Whisper Small re-tuned_1

This model is a fine-tuned version of eolang/whisper-small-sw-WER-13-zindi on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.8944
• Wer Ortho: 40.9465
• Wer: 39.9160

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
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.0018	41.6667	500	1.6457	45.0617	43.6975
0.0011	83.3333	1000	1.7081	46.5021	45.3782
0.0009	125.0	1500	1.7469	67.9012	67.2269
0.0009	166.6667	2000	1.7766	68.9300	68.2773
0.0009	208.3333	2500	1.8003	70.3704	69.7479
0.0008	250.0	3000	1.8225	70.3704	69.7479
0.0008	291.6667	3500	1.8417	40.7407	39.9160
0.0008	333.3333	4000	1.8590	40.3292	39.4958
0.0008	375.0	4500	1.8773	40.9465	40.1261
0.0008	416.6667	5000	1.8944	40.9465	39.9160

Framework versions

• Transformers 4.46.1
• Pytorch 2.5.1+cu121
• Datasets 3.3.2
• Tokenizers 0.20.3