wav2vec2-large-ft-fake-detection

This model is a fine-tuned version of facebook/wav2vec2-large on the alexandreacff/kaggle-fake-detection dataset. It achieves the following results on the evaluation set:

• Loss: 0.6694
• Accuracy: 0.7103

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: 32
• eval_batch_size: 16
• seed: 0
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.6274	0.9851	33	0.6254	0.6206
0.4961	2.0	67	0.9477	0.6159
0.3391	2.9851	100	0.9273	0.6411
0.2857	4.0	134	0.6611	0.6617
0.3186	4.9851	167	0.7654	0.6215
0.2483	6.0	201	0.9395	0.6224
0.239	6.9851	234	0.8367	0.6542
0.2049	8.0	268	0.7709	0.6860
0.224	8.9851	301	0.6694	0.7103
0.2279	9.8507	330	0.7867	0.6822

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.1.0a0+32f93b1
• Datasets 2.19.1
• Tokenizers 0.19.1