alkiskoudounas/whisper-el-medium-augmented

Whisper Medium Greek - Robust

This model is a fine-tuned version of openai/whisper-medium on the mozilla-foundation/common_voice_11_0 el dataset. It achieves the following results on the evaluation set:

• Loss: 0.3168
• Wer: 21.6846

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.3865	1.17	500	0.5842	51.4487
0.2302	2.35	1000	0.4861	39.3202
0.1321	3.52	1500	0.4536	37.4257
0.0916	4.69	2000	0.4103	39.6824
0.0497	5.87	2500	0.4101	29.1883
0.03	7.04	3000	0.4121	28.0089
0.0156	8.22	3500	0.3842	26.7459
0.0037	9.39	4000	0.3433	28.7054
0.0008	10.56	4500	0.3244	21.8332
0.0006	11.74	5000	0.3178	21.5267

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 1.11.0+cu113
• Datasets 2.7.1
• Tokenizers 0.12.1