cheonanAddresses_torch_large_model_model

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

• Loss: 0.0534
• Cer: 1.8909

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.0757	0.3101	1000	0.0772	2.2598
0.0714	0.6202	2000	0.0720	2.3372
0.0624	0.9302	3000	0.0676	2.2765
0.0583	1.2403	4000	0.0642	2.2266
0.0611	1.5504	5000	0.0604	2.0681
0.0552	1.8605	6000	0.0590	2.1282
0.0497	2.1705	7000	0.0569	2.0233
0.0423	2.4806	8000	0.0560	1.9835
0.0495	2.7907	9000	0.0538	1.9387
0.0364	3.1008	10000	0.0534	1.8909

Framework versions

• Transformers 4.41.2
• Pytorch 2.2.2+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1