misterkissi/whisper-small-haitian-creole

wisper-small-haitian-creole

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

• Loss: 0.2371
• Wer: 0.1675

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: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Use 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: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
1.0385	1.6447	500	0.2865	0.2843
0.1426	3.2895	1000	0.2107	0.1897
0.0475	4.9342	1500	0.2010	0.2111
0.0174	6.5789	2000	0.2093	0.2216
0.0075	8.2237	2500	0.2182	0.1757
0.0039	9.8684	3000	0.2245	0.1750
0.0027	11.5132	3500	0.2287	0.1591
0.0023	13.1579	4000	0.2330	0.1706
0.002	14.8026	4500	0.2356	0.1706
0.0018	16.4474	5000	0.2371	0.1675

Framework versions

• Transformers 4.53.2
• Pytorch 2.6.0+cu124
• Datasets 2.18.0
• Tokenizers 0.21.2