miosipof/whisper_medium_BB_and_EC_v1

Whisper Medium

This model is a fine-tuned version of b-brave/asr_double_training_15-10-2024_merged on the b-brave/speech_disorders_voice_edit dataset. It achieves the following results on the evaluation set:

• Loss: 0.4541
• Wer: 38.9095

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• 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: 150
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.7263	0.6849	100	0.5099	41.0161
0.2741	1.3699	200	0.4735	41.5118
0.1736	2.0548	300	0.4578	40.6444
0.0996	2.7397	400	0.4506	38.7856
0.0726	3.4247	500	0.4620	41.0161
0.0605	4.1096	600	0.4531	39.5291
0.0432	4.7945	700	0.4541	38.9095

Framework versions

• PEFT 0.13.2
• Transformers 4.45.2
• Pytorch 2.2.0
• Datasets 3.1.0
• Tokenizers 0.20.3