wav2vec2-base-checkpoint-6

This model is a fine-tuned version of jiobiala24/wav2vec2-base-checkpoint-5 on the common_voice dataset. It achieves the following results on the evaluation set:

• Loss: 0.9738
• Wer: 0.3323

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
0.3435	1.82	1000	0.5637	0.3419
0.2599	3.65	2000	0.5804	0.3473
0.2043	5.47	3000	0.6481	0.3474
0.1651	7.3	4000	0.6937	0.3452
0.1376	9.12	5000	0.7221	0.3429
0.118	10.95	6000	0.7634	0.3441
0.105	12.77	7000	0.7789	0.3444
0.0925	14.6	8000	0.8209	0.3444
0.0863	16.42	9000	0.8293	0.3440
0.0756	18.25	10000	0.8553	0.3412
0.0718	20.07	11000	0.9006	0.3430
0.0654	21.9	12000	0.9541	0.3458
0.0605	23.72	13000	0.9400	0.3350
0.0552	25.55	14000	0.9547	0.3363
0.0543	27.37	15000	0.9715	0.3348
0.0493	29.2	16000	0.9738	0.3323

Framework versions

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