whisper-id-jv-multilingual-fp16-base

This model is a fine-tuned version of openai/whisper-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2707
• Wer: 38.7801

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: Use OptimizerNames.ADAMW_TORCH_FUSED 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: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.6752	0.2568	500	0.6657	63.7640
0.4742	0.5136	1000	0.4755	44.4623
0.3744	0.7704	1500	0.3932	41.3082
0.2569	1.0272	2000	0.3418	41.2761
0.2278	1.2840	2500	0.3157	42.1188
0.2222	1.5408	3000	0.2932	35.7303
0.2083	1.7976	3500	0.2772	35.7544
0.1588	2.0544	4000	0.2707	38.7801

Framework versions

• Transformers 4.57.2
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1