ssc-koo-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: 0.9833
• Cer: 0.2078
• Wer: 0.7198

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
0.5519	0.2380	200	1.0491	0.2181	0.7486
0.6695	0.4759	400	1.0794	0.2235	0.7537
0.6281	0.7139	600	0.9872	0.2197	0.7597
1.0204	0.9518	800	1.0147	0.2269	0.8161
0.7552	1.1892	1000	0.9663	0.2169	0.7399
0.5822	1.4271	1200	0.9698	0.2201	0.7370
0.7803	1.6651	1400	0.9582	0.2174	0.7369
0.5241	1.9030	1600	1.0238	0.2145	0.7384
0.6677	2.1404	1800	0.9963	0.2176	0.7454
0.5136	2.3783	2000	0.9709	0.2121	0.7340
0.5906	2.6163	2200	0.9465	0.2165	0.7346
0.4955	2.8543	2400	0.9719	0.2153	0.7393
0.5493	3.0916	2600	1.0170	0.2098	0.7257
0.5162	3.3296	2800	1.0054	0.2112	0.7302
0.5153	3.5675	3000	0.9765	0.2103	0.7255
0.4576	3.8055	3200	0.9748	0.2113	0.7238
0.5859	4.0428	3400	0.9813	0.2115	0.7242
0.5492	4.2808	3600	0.9603	0.2117	0.7283
0.5105	4.5187	3800	0.9731	0.2130	0.7323
0.5012	4.7567	4000	0.9905	0.2090	0.7310
0.4848	4.9946	4200	0.9833	0.2078	0.7198

Framework versions

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