 0.6645 0.6563 0.6595
0.009 17.93 4464 3.1200 0.7223 0.6589 0.6561 0.6569
0.012 18.92 4712 3.1364 0.7203 0.6573 0.6503 0.6525
0.017 19.92 4960 3.1340 0.7203 0.6584 0.6548 0.6558

Framework versions

  • Transformers 4.20.1
  • Pytorch 1.10.1+cu111
  • Datasets 2.3.2
  • Tokenizers 0.12.1
Downloads last month
24
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.