UDA-LIDI-Whisper-large-v3-ECU-911

This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8777
• Wer: 37.9051

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: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• 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: constant
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6583	1.0	91	0.5713	39.8617
0.3725	2.0	182	0.5667	37.7866
0.2317	3.0	273	0.6098	37.6285
0.1397	4.0	364	0.6432	37.1937
0.0841	5.0	455	0.7177	39.4466
0.0539	6.0	546	0.7817	39.1700
0.036	7.0	637	0.8725	38.7747
0.0281	8.0	728	0.8485	39.6245
0.0228	9.0	819	0.8553	37.9051
0.0181	9.8950	900	0.8777	37.9051

Framework versions

• Transformers 4.47.0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0