versae/whisper-large-nob-ncc-s

Whisper Large Norwegian

This model is a fine-tuned version of openai/whisper-large-v2 on the NbAiLab/NCC_S dataset. It achieves the following results on the evaluation set:

• Loss: 0.2776
• Wer: 12.5152

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: 12
• eval_batch_size: 6
• seed: 42
• distributed_type: multi-GPU
• 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
0.6892	0.2	1000	0.3177	15.1035
0.6782	0.4	2000	0.3033	13.4592
0.6317	0.6	3000	0.2909	13.7637
0.5609	0.8	4000	0.2803	12.6675
0.5726	1.0	5000	0.2776	12.5152

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.7.1.dev0
• Tokenizers 0.11.0