wav2vec2-base-timit-demo-google-colab-Ezra_William

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: 0.5198
• Wer: 0.3335

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

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
3.4828	1.0	500	1.6354	1.0429
0.8406	2.01	1000	0.5389	0.5405
0.4345	3.01	1500	0.4202	0.4438
0.2912	4.02	2000	0.4195	0.4216
0.2316	5.02	2500	0.4253	0.4051
0.1917	6.02	3000	0.3969	0.3958
0.1545	7.03	3500	0.4291	0.3912
0.1423	8.03	4000	0.4349	0.3731
0.1234	9.04	4500	0.4419	0.3784
0.1124	10.04	5000	0.4713	0.3741
0.0991	11.04	5500	0.4711	0.3692
0.0924	12.05	6000	0.4994	0.3699
0.0809	13.05	6500	0.4888	0.3643
0.0715	14.06	7000	0.4828	0.3634
0.0646	15.06	7500	0.5058	0.3570
0.0604	16.06	8000	0.5586	0.3637
0.0571	17.07	8500	0.4991	0.3553
0.0532	18.07	9000	0.5317	0.3566
0.0471	19.08	9500	0.5308	0.3508
0.0449	20.08	10000	0.5362	0.3486
0.0373	21.08	10500	0.5211	0.3479
0.0351	22.09	11000	0.5132	0.3445
0.0333	23.09	11500	0.4927	0.3381
0.0302	24.1	12000	0.5330	0.3413
0.0282	25.1	12500	0.5295	0.3396
0.024	26.1	13000	0.5022	0.3356
0.0262	27.11	13500	0.5320	0.3329
0.0242	28.11	14000	0.5133	0.3326
0.0201	29.12	14500	0.5198	0.3335

Framework versions

• Transformers 4.32.0
• Pytorch 2.0.1+cu118
• Datasets 1.18.3
• Tokenizers 0.13.3