alakxender/whisper-large-dv-a40

This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0252
• Wer: 2.0163
• Wer Ortho: 15.2648

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: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: constant_with_warmup
• lr_scheduler_warmup_steps: 50
• training_steps: 4000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Wer Ortho
0.0426	0.0354	500	0.0501	3.6048	24.5780
0.0293	0.0709	1000	0.0367	2.6889	21.3792
0.0251	0.1063	1500	0.0317	2.3869	17.6751
0.0244	0.1418	2000	0.0296	2.2782	16.7890
0.0209	0.1772	2500	0.0284	2.2486	16.2831
0.0205	0.2126	3000	0.0254	1.9749	14.9776
0.0234	0.2481	3500	0.0261	2.1892	15.1784
0.0229	0.2835	4000	0.0252	2.0163	15.2648

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.0
• Tokenizers 0.19.1