Edit model card

bert-finetuned-ner

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

  • Loss: 0.0209
  • Precision: 0.8249
  • Recall: 0.8825
  • F1: 0.8527
  • Accuracy: 0.9946
  • B-location-precision: 0.9446
  • B-location-recall: 0.9653
  • B-location-f1: 0.9549
  • I-location-precision: 0.9358
  • I-location-recall: 0.9745
  • I-location-f1: 0.9548
  • B-group-precision: 0.8819
  • B-group-recall: 0.8485
  • B-group-f1: 0.8649
  • I-group-precision: 0.8879
  • I-group-recall: 0.8358
  • I-group-f1: 0.8610
  • B-corporation-precision: 0.8475
  • B-corporation-recall: 0.8552
  • B-corporation-f1: 0.8514
  • I-corporation-precision: 0.8158
  • I-corporation-recall: 0.7294
  • I-corporation-f1: 0.7702
  • B-person-precision: 0.9583
  • B-person-recall: 0.9742
  • B-person-f1: 0.9662
  • I-person-precision: 0.9596
  • I-person-recall: 0.95
  • I-person-f1: 0.9548
  • B-creative-work-precision: 0.8102
  • B-creative-work-recall: 0.7929
  • B-creative-work-f1: 0.8014
  • I-creative-work-precision: 0.8131
  • I-creative-work-recall: 0.8354
  • I-creative-work-f1: 0.8241
  • B-product-precision: 0.8682
  • B-product-recall: 0.7887
  • B-product-f1: 0.8266
  • I-product-precision: 0.8862
  • I-product-recall: 0.8886
  • I-product-f1: 0.8874
  • Corporation-precision: 0.6972
  • Corporation-recall: 0.7919
  • Corporation-f1: 0.7415
  • Corporation-number: 221
  • Creative-work-precision: 0.6433
  • Creative-work-recall: 0.7214
  • Creative-work-f1: 0.6801
  • Creative-work-number: 140
  • Group-precision: 0.7465
  • Group-recall: 0.8144
  • Group-f1: 0.7790
  • Group-number: 264
  • Location-precision: 0.9026
  • Location-recall: 0.9471
  • Location-f1: 0.9243
  • Location-number: 548
  • Person-precision: 0.9101
  • Person-recall: 0.9515
  • Person-f1: 0.9304
  • Person-number: 660
  • Product-precision: 0.6908
  • Product-recall: 0.7394
  • Product-f1: 0.7143
  • Product-number: 142

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: 0.0002
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 3

Training results

Training Loss Epoch Step Validation Loss Precision Recall F1 Accuracy B-location-precision B-location-recall B-location-f1 I-location-precision I-location-recall I-location-f1 B-group-precision B-group-recall B-group-f1 I-group-precision I-group-recall I-group-f1 B-corporation-precision B-corporation-recall B-corporation-f1 I-corporation-precision I-corporation-recall I-corporation-f1 B-person-precision B-person-recall B-person-f1 I-person-precision I-person-recall I-person-f1 B-creative-work-precision B-creative-work-recall B-creative-work-f1 I-creative-work-precision I-creative-work-recall I-creative-work-f1 B-product-precision B-product-recall B-product-f1 I-product-precision I-product-recall I-product-f1 Corporation-precision Corporation-recall Corporation-f1 Corporation-number Creative-work-precision Creative-work-recall Creative-work-f1 Creative-work-number Group-precision Group-recall Group-f1 Group-number Location-precision Location-recall Location-f1 Location-number Person-precision Person-recall Person-f1 Person-number Product-precision Product-recall Product-f1 Product-number
No log 1.0 107 0.1175 0.5693 0.4076 0.4751 0.9701 0.6320 0.7646 0.6920 0.7752 0.3929 0.5215 1.0 0.0114 0.0225 0.6667 0.0176 0.0343 0.9787 0.2081 0.3433 nan 0.0 nan 0.8123 0.7409 0.7750 0.9117 0.555 0.6900 nan 0.0 nan nan 0.0 nan nan 0.0 nan nan 0.0 nan 0.9787 0.2081 0.3433 221 0.0 0.0 0.0 140 0.3333 0.0152 0.0290 264 0.4682 0.6040 0.5275 548 0.6543 0.6424 0.6483 660 0.0 0.0 0.0 142
No log 2.0 214 0.0411 0.6931 0.7489 0.7199 0.9886 0.8194 0.9270 0.8699 0.8701 0.9214 0.8950 0.7919 0.5909 0.6768 0.6897 0.7625 0.7242 0.8297 0.6833 0.7494 0.8548 0.3118 0.4569 0.9139 0.9485 0.9309 0.8996 0.9075 0.9035 0.7541 0.3286 0.4577 0.7952 0.5091 0.6208 0.7407 0.5634 0.64 0.6740 0.8315 0.7445 0.6515 0.5837 0.6158 221 0.2941 0.2143 0.2479 140 0.5051 0.5682 0.5348 264 0.7617 0.8923 0.8218 548 0.8470 0.9227 0.8832 660 0.4091 0.5070 0.4528 142
No log 3.0 321 0.0209 0.8249 0.8825 0.8527 0.9946 0.9446 0.9653 0.9549 0.9358 0.9745 0.9548 0.8819 0.8485 0.8649 0.8879 0.8358 0.8610 0.8475 0.8552 0.8514 0.8158 0.7294 0.7702 0.9583 0.9742 0.9662 0.9596 0.95 0.9548 0.8102 0.7929 0.8014 0.8131 0.8354 0.8241 0.8682 0.7887 0.8266 0.8862 0.8886 0.8874 0.6972 0.7919 0.7415 221 0.6433 0.7214 0.6801 140 0.7465 0.8144 0.7790 264 0.9026 0.9471 0.9243 548 0.9101 0.9515 0.9304 660 0.6908 0.7394 0.7143 142

Framework versions

  • Transformers 4.35.0
  • Pytorch 2.1.0+cpu
  • Datasets 2.14.6
  • Tokenizers 0.14.1
Downloads last month
8
Safetensors
Model size
109M params
Tensor type
F32
·

Finetuned from