l3-whisper-tiny-l2c_v3

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

• Loss: 0.5051
• Wer: 41.3818

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: 1
• 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
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.4817	0.0933	1000	0.6047	47.5560
0.4557	0.1866	2000	0.5816	46.0588
0.4367	0.2799	3000	0.5620	44.7085
0.4253	0.3731	4000	0.5468	43.6378
0.4124	0.4664	5000	0.5378	43.1714
0.4017	0.5597	6000	0.5258	42.2671
0.3952	0.6530	7000	0.5193	41.7005
0.3901	0.7463	8000	0.5125	41.6003
0.3888	0.8396	9000	0.5077	41.3810
0.3804	0.9328	10000	0.5051	41.3818

Framework versions

• Transformers 4.51.3
• Pytorch 2.7.1+cu128
• Datasets 3.6.0
• Tokenizers 0.21.4