shuff_100

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

• Loss: 322.2777
• Wer: 0.6311
• Cer: 0.2590

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: 0.0005
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.15
• num_epochs: 60.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2829.8126	1.0	3265	346.9778	0.7165	0.2613
761.7557	2.0	6530	242.2776	0.5436	0.2114
672.5873	3.0	9795	339.8657	0.5724	0.2624
668.1343	4.0	13060	278.7672	0.5458	0.2289
674.2274	5.0	16325	305.6685	0.6026	0.2488
707.3097	6.0	19590	285.1494	0.5852	0.2355
741.8663	7.0	22855	350.9450	0.6435	0.2659
764.1505	8.0	26120	320.6730	0.6421	0.2611
814.4658	9.0	29385	319.7635	0.6572	0.2670
842.7828	10.0	32650	375.5847	0.6620	0.2833
827.2463	11.0	35915	329.4790	0.6426	0.2613
812.5209	12.0	39180	322.2777	0.6311	0.2590

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1
• Datasets 3.2.0
• Tokenizers 0.21.0