aquaqamelob/LightOnOCR-2-ft-polish_synthetic

<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->

LightOnOCR-2-ft-polish_synthetic

This model is a fine-tuned version of lightonai/LightOnOCR-2-1B on the None dataset. It achieves the following results on the evaluation set:

• —eval_loss: 0.0056
• —eval_runtime: 27.716
• —evalsamplesper_second: 7.216
• —evalstepsper_second: 1.227
• —epoch: 1.0038
• —step: 800

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: 0.0001
• —trainbatchsize: 4
• —evalbatchsize: 6
• —seed: 42
• —gradientaccumulationsteps: 4
• —totaltrainbatch_size: 16
• —optimizer: Use OptimizerNames.ADAMWTORCHFUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• —lrschedulertype: cosine
• —lrschedulerwarmup_steps: 0.05
• —num_epochs: 3

Framework versions

• —PEFT 0.19.1
• —Transformers 5.0.0
• —Pytorch 2.10.0+cu128
• —Datasets 4.8.5
• —Tokenizers 0.22.2