xls-r-300m-toratan-120

This model is a fine-tuned version of Benjaminpwh/xlsr-toratan-240-copt-base_K on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0121
• Cer: 0.0036

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: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: 500
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Cer
4.4013	4.1667	400	1.6662	0.4168
1.4361	8.3333	800	0.7861	0.2370
0.8924	12.5	1200	0.4475	0.1519
0.575	16.6667	1600	0.2061	0.0790
0.353	20.8333	2000	0.0794	0.0311
0.2164	25.0	2400	0.0308	0.0113
0.1399	29.1667	2800	0.0121	0.0036

Framework versions

• Transformers 4.52.0.dev0
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1