JustANormalTinkerer/hayai-ocr-v2

Hayai OCR v2

A compact 155M-parameter OCR model that maps images directly to transcribed text. Pairs a SigLIP2 vision encoder with a causal transformer decoder to read text directly out of images without a separate detection stage.

Supports Japanese, Chinese, and English text recognition.

Model

• —Repository: JustANormalTinkerer/hayai-ocr-v2
• —Architecture: SigLIP2 vision encoder + 12-layer causal transformer decoder
• —License: Apache 2.0

Example images

Example images are sourced from the hayai-finetuning-dataset (Apache 2.0).