wav2vec2-base-timit-demo-colab9

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: 3.1922
• Wer: 1.0

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
5.0683	1.42	500	3.2471	1.0
3.1349	2.85	1000	3.2219	1.0
3.1317	4.27	1500	3.2090	1.0
3.1262	5.7	2000	3.2152	1.0
3.1307	7.12	2500	3.2147	1.0
3.1264	8.55	3000	3.2072	1.0
3.1279	9.97	3500	3.2158	1.0
3.1287	11.4	4000	3.2190	1.0
3.1256	12.82	4500	3.2069	1.0
3.1254	14.25	5000	3.2134	1.0
3.1259	15.67	5500	3.2231	1.0
3.1269	17.09	6000	3.2005	1.0
3.1279	18.52	6500	3.1988	1.0
3.1246	19.94	7000	3.1929	1.0
3.128	21.37	7500	3.1864	1.0
3.1245	22.79	8000	3.1868	1.0
3.1266	24.22	8500	3.1852	1.0
3.1239	25.64	9000	3.1855	1.0
3.125	27.07	9500	3.1917	1.0
3.1233	28.49	10000	3.1929	1.0
3.1229	29.91	10500	3.1922	1.0

Framework versions

• Transformers 4.11.3
• Pytorch 1.11.0+cu113
• Datasets 1.18.3
• Tokenizers 0.10.3