wav2vec2-base-finetuned-ks

This model is a fine-tuned version of facebook/wav2vec2-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.2562
• Accuracy: 0.9869

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: 32
• seed: 42
• 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: 16

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.4691	0.99	26	2.3935	0.2310
2.1621	1.99	52	2.0155	0.3202
1.8731	2.99	78	1.6397	0.7929
1.4521	3.99	104	1.2337	0.8940
1.101	4.99	130	0.9519	0.9393
0.9401	5.99	156	0.7686	0.975
0.7463	6.99	182	0.6338	0.9774
0.6555	7.99	208	0.5214	0.9810
0.5095	8.99	234	0.4228	0.9869
0.4152	9.99	260	0.3658	0.9857
0.3764	10.99	286	0.3311	0.9857
0.3325	11.99	312	0.2954	0.9881
0.3121	12.99	338	0.2797	0.9869
0.281	13.99	364	0.2650	0.9857
0.2627	14.99	390	0.2571	0.9869
0.2655	15.99	416	0.2562	0.9869

Framework versions

• Transformers 4.21.1
• Pytorch 1.12.1+cu113
• Datasets 1.14.0
• Tokenizers 0.12.1