malayalam-bert-FakeNews-Dravidian

This model is a fine-tuned version of l3cube-pune/malayalam-bert on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7928
• Accuracy: 0.7840
• Weighted f1 score: 0.7819
• Macro f1 score: 0.7818

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-06
• 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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Weighted f1 score	Macro f1 score
1.0681	1.0	204	1.0238	0.4982	0.3313	0.3325
1.0077	2.0	408	0.9829	0.5031	0.3421	0.3433
0.9738	3.0	612	0.9529	0.5877	0.5499	0.5503
0.946	4.0	816	0.9267	0.6466	0.6401	0.6399
0.9204	5.0	1020	0.9019	0.7006	0.6877	0.6875
0.8961	6.0	1224	0.8754	0.7644	0.7629	0.7628
0.8715	7.0	1428	0.8540	0.7607	0.7544	0.7543
0.8485	8.0	1632	0.8362	0.7828	0.7789	0.7788
0.8323	9.0	1836	0.8244	0.7791	0.7749	0.7748
0.8182	10.0	2040	0.8151	0.7816	0.7773	0.7772
0.8063	11.0	2244	0.8069	0.7816	0.7792	0.7791
0.7973	12.0	2448	0.8011	0.7828	0.7799	0.7798
0.791	13.0	2652	0.7950	0.7853	0.7840	0.7839
0.7857	14.0	2856	0.7939	0.7816	0.7793	0.7792
0.7826	15.0	3060	0.7928	0.7840	0.7819	0.7818

Framework versions

• Transformers 4.35.0
• Pytorch 2.0.0
• Datasets 2.11.0
• Tokenizers 0.14.1