bert-question-classifier

This model is a fine-tuned version of google-bert/bert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.9776
  • Accuracy: 0.9688
  • Recall: 0.8356
  • Precision: 0.8233
  • F1: 0.8294

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: 4
  • eval_batch_size: 4
  • seed: 42
  • optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 5

Training results

Training Loss Epoch Step Validation Loss Accuracy Recall Precision F1
No log 0.0959 100 5.2538 0.8954 0.4113 0.4226 0.4169
No log 0.1918 200 4.8785 0.9119 0.4529 0.5177 0.4831
No log 0.2876 300 4.5415 0.9206 0.5307 0.5675 0.5485
No log 0.3835 400 4.2158 0.9319 0.6238 0.6261 0.6249
4.7973 0.4794 500 3.9246 0.9365 0.6445 0.6528 0.6487
4.7973 0.5753 600 3.7378 0.9393 0.6749 0.6630 0.6689
4.7973 0.6711 700 3.5841 0.9420 0.6849 0.6794 0.6821
4.7973 0.7670 800 3.3128 0.9471 0.7255 0.7024 0.7138
4.7973 0.8629 900 3.1501 0.9476 0.7261 0.7058 0.7158
3.4614 0.9588 1000 2.9866 0.9498 0.7354 0.7189 0.7271
3.4614 1.0547 1100 2.8197 0.9532 0.7509 0.7386 0.7447
3.4614 1.1505 1200 2.6852 0.9544 0.7559 0.7457 0.7508
3.4614 1.2464 1300 2.5284 0.9585 0.7776 0.7685 0.7730
3.4614 1.3423 1400 2.4018 0.9612 0.7981 0.7802 0.7890
2.6456 1.4382 1500 2.3255 0.9620 0.7996 0.7860 0.7927
2.6456 1.5340 1600 2.2501 0.9630 0.8089 0.7891 0.7989
2.6456 1.6299 1700 2.1575 0.9655 0.8229 0.8022 0.8124
2.6456 1.7258 1800 2.0949 0.9658 0.8158 0.8095 0.8126
2.6456 1.8217 1900 2.0348 0.9666 0.8282 0.8086 0.8183
2.1476 1.9175 2000 1.9776 0.9688 0.8356 0.8233 0.8294
2.1476 2.0134 2100 1.9170 0.9685 0.8390 0.8187 0.8287
2.1476 2.1093 2200 1.8676 0.9685 0.8347 0.8214 0.828

Framework versions

  • Transformers 4.48.1
  • Pytorch 2.5.1+cu124
  • Datasets 3.2.0
  • Tokenizers 0.21.0
Downloads last month
112
Safetensors
Model size
110M params
Tensor type
F32
·
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.

Model tree for lekhnathrijal/bert-question-classifier

Finetuned
(2415)
this model