med_v1_M01_1e-05 – AI Model by mprzibilla | AlphaNeural AI

You can deploy this model and start earning money today!

mprzibilla/
med_v1_M01_1e-05

like

0

automatic-speech-recognition

generated_from_trainer

endpoints_compatible

Views

No views yet

med_v1_M01_1e-05

This model is a fine-tuned version of facebook/wav2vec2-base on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 3.2024
Wer: 1.0
Cer: 0.9785

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: 1e-05
train_batch_size: 20
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 13500
num_epochs: 2000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
47.8457	200.0	27000	3.4905	1.0	0.9851
3.3161	400.0	54000	3.4257	1.0	0.9851
3.1908	600.0	81000	3.3574	1.0	0.9851
3.072	800.0	108000	3.2488	1.0	0.9851
2.9895	1000.0	135000	3.2019	1.0	0.9851
2.9343	1200.0	162000	3.1951	1.0	0.9835
2.8974	1400.0	189000	3.1969	1.0	0.9818
2.8738	1600.0	216000	3.2007	1.0	0.9818
2.8588	1800.0	243000	3.2020	1.0	0.9785
2.852	2000.0	270000	3.2024	1.0	0.9785

Framework versions

Transformers 4.18.0
Pytorch 1.10.2+cu102
Datasets 2.3.2
Tokenizers 0.12.1