Hungarians/whisper-large-v2-hu-cleaned

Whisper Large-v2 Hu - cleaned

This model is a fine-tuned version of openai/whisper-large-v2 on the Common Voice 16.1 hu cleaned dataset. It achieves the following results on the evaluation set:

• Loss: 0.0393
• Wer Ortho: 4.1403
• Wer: 3.5518

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: 5e-06
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 128
• total_train_batch_size: 128
• 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: 600
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
0.0716	0.34	100	0.0690	6.5849	5.9493
0.0539	0.69	200	0.0520	4.9650	4.4825
0.0381	1.03	300	0.0457	4.4900	4.0385
0.0235	1.37	400	0.0423	4.2854	3.7458
0.0221	1.72	500	0.0386	3.9786	3.5518
0.0158	2.06	600	0.0393	4.1403	3.6768

Framework versions

• Transformers 4.37.2
• Pytorch 2.1.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.1