bangla-bert-base-VITD

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

• Loss: 1.9621
• Accuracy: 0.7263
• F1 score: 0.7237

Training and evaluation data

banglaVITD is used for training and validation.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 score
0.8085	1.0	169	0.7044	0.6970	0.7005
0.5263	2.0	338	0.6754	0.7211	0.7124
0.3115	3.0	507	0.7353	0.7301	0.7261
0.1598	4.0	676	1.0257	0.7180	0.7156
0.0906	5.0	845	1.5686	0.7030	0.7032
0.0702	6.0	1014	1.5592	0.7383	0.7307
0.0166	7.0	1183	1.7670	0.7293	0.7250
0.0092	8.0	1352	1.9016	0.7135	0.7129
0.0056	9.0	1521	1.9197	0.7248	0.7216
0.0018	10.0	1690	1.9621	0.7263	0.7237

Framework versions

• Transformers 4.30.2

• Pytorch 2.0.0

• Datasets 2.1.0

• Tokenizers 0.13.3