fs-w-xavier-base

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

• Loss: 0.3940
• Wer: 97.2460
• Cer: 75.7386

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: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
3.7367	4.5872	500	3.7672	106.7426	75.6699
0.9212	9.1743	1000	1.1453	112.3457	89.8145
0.4126	13.7615	1500	0.6488	104.0361	81.1491
0.3365	18.3486	2000	0.5043	106.0304	81.9048
0.304	22.9358	2500	0.4632	109.7341	85.4174
0.2752	27.5229	3000	0.4296	103.2764	79.6633
0.2388	32.1101	3500	0.4201	101.7569	78.3494
0.218	36.6972	4000	0.4022	98.6705	76.3140
0.1995	41.2844	4500	0.3983	98.4330	76.5201
0.1743	45.8716	5000	0.3940	97.2460	75.7386

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.1+cu121
• Datasets 3.0.1
• Tokenizers 0.20.0