addison6

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: 1.4278
• Wer: 0.3784

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
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
4.0707	1.03	500	2.8972	0.9998
1.8869	2.05	1000	1.3286	0.6772
1.218	3.08	1500	1.0999	0.5714
1.0183	4.11	2000	1.0781	0.5523
0.8743	5.13	2500	0.9991	0.4951
0.7561	6.16	3000	1.0861	0.4869
0.6883	7.19	3500	1.1915	0.4704
0.6185	8.21	4000	1.1553	0.4634
0.5761	9.24	4500	1.1144	0.4513
0.5315	10.27	5000	1.1818	0.4298
0.4982	11.29	5500	1.1973	0.4329
0.4719	12.32	6000	1.1456	0.4259
0.4369	13.35	6500	1.1701	0.4292
0.4083	14.37	7000	1.1929	0.4132
0.3886	15.4	7500	1.3307	0.4163
0.3752	16.43	8000	1.3405	0.4081
0.349	17.45	8500	1.3283	0.3952
0.3261	18.48	9000	1.2956	0.4128
0.3094	19.51	9500	1.2671	0.4004
0.3041	20.53	10000	1.3534	0.3964
0.2796	21.56	10500	1.3730	0.3899
0.25	22.59	11000	1.3952	0.3942
0.2303	23.61	11500	1.4792	0.3923
0.2321	24.64	12000	1.4228	0.3847
0.2	25.67	12500	1.4469	0.3837
0.2009	26.69	13000	1.4532	0.3820
0.195	27.72	13500	1.4329	0.3821
0.1804	28.75	14000	1.4265	0.3799
0.1713	29.77	14500	1.4278	0.3784

Framework versions

• Transformers 4.17.0
• Pytorch 2.0.0+cu118
• Datasets 1.18.3
• Tokenizers 0.13.2