This is a fine-tuned and 4-bit quantized version of the Qwen2.5-VL-3B vision-language model specialized for Arabic handwritten OCR. It accepts images containing handwritten Arabic text and outputs transcribed text. The model is hosted on Hugging Face for easy integration into OCR pipelines and runs efficiently on consumer hardware due to quantization.
Arabic Handwritten OCR 4bit Qwen2.5 VL 3B sits in PulseGate's Foundation models & chat category. Accurately recognizing and transcribing handwritten Arabic text from images using a lightweight quantized vision-language model. It is built as an open-source project for developers. The project is open source (Open Source). It runs on the web and API.
It is developed by sherif1313, and it first shipped in 2025. Key capabilities include Handwritten OCR, Arabic Language, and 4-bit Quantized.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do