Edit model card

roberta-base-ner-demo-turshilt4

This model is a fine-tuned version of bayartsogt/mongolian-roberta-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.1101
  • Precision: 0.9354
  • Recall: 0.9427
  • F1: 0.9390
  • Accuracy: 0.9815

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: 32
  • eval_batch_size: 64
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 15
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy
0.9016 0.9958 119 0.1663 0.7064 0.7773 0.7401 0.9416
0.1182 2.0 239 0.1002 0.8118 0.8752 0.8423 0.9656
0.064 2.9958 358 0.0733 0.9162 0.9282 0.9221 0.9788
0.0366 4.0 478 0.0826 0.9221 0.9297 0.9259 0.9791
0.0233 4.9958 597 0.0843 0.9251 0.9359 0.9305 0.9798
0.0157 6.0 717 0.0898 0.9305 0.9395 0.9350 0.9802
0.0121 6.9958 836 0.0964 0.9269 0.9369 0.9319 0.9795
0.0093 8.0 956 0.0995 0.9332 0.9408 0.9370 0.9805
0.0072 8.9958 1075 0.1046 0.9306 0.9387 0.9346 0.9799
0.0057 10.0 1195 0.1060 0.9332 0.9410 0.9371 0.9804
0.0049 10.9958 1314 0.1052 0.9322 0.9399 0.9360 0.9810
0.0042 12.0 1434 0.1091 0.9350 0.9416 0.9383 0.9811
0.0037 12.9958 1553 0.1098 0.9350 0.9414 0.9382 0.9812
0.0035 14.0 1673 0.1099 0.9362 0.9426 0.9394 0.9815
0.0033 14.9372 1785 0.1101 0.9354 0.9427 0.9390 0.9815

Framework versions

  • Transformers 4.40.2
  • Pytorch 2.2.1+cu121
  • Datasets 2.19.1
  • Tokenizers 0.19.1
Downloads last month
2
Safetensors
Model size
124M params
Tensor type
F32
·

Finetuned from