wav2vec2_hassaniya_model

This model is a fine-tuned version of abscheik/wav2vec2_hassaniya_model on the None dataset. It achieves the following results on the evaluation set:

• Loss: 1.0391
• Model Preparation Time: 0.006
• Wer: 57.3061
• Cer: 18.8614

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• 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
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Wer	Cer
0.3625	0.9888	44	0.9335	0.006	60.2041	19.3127
0.3419	2.0	89	0.9105	0.006	58.6122	18.9405
0.3167	2.9888	133	0.9591	0.006	58.8980	18.9960
0.2996	4.0	178	0.9696	0.006	58.8163	19.1147
0.2699	4.9888	222	0.9683	0.006	58.1633	18.8534
0.2553	6.0	267	1.0324	0.006	57.5918	18.8614
0.2574	6.9888	311	1.0255	0.006	58.0408	18.9485
0.2333	8.0	356	1.0391	0.006	57.3061	18.8614
0.2327	8.9888	400	1.0831	0.006	57.5510	18.9168
0.2266	9.8876	440	1.0784	0.006	57.7551	18.9089

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1