output_model_whisper_base_shunya_data_augmentation_enabled

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: 0.6446
• Wer: 26.6135

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: 0.0001
• train_batch_size: 2
• eval_batch_size: 4
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 50000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6163	1.5625	2000	0.8828	49.4462
0.1531	3.125	4000	0.6449	36.9222
0.0601	4.6875	6000	0.5940	31.8712
0.0082	6.25	8000	0.5382	26.9384
0.0033	7.8125	10000	0.5168	25.2105
0.0343	9.375	12000	0.6493	31.9598
0.0085	10.9375	14000	0.6135	28.1790
0.0596	12.5	16000	0.7192	34.4262
0.0233	14.0625	18000	0.7470	35.3124
0.0272	15.625	20000	0.7101	32.4915
0.0103	17.1875	22000	0.7010	31.3395
0.0098	18.75	24000	0.7054	30.7488
0.0071	20.3125	26000	0.6713	28.8288
0.0040	21.875	28000	0.6613	27.5735
0.0020	23.4375	30000	0.6940	27.9575
0.0035	25.0	32000	0.6592	27.7655
0.0013	26.5625	34000	0.6453	26.2443
0.0001	28.125	36000	0.6450	27.6030
0.0000	29.6875	38000	0.6406	26.6578
0.0002	31.25	40000	0.6170	25.1071
0.0002	32.8125	42000	0.6494	26.3329
0.0100	34.375	44000	0.6631	27.7507
0.0001	35.9375	46000	0.6504	26.2886
0.0003	37.5	48000	0.6447	27.0713
0.0000	39.0625	50000	0.6446	26.6135

Framework versions

• Transformers 5.3.0.dev0
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.2