akan-whisper-medium

This model is a fine-tuned version of akan-whisper-medium on the audiofolder dataset. It achieves the following results on the evaluation set:

• Loss: 1.3225
• Wer: 65.3783

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: 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6326	7.5758	500	0.8675	71.1922
0.1649	15.1515	1000	0.8976	65.7903
0.0284	22.7273	1500	1.0661	66.8563
0.0067	30.3030	2000	1.1925	66.7517
0.0034	37.8788	2500	1.2540	68.6221
0.0022	45.4545	3000	1.2904	67.0721
0.0018	53.0303	3500	1.3139	67.5038
0.0017	60.6061	4000	1.3225	65.3783

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1