predika-org/ayira-large-turbo-v3

Ayira Large Turbo V3

This model is a fine-tuned version of unsloth/whisper-large-v3-turbo on the Haitian Creole ASR Dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.2041
• Wer: 5.6131

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: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• 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
0.0906	2.9499	1000	0.1945	15.4483
0.0296	5.8997	2000	0.1862	6.4095
0.0032	8.8496	3000	0.1946	5.6375
0.0004	11.7994	4000	0.2041	5.6131

Framework versions

• Transformers 4.52.4
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1