wav2vec2-base-cynthia-tedlium-2500-v2

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6425
• Wer: 0.2033

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1000
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
0.1196	6.58	500	0.6498	0.2103
0.1176	13.16	1000	0.6490	0.2169
0.1227	19.73	1500	0.6241	0.2127
0.1078	26.31	2000	0.6359	0.2118
0.0956	32.89	2500	0.6330	0.2073
0.1008	39.47	3000	0.6816	0.2036
0.09	46.05	3500	0.6425	0.2033

Framework versions

• Transformers 4.11.3
• Pytorch 1.10.0+cu113
• Datasets 1.13.3
• Tokenizers 0.10.3