hing-roberta-NCM-run-1

This model is a fine-tuned version of l3cube-pune/hing-roberta on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 3.2912
• Accuracy: 0.6667
• Precision: 0.6513
• Recall: 0.6494
• F1: 0.6502

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: 3e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.8968	1.0	927	0.8552	0.6257	0.6508	0.5961	0.5969
0.7022	2.0	1854	1.1142	0.3937	0.3270	0.3273	0.2051
0.5569	3.0	2781	0.9130	0.6591	0.6566	0.6612	0.6509
0.363	4.0	3708	1.6630	0.6526	0.6634	0.6414	0.6436
0.2801	5.0	4635	2.0458	0.6451	0.6339	0.6345	0.6330
0.1925	6.0	5562	2.3378	0.6570	0.6439	0.6254	0.6277
0.1297	7.0	6489	2.5205	0.6839	0.6719	0.6651	0.6675
0.114	8.0	7416	2.8373	0.6505	0.6379	0.6249	0.6280
0.0994	9.0	8343	2.5358	0.6634	0.6539	0.6446	0.6474
0.0977	10.0	9270	2.8244	0.6537	0.6489	0.6210	0.6238
0.0623	11.0	10197	2.7593	0.6764	0.6602	0.6487	0.6510
0.0537	12.0	11124	2.9823	0.6677	0.6679	0.6450	0.6488
0.0432	13.0	12051	3.0792	0.6537	0.6465	0.6352	0.6378
0.0406	14.0	12978	3.0707	0.6688	0.6592	0.6509	0.6534
0.0296	15.0	13905	3.3289	0.6667	0.6596	0.6452	0.6486
0.0288	16.0	14832	3.2147	0.6645	0.6592	0.6512	0.6528
0.024	17.0	15759	3.3284	0.6645	0.6470	0.6405	0.6425
0.0201	18.0	16686	3.2428	0.6688	0.6515	0.6515	0.6515
0.0176	19.0	17613	3.2680	0.6710	0.6574	0.6536	0.6547
0.0168	20.0	18540	3.2912	0.6667	0.6513	0.6494	0.6502

Framework versions

• Transformers 4.20.1
• Pytorch 1.10.1+cu111
• Datasets 2.3.2
• Tokenizers 0.12.1