Akan-3-3000ms – AI Model by devkyle | AlphaNeural AI

You can deploy this model and start earning money today!

devkyle/
Akan-3-3000ms

like

0

automatic-speech-recognition

generated_from_trainer

openai/whisper-tiny

endpoints_compatible

Views

No views yet

whisper-tiny-akan

This model is a fine-tuned version of openai/whisper-tiny on the None dataset. It achieves the following results on the evaluation set:

Loss: 1.0747
Wer: 43.6101

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
train_batch_size: 16
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: 500
training_steps: 3000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4857	10.0	250	0.7120	57.1555
0.0806	20.0	500	0.8478	49.9411
0.0347	30.0	750	0.9223	48.1743
0.0168	40.0	1000	1.0079	55.1826
0.0085	50.0	1250	1.0402	47.3498
0.0051	60.0	1500	1.0890	46.7314
0.0029	70.0	1750	1.0639	44.9352
0.002	80.0	2000	1.0707	44.6702
0.0005	90.0	2250	1.0705	43.7574
0.0005	100.0	2500	1.0721	44.4052
0.0002	110.0	2750	1.0730	43.3451
0.0003	120.0	3000	1.0747	43.6101

Framework versions

Transformers 4.44.2
Pytorch 2.4.0+cu121
Datasets 2.21.0
Tokenizers 0.19.1