emilios/whisper-lg-v2-parsifal-el

Whisper Large V2 Farsipal and El Greco

This model is a fine-tuned version of emilios/whisper-lg-v2-parsifal-el-0 on the mozilla-foundation/common_voice_11_0 el dataset. It achieves the following results on the evaluation set:

• Loss: 0.2866
• Wer: 9.1103

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: 32
• eval_batch_size: 16
• seed: 42
• distributed_type: multi-GPU
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 5000

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.0281	9.35	1000	0.2515	9.5561
0.0131	18.69	2000	0.2866	9.1103
0.0069	28.04	3000	0.3118	9.2403
0.004	37.38	4000	0.3279	9.3796
0.0025	46.73	5000	0.3464	9.3146

Framework versions

• Transformers 4.26.0.dev0
• Pytorch 2.0.0.dev20221216+cu116
• Datasets 2.7.1.dev0
• Tokenizers 0.13.2