ssc-bxk-mms-model-mix-adapt-max

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

• Loss: 0.5747
• Cer: 0.1533
• Wer: 0.5485

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: 2
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 4
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED 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
0.492	0.4872	200	0.6465	0.1689	0.6131
0.4916	0.9744	400	0.6220	0.1598	0.5790
0.4791	1.4604	600	0.6018	0.1567	0.5703
0.4661	1.9476	800	0.6045	0.1590	0.5910
0.4444	2.4336	1000	0.5895	0.1581	0.5696
0.4369	2.9208	1200	0.5877	0.1558	0.5683
0.4305	3.4068	1400	0.5883	0.1565	0.5593
0.4214	3.8940	1600	0.5816	0.1558	0.5549
0.4028	4.3800	1800	0.5771	0.1550	0.5607
0.4118	4.8672	2000	0.5747	0.1533	0.5485

Framework versions

• Transformers 4.57.2
• Pytorch 2.9.1+cu128
• Datasets 3.6.0
• Tokenizers 0.22.0