UDA-LIDI-Whisper-large-v2-ECU-911

This model is a fine-tuned version of openai/whisper-large-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8833
• Wer: 40.0395

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
0.7079	1.0	91	0.6057	39.6640
0.4014	2.0	182	0.5828	39.2292
0.2505	3.0	273	0.6180	40.7115
0.1528	4.0	364	0.6764	40.0791
0.0971	5.0	455	0.7001	39.8221
0.0637	6.0	546	0.7852	42.6680
0.0445	7.0	637	0.8403	39.6640
0.0341	8.0	728	0.8778	40.9684
0.0304	9.0	819	0.8678	39.2292
0.0256	9.8950	900	0.8833	40.0395

Framework versions

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