subhadeep_whisper_base_finetune_teacher_babble_noise_libri_360_hours_100_epochs_batch_8

This model is a fine-tuned version of openai/whisper-base.en on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2491
• Wer: 13.5528

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: 0.0005
• train_batch_size: 8
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 256
• total_train_batch_size: 2048
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.2
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7942	1.98	100	0.2872	16.8523
0.1675	3.98	200	0.2003	13.5730
0.0819	5.98	300	0.1944	13.1208
0.0418	7.98	400	0.2070	13.0639
0.0264	9.98	500	0.2199	13.0289
0.0227	11.98	600	0.2310	13.3690
0.0218	13.98	700	0.2322	13.1870
0.02	15.98	800	0.2405	13.1466
0.0207	17.98	900	0.2496	13.4444
0.0226	19.98	1000	0.2491	13.5528

Framework versions

• Transformers 4.25.1
• Pytorch 1.12.1
• Datasets 2.8.0
• Tokenizers 0.13.2