ssc-led-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: 0.5913
• Cer: 0.1439
• Wer: 0.3363

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
0.6596	0.5362	200	0.7025	0.1633	0.3719
0.5722	1.0724	400	0.6537	0.1549	0.3569
0.5542	1.6086	600	0.6368	0.1519	0.3518
0.5215	2.1448	800	0.6442	0.1535	0.3566
0.4808	2.6810	1000	0.6213	0.1483	0.3447
0.4893	3.2172	1200	0.6082	0.1484	0.3395
0.4566	3.7534	1400	0.6005	0.1456	0.3407
0.4779	4.2895	1600	0.5993	0.1465	0.3396
0.442	4.8257	1800	0.5913	0.1439	0.3363

Framework versions

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