cportoca/whisper-small-hi

Whisper Small Ja-Qve - cportoca

This model is a fine-tuned version of openai/whisper-small on the Quechua_dataset dataset. It achieves the following results on the evaluation set:

• Loss: 0.2232
• Wer: 19.7678

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 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
0.2739	1.3550	1000	0.3388	31.3461
0.1322	2.7100	2000	0.2377	25.1961
0.0275	4.0650	3000	0.2216	20.6464
0.0127	5.4201	4000	0.2232	19.7678

Framework versions

• Transformers 4.46.3
• Pytorch 2.4.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3