ssc-kcn-mms-model-mix-adapt-max3-devtrain

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

• Loss: 0.7720
• Cer: 0.1878
• Wer: 0.4832

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: 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
0.6953	0.2867	200	0.9526	0.2109	0.5338
0.7399	0.5735	400	0.9448	0.2104	0.5283
0.736	0.8602	600	0.8170	0.1996	0.5173
0.6204	1.1462	800	0.8343	0.2048	0.5133
0.77	1.4330	1000	0.9082	0.2060	0.5184
0.6475	1.7197	1200	0.7871	0.2086	0.5469
0.7394	2.0057	1400	0.8106	0.2012	0.5046
0.6703	2.2925	1600	0.8129	0.1991	0.4990
0.5731	2.5792	1800	0.7972	0.1989	0.5058
0.6649	2.8659	2000	0.7763	0.1940	0.4984
0.6588	3.1520	2200	0.8051	0.1953	0.4982
0.6042	3.4387	2400	0.7961	0.1877	0.4847
0.5617	3.7254	2600	0.7986	0.1898	0.4824
0.5678	4.0115	2800	0.7434	0.1900	0.4954
0.598	4.2982	3000	0.7335	0.1911	0.4956
0.5983	4.5849	3200	0.7530	0.1876	0.4861
0.5272	4.8717	3400	0.7720	0.1878	0.4832

Framework versions

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