hing-roberta-NCM-run-4

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: 3.3405
• Accuracy: 0.6505
• Precision: 0.6410
• Recall: 0.6318
• F1: 0.6350

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
0.8975	1.0	927	0.9553	0.6127	0.5994	0.6026	0.5930
0.6924	2.0	1854	0.8426	0.6505	0.6535	0.6344	0.6372
0.472	3.0	2781	1.0533	0.6570	0.6449	0.6442	0.6442
0.3271	4.0	3708	1.8111	0.6624	0.6635	0.6407	0.6448
0.2368	5.0	4635	2.1234	0.6483	0.6297	0.6288	0.6267
0.172	6.0	5562	2.5340	0.6419	0.6312	0.6164	0.6199
0.1251	7.0	6489	2.5758	0.6472	0.6405	0.6311	0.6336
0.0943	8.0	7416	2.9090	0.6332	0.6337	0.6090	0.6124
0.0919	9.0	8343	2.8236	0.6494	0.6394	0.6301	0.6329
0.0851	10.0	9270	2.9368	0.6570	0.6448	0.6405	0.6422
0.0602	11.0	10197	3.2925	0.6289	0.6221	0.6111	0.6140
0.0551	12.0	11124	3.1185	0.6397	0.6239	0.6108	0.6131
0.0498	13.0	12051	3.0170	0.6559	0.6400	0.6322	0.6341
0.0309	14.0	12978	3.0934	0.6537	0.6481	0.6386	0.6410
0.0303	15.0	13905	3.1530	0.6440	0.6292	0.6258	0.6272
0.028	16.0	14832	3.1491	0.6570	0.6502	0.6346	0.6385
0.0199	17.0	15759	3.2515	0.6526	0.6394	0.6295	0.6324
0.0245	18.0	16686	3.2644	0.6526	0.6494	0.6315	0.6356
0.0159	19.0	17613	3.3344	0.6483	0.6377	0.6295	0.6324
0.0116	20.0	18540	3.3405	0.6505	0.6410	0.6318	0.6350

Framework versions

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