xls-r-1b-bemgen-male-model

This model is a fine-tuned version of facebook/wav2vec2-xls-r-1b on the BEMGEN - NA dataset. It achieves the following results on the evaluation set:

• Loss: 0.3453
• Wer: 0.4416

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-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	0.3604	100	3.5018	1.0
No log	0.7207	200	2.8616	1.0
No log	1.0793	300	1.1005	0.9874
No log	1.4396	400	0.6285	0.8360
5.4329	1.8	500	0.5177	0.7589
5.4329	2.1586	600	0.4120	0.5989
5.4329	2.5189	700	0.3998	0.5496
5.4329	2.8793	800	0.3715	0.5654
5.4329	3.2378	900	0.3351	0.4872
0.644	3.5982	1000	0.3334	0.5015
0.644	3.9586	1100	0.3192	0.4796
0.644	4.3171	1200	0.3246	0.4594
0.644	4.6775	1300	0.3242	0.4612
0.644	5.0360	1400	0.3203	0.4442
0.3852	5.3964	1500	0.3453	0.4416

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0