w2v2-base-pretrained_lr5e-5_at0.7_da1 – AI Model by MelanieKoe | AlphaNeural AI

You can deploy this model and start earning money today!

MelanieKoe/
w2v2-base-pretrained_lr5e-5_at0.7_da1

like

0

automatic-speech-recognition

generated_from_trainer

facebook/wav2vec2-base

endpoints_compatible

Views

No views yet

w2v2-base-pretrained_lr5e-5_at0.7_da1

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

Loss: 1.8380
Wer: 0.1653

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: 5e-05
train_batch_size: 32
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: 1000
training_steps: 4000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
18.7456	4.46	250	4.0623	1.0
3.321	8.93	500	3.1987	1.0
3.0779	13.39	750	3.0954	1.0
1.8257	17.86	1000	0.8216	0.6275
0.3449	22.32	1250	1.1757	0.3268
0.2036	26.79	1500	1.2369	0.1982
0.1299	31.25	1750	1.1629	0.1991
0.0998	35.71	2000	1.3491	0.1743
0.0772	40.18	2250	1.4032	0.1730
0.0667	44.64	2500	1.7240	0.1756
0.0558	49.11	2750	1.7005	0.1709
0.0488	53.57	3000	1.7088	0.1717
0.0427	58.04	3250	1.6884	0.1649
0.0389	62.5	3500	1.7467	0.1670
0.035	66.96	3750	1.8174	0.1653
0.0332	71.43	4000	1.8380	0.1653

Framework versions

Transformers 4.35.0
Pytorch 2.0.0
Datasets 2.14.6
Tokenizers 0.14.1