WAV2VEC2_CAPSTONE_MODEL

This model is a fine-tuned version of facebook/wav2vec2-base on a subset of the common_voice_16_1 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3952
• Accuracy: 0.9098
• F1 score: 0.9097

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: 0.0003
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 80
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 8
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1 score
0.9339	1.0	776	1.4214	0.7162	0.7094
0.5663	2.0	1552	1.0182	0.8318	0.8277
0.4408	3.0	2328	0.6117	0.8795	0.8784
0.3521	4.0	3105	0.5092	0.8998	0.9001
0.2305	5.0	3881	0.3896	0.9004	0.9013
0.1219	6.0	4657	0.3096	0.9196	0.9194
0.0672	6.99	3591	0.3952	0.9098	0.9097

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.2+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0