asr-somali-may – AI Model by ayaazlion | AlphaNeural AI

You can deploy this model and start earning money today!

ayaazlion/
asr-somali-may

like

0

automatic-speech-recognition

generated_from_trainer

facebook/wav2vec2-xls-r-300m

endpoints_compatible

Views

No views yet

asr-somali-may

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

Loss: inf
Wer: 0.5869

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: 0.0003
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 32
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 74

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.5445	3.65	400	inf	0.9480
1.0779	7.31	800	inf	0.7415
0.7285	10.96	1200	inf	0.7151
0.5235	14.61	1600	inf	0.6895
0.3967	18.26	2000	inf	0.6626
0.3071	21.92	2400	inf	0.6900
0.2488	25.57	2800	inf	0.6573
0.2106	29.22	3200	inf	0.6505
0.1808	32.88	3600	inf	0.6353
0.1556	36.53	4000	inf	0.6443
0.1328	40.18	4400	inf	0.6241
0.1152	43.84	4800	inf	0.6205
0.0993	47.49	5200	inf	0.6183
0.0861	51.14	5600	inf	0.6102
0.0741	54.79	6000	inf	0.5923
0.0621	58.45	6400	inf	0.5959
0.0514	62.1	6800	inf	0.6048
0.0446	65.75	7200	inf	0.5954
0.0372	69.41	7600	inf	0.5909
0.034	73.06	8000	inf	0.5869

Framework versions

Transformers 4.33.1
Pytorch 2.1.2
Datasets 2.19.2
Tokenizers 0.13.3