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: 0.3748
• Wer: 97.8158
• Cer: 77.1470

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
3.6366	4.5872	500	3.6746	94.9193	77.2587
0.843	9.1743	1000	1.0679	99.8575	79.0879
0.388	13.7615	1500	0.6110	102.8965	81.0031
0.3226	18.3486	2000	0.4898	104.7958	86.9718
0.2976	22.9358	2500	0.4571	113.0579	88.8956
0.2679	27.5229	3000	0.4198	103.5613	80.5050
0.2327	32.1101	3500	0.4085	103.8462	82.1711
0.2099	36.6972	4000	0.3883	101.3295	78.2807
0.1959	41.2844	4500	0.3780	100.1425	77.5936
0.1681	45.8716	5000	0.3748	97.8158	77.1470

Framework versions

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