miosipof/whisper_medium_bbrave_12-11-2024_v3.2

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: 0.4710
• Wer: 35.6877

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: 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: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1179	1.3245	100	0.4789	35.8116
0.1085	2.6490	200	0.4765	36.6791
0.1207	3.9735	300	0.4743	36.8030
0.1101	5.2980	400	0.4723	35.9356
0.1	6.6225	500	0.4714	35.6877
0.1133	7.9470	600	0.4710	35.6877

Framework versions

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