DarliAI/kissi-wav2vec2-fula-fleurs-full

wav2vec2-fula-fleurs-full

This model is a fine-tuned version of asr-africa/wav2vec2-xlsr-fula-google-fleurs-5-hours-plus-lm on the fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 0.8529
• Wer: 0.5941

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: 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: 800
• num_epochs: 16
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.8429	3.6585	600	2.7091	1.0610
0.6533	7.3171	1200	0.8250	0.6230
0.5052	10.9756	1800	0.8203	0.5984
0.4041	14.6341	2400	0.8529	0.5941

Framework versions

• Transformers 4.52.3
• Pytorch 2.6.0+cu124
• Datasets 3.6.0
• Tokenizers 0.21.1