Modelpublic

KasuleTrevor/multilingual_speech_to_intent_nyn_xlsr

sourceHugging Faceupdated 1y agoView on Hugging Face

0likes2downloads

Model Card

multilingualspeechtointentnyn_xlsr

This model is a fine-tuned version of KasuleTrevor/wav2vec2-xls-r-300m-nyn_filtered-yogera-v3 on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.1478
—Accuracy: 0.9787
—Precision: 0.9795
—Recall: 0.9787
—F1: 0.9789

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: 0.0001
—trainbatchsize: 32
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 64
—optimizer: Use adamwtorch with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: cosine
—lrschedulerwarmup_steps: 500
—num_epochs: 80
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
2.396	1.0	219	1.1252	0.7502	0.7706	0.7502	0.7244
0.3037	2.0	438	0.0997	0.9837	0.9840	0.9837	0.9837
0.1717	3.0	657	0.1135	0.9752	0.9763	0.9752	0.9752
0.1365	4.0	876	0.1008	0.9809	0.9812	0.9809	0.9809
0.1115	5.0	1095	0.0926	0.9816	0.9820	0.9816	0.9816
0.0817	6.0	1314	0.1174	0.9781	0.9788	0.9781	0.9780
0.0877	7.0	1533	0.1599	0.9611	0.9639	0.9611	0.9610
0.0596	8.0	1752	0.1075	0.9802	0.9806	0.9802	0.9802
0.0539	9.0	1971	0.1284	0.9682	0.9697	0.9682	0.9681
0.0627	10.0	2190	0.1918	0.9533	0.9555	0.9533	0.9532

Framework versions

—Transformers 4.51.3
—Pytorch 2.1.0+cu118
—Datasets 3.6.0
—Tokenizers 0.21.2