Arbi-Houssem/TunLangModel1.2

Whisper Tunisien

This model is a fine-tuned version of openai/whisper-small on the comondov dataset. It achieves the following results on the evaluation set:

• Loss: 1.7109
• Wer: 61.0169

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: 8
• 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: 3000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.004	62.5	500	1.5896	61.0169
0.0003	125.0	1000	1.5918	59.1680
0.0002	187.5	1500	1.6245	59.4761
0.0002	250.0	2000	1.6482	61.9414
0.0001	312.5	2500	1.7052	60.8629
0.0001	375.0	3000	1.7109	61.0169

Framework versions

• Transformers 4.42.0.dev0
• Pytorch 2.3.0+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1