Modelpublic

robello2/wav2vec2-xlsr-53-afrispeech-clinical

sourceHugging Faceapache-2.0updated 7mo agoView on Hugging Face

0likes4downloads

Model Card

wav2vec2-xlsr-afrispeech-clinical

This model is a fine-tuned version of jonatasgrosman/wav2vec2-large-xlsr-53-english on the afrispeech-200 dataset. It achieves the following results on the evaluation set:

—Loss: 0.3815
—Model Preparation Time: 0.003
—Wer: 0.295

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: 8
—evalbatchsize: 8
—seed: 42
—gradientaccumulationsteps: 2
—totaltrainbatch_size: 16
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_steps: 2239
—num_epochs: 10
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer
1.3513	1.0	2240	0.5410	0.003	0.391
1.0632	2.0	4480	0.4555	0.003	0.351
0.9540	3.0	6720	0.4268	0.003	0.334
0.8654	4.0	8960	0.4057	0.003	0.328
0.7554	5.0	11200	0.3871	0.003	0.316
0.6892	6.0	13440	0.3936	0.003	0.312
0.6096	7.0	15680	0.3836	0.003	0.303
0.5333	8.0	17920	0.3859	0.003	0.298
0.5335	9.0	20160	0.3858	0.003	0.296
0.5038	10.0	22400	0.3815	0.003	0.295

Framework versions

—Transformers 5.3.0
—Pytorch 2.12.0.dev20260307+cu128
—Datasets 2.19.0
—Tokenizers 0.22.2