bert-base-multilingual-cased-VITD

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

• Loss: 1.4876
• Accuracy: 0.7338
• F1 score: 0.7300

Training and evaluation data

banglaVITD is used for training and validation.

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
0.9505	1.0	169	0.8351	0.6361	0.5587
0.7782	2.0	338	0.7390	0.7120	0.6829
0.5627	3.0	507	0.6598	0.7278	0.7265
0.4067	4.0	676	0.7841	0.7233	0.7239
0.2826	5.0	845	0.9663	0.7353	0.7265
0.219	6.0	1014	1.0408	0.7391	0.7294
0.1434	7.0	1183	1.2585	0.7346	0.7299
0.1018	8.0	1352	1.3435	0.7331	0.7299
0.0734	9.0	1521	1.4392	0.7346	0.7319
0.0558	10.0	1690	1.4876	0.7338	0.7300

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.0
• Datasets 2.1.0
• Tokenizers 0.13.3