ssc-hch-mms-model-mix-adapt-max3

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

• Loss: 1.0191
• Cer: 0.1685
• Wer: 0.8244

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
0.3264	0.4739	200	0.9296	0.1810	0.8401
0.3411	0.9479	400	0.9353	0.1776	0.8480
0.2961	1.4218	600	0.9733	0.1791	0.8408
0.3263	1.8957	800	0.9891	0.1707	0.8281
0.2613	2.3697	1000	1.0369	0.1749	0.8384
0.2602	2.8436	1200	1.0332	0.1712	0.8317
0.2309	3.3175	1400	1.0195	0.1725	0.8274
0.236	3.7915	1600	1.0189	0.1723	0.8354
0.2158	4.2654	1800	1.0199	0.1717	0.8276
0.2128	4.7393	2000	1.0191	0.1685	0.8244

Framework versions

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