ssc-lke-mms-model-mix-adapt-max-lowlr

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

• Loss: 0.6119
• Cer: 0.1626
• Wer: 0.5822

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: 0.0003
• train_batch_size: 2
• eval_batch_size: 6
• 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
0.65	0.4678	200	0.6444	0.1672	0.5986
0.6463	0.9357	400	0.6297	0.1638	0.5892
0.6016	1.4023	600	0.6275	0.1691	0.6015
0.6109	1.8702	800	0.6211	0.1666	0.5957
0.6679	2.3368	1000	0.6175	0.1661	0.5894
0.533	2.8047	1200	0.6284	0.1644	0.5912
0.6094	3.2713	1400	0.6141	0.1632	0.5866
0.6165	3.7392	1600	0.6135	0.1607	0.5758
0.5445	4.2058	1800	0.6126	0.1619	0.5810
0.5602	4.6737	2000	0.6119	0.1626	0.5822

Framework versions

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