Makkoen/whisper-medium.en-cit-do015-wd0-lr1e-06-SF-400

./400

This model is a fine-tuned version of openai/whisper-medium.en on the 400 SF 1000 dataset. It achieves the following results on the evaluation set:

• Loss: 0.7217
• Wer Ortho: 29.6283
• Wer: 19.5551

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-06
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• gradient_accumulation_steps: 2
• total_train_batch_size: 16
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• training_steps: 500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
1.6015	4.0	100	1.1045	41.2901	30.3911
0.8237	8.0	200	0.7959	31.3776	20.4880
0.5824	12.0	300	0.7417	29.7741	19.3757
0.4724	16.0	400	0.7256	29.5554	19.4474
0.4232	20.0	500	0.7217	29.6283	19.5551

Framework versions

• Transformers 4.44.0
• Pytorch 1.13.1+cu117
• Datasets 2.20.0
• Tokenizers 0.19.1