devkyle/Akan-tiny-2000ms-1.5k

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: 0.2466
• Wer: 18.4363

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: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.8631	2.9070	250	0.8110	57.7923
0.4257	5.8140	500	0.7295	50.5545
0.1859	8.7209	750	0.7882	50.0381
0.0717	11.6279	1000	0.8608	49.2847
0.0758	14.5349	1250	0.2162	17.7826
0.0242	17.4419	1500	0.2390	19.1234
0.0105	20.3488	1750	0.2467	19.7519
0.0054	23.2558	2000	0.2466	18.4363

Framework versions

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