NewSpeechModel-V3.0

This model was trained from scratch on the fleurs dataset. It achieves the following results on the evaluation set:

• Loss: 2.8201
• 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: 0.0003
• train_batch_size: 4
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 10
• total_train_batch_size: 40
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 50
• num_epochs: 1
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
18.0665	0.0960	10	11.8649	1.0
9.3649	0.1919	20	4.3831	1.0
3.9649	0.2879	30	8.1376	1.0
4.9612	0.3839	40	2.9464	1.0
2.8949	0.4798	50	2.8598	1.0
2.9052	0.5758	60	2.8362	1.0
2.8514	0.6718	70	2.8395	1.0
3.1088	0.7678	80	2.8443	1.0
2.8454	0.8637	90	2.8211	1.0
2.8211	0.9597	100	2.8201	1.0

Framework versions

• Transformers 4.42.4
• Pytorch 2.3.1+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1