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: 0.3550
• Accuracy: 0.8727

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: 5e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.1
• lr_scheduler_warmup_steps: 500
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	8	0.6840	0.6
0.6867	2.0	16	0.6780	0.6364
0.6742	3.0	24	0.6601	0.6182
0.6446	4.0	32	0.6294	0.6364
0.6299	5.0	40	0.6002	0.6727
0.6299	6.0	48	0.5755	0.7091
0.6021	7.0	56	0.5530	0.7273
0.5678	8.0	64	0.5036	0.8182
0.5512	9.0	72	0.4753	0.8545
0.4784	10.0	80	0.4184	0.9273
0.4784	11.0	88	0.4102	0.8909
0.4515	12.0	96	0.4444	0.8182
0.4878	13.0	104	0.3780	0.9091
0.4418	14.0	112	0.4570	0.8
0.4746	15.0	120	0.3870	0.8545
0.4746	16.0	128	0.3932	0.8364
0.4226	17.0	136	0.2779	0.9636
0.4301	18.0	144	0.3125	0.9455
0.3482	19.0	152	0.3212	0.9091
0.3611	20.0	160	0.3925	0.8364
0.3611	21.0	168	0.3389	0.8909
0.3507	22.0	176	0.3099	0.8727
0.3241	23.0	184	0.3120	0.8727
0.2533	24.0	192	0.2313	0.9455
0.2466	25.0	200	0.3550	0.8727

Framework versions

• Transformers 4.30.2
• Pytorch 2.0.1+cu118
• Datasets 2.13.1
• Tokenizers 0.13.3