hing-roberta-NCM-run-3

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: 3.2053
• Accuracy: 0.6645
• Precision: 0.6565
• Recall: 0.6479
• F1: 0.6505

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
0.9077	1.0	927	0.8070	0.6397	0.6581	0.6439	0.6382
0.6915	2.0	1854	0.8635	0.6462	0.6368	0.6439	0.6357
0.4785	3.0	2781	1.0961	0.6613	0.6510	0.6556	0.6505
0.3356	4.0	3708	1.6867	0.6667	0.6623	0.6611	0.6595
0.2622	5.0	4635	2.0271	0.6602	0.6589	0.6451	0.6482
0.1957	6.0	5562	2.2565	0.6634	0.6763	0.6517	0.6541
0.1419	7.0	6489	2.4627	0.6440	0.6487	0.6203	0.6230
0.1126	8.0	7416	2.7844	0.6483	0.6347	0.6268	0.6295
0.091	9.0	8343	2.8776	0.6440	0.6302	0.6315	0.6307
0.0758	10.0	9270	3.0246	0.6451	0.6325	0.6227	0.6256
0.0674	11.0	10197	2.9389	0.6721	0.6605	0.6501	0.6530
0.0542	12.0	11124	3.0503	0.6429	0.6456	0.6315	0.6330
0.0576	13.0	12051	3.0252	0.6483	0.6427	0.6435	0.6398
0.0337	14.0	12978	3.1160	0.6731	0.6676	0.6545	0.6575
0.0318	15.0	13905	3.0740	0.6807	0.6733	0.6647	0.6671
0.0188	16.0	14832	3.0890	0.6721	0.6633	0.6574	0.6589
0.0258	17.0	15759	3.1519	0.6634	0.6602	0.6456	0.6490
0.017	18.0	16686	3.1503	0.6688	0.6638	0.6547	0.6568
0.0146	19.0	17613	3.2083	0.6688	0.6621	0.6516	0.6545
0.0125	20.0	18540	3.2053	0.6645	0.6565	0.6479	0.6505

Framework versions

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