chungnam_firestation_small_model

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

• Loss: 0.0569
• Cer: 204.7194

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: 5e-06
• 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: 250
• training_steps: 2000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
0.9925	1.3245	200	0.9694	214.5036
0.1173	2.6490	400	0.2153	108.6043
0.0366	3.9735	600	0.0958	225.2086
0.0056	5.2980	800	0.0643	224.0576
0.0028	6.6225	1000	0.0559	180.8058
0.0011	7.9470	1200	0.0553	196.0288
0.0008	9.2715	1400	0.0560	228.6906
0.0007	10.5960	1600	0.0566	212.2878
0.0007	11.9205	1800	0.0569	211.7986
0.0006	13.2450	2000	0.0569	204.7194

Framework versions

• Transformers 4.41.2
• Pytorch 2.2.2+cu121
• Datasets 3.2.0
• Tokenizers 0.19.1