wav2vec2-base-checkpoint-14

This model is a fine-tuned version of jiobiala24/wav2vec2-base-checkpoint-13 on the common_voice dataset. It achieves the following results on the evaluation set:

• Loss: 1.2822
• Wer: 0.4068

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: 32
• 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
0.1996	1.59	1000	0.7181	0.4079
0.1543	3.17	2000	0.7735	0.4113
0.1171	4.76	3000	0.8152	0.4045
0.0969	6.35	4000	0.8575	0.4142
0.082	7.94	5000	0.9005	0.4124
0.074	9.52	6000	0.9232	0.4151
0.0653	11.11	7000	0.9680	0.4223
0.0587	12.7	8000	1.0633	0.4232
0.0551	14.29	9000	1.0875	0.4171
0.0498	15.87	10000	1.0281	0.4105
0.0443	17.46	11000	1.2164	0.4274
0.0421	19.05	12000	1.1868	0.4191
0.0366	20.63	13000	1.1678	0.4173
0.0366	22.22	14000	1.2444	0.4187
0.0346	23.81	15000	1.2042	0.4169
0.0316	25.4	16000	1.3019	0.4127
0.0296	26.98	17000	1.2001	0.4081
0.0281	28.57	18000	1.2822	0.4068

Framework versions

• Transformers 4.11.3
• Pytorch 1.10.0+cu111
• Datasets 1.13.3
• Tokenizers 0.10.3