Makkoen/whisper-medium.en-cit-do015-wd0-lr3e-06-SF-800

./800

This model is a fine-tuned version of openai/whisper-medium.en on the 800 SF 1000 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6191
• Wer Ortho: 30.5394
• Wer: 20.0215

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: 3e-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: 800
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer Ortho	Wer
1.2835	2.0	100	0.7681	30.5758	19.3039
0.5883	4.0	200	0.6235	27.6968	17.5099
0.3246	6.0	300	0.5332	29.4461	19.6268
0.1851	8.0	400	0.5366	34.6574	23.3226
0.1133	10.0	500	0.5747	29.9198	19.0886
0.0837	12.0	600	0.5947	30.1020	19.9498
0.0697	14.0	700	0.6128	30.3571	20.4521
0.0622	16.0	800	0.6191	30.5394	20.0215

Framework versions

• Transformers 4.44.0
• Pytorch 1.13.1+cu117
• Datasets 2.21.0
• Tokenizers 0.19.1