ssc-meh-mms-model-mix-adapt-max2

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

• Loss: 0.9357
• Cer: 0.1848
• Wer: 0.5245

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: 0.001
• train_batch_size: 1
• eval_batch_size: 6
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 2
• 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_steps: 100
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer	Wer
0.3852	0.6231	200	0.9490	0.1866	0.5388
0.3834	1.2461	400	0.8982	0.1913	0.5373
0.3539	1.8692	600	0.9107	0.1877	0.5265
0.3641	2.4922	800	0.9277	0.1873	0.5250
0.3281	3.1153	1000	0.9339	0.1895	0.5360
0.3245	3.7383	1200	0.9323	0.1871	0.5261
0.2834	4.3614	1400	0.9389	0.1841	0.5205
0.278	4.9844	1600	0.9357	0.1848	0.5245

Framework versions

• Transformers 4.52.1
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.21.4