Modelpublic

ramesh070/xlm-r-hatespeechdetection-telugu

sourceHugging Facemitupdated 10mo agoView on Hugging Face

0likes11downloads

Model Card

xlm-r-hatespeechdetection-telugu

This model is a fine-tuned version of FacebookAI/xlm-roberta-base on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.1518
—Accuracy: 0.9643
—F1: 0.9722

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-05
—trainbatchsize: 32
—evalbatchsize: 32
—seed: 42
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.6729	1.0	14	0.6153	0.6429	0.7826
0.6196	2.0	28	0.5032	0.6429	0.7826
0.5297	3.0	42	0.4319	0.6429	0.7826
0.4444	4.0	56	0.3296	0.9643	0.9730
0.3435	5.0	70	0.1656	1.0	1.0
0.2417	6.0	84	0.1390	0.9286	0.9474
0.1392	7.0	98	0.0760	0.9643	0.9722
0.0815	8.0	112	0.1240	0.9643	0.9730
0.1242	9.0	126	0.0870	0.9643	0.9722
0.0776	10.0	140	0.2582	0.9107	0.9333
0.0444	11.0	154	0.3596	0.9107	0.9333
0.0449	12.0	168	0.3502	0.9286	0.9459
0.0534	13.0	182	0.1307	0.9821	0.9859
0.0205	14.0	196	0.2319	0.9464	0.9589
0.0067	15.0	210	0.4442	0.9286	0.9459
0.0045	16.0	224	0.2420	0.9464	0.9589
0.0132	17.0	238	0.4973	0.9286	0.9459
0.0025	18.0	252	0.2707	0.9464	0.9589
0.0015	19.0	266	0.1389	0.9821	0.9859
0.0029	20.0	280	0.1518	0.9643	0.9722

Framework versions

—Transformers 4.53.3
—Pytorch 2.9.1+cu128
—Datasets 4.4.1
—Tokenizers 0.21.2