GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
AhmedZaky1
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
AhmedZaky1
Model Tree
unsloth/Qwen2.5-VL-7B-Instruct-bnb-4bit
Input Modalities
Output Modalities
Supported Functionality
DIMI-Arabic-OCR is a fine-tuned vision-language model (VLM) specialized for Arabic Optical Character Recognition (OCR).
It extracts printed Arabic text from images with high accuracy — including diacritics (tashkeel) and punctuation.
python
# IMPORTANT: Import unsloth first!import unslothfrom unsloth import FastVisionModelfrom PIL import Imageimport torch # Load the modelmodel, tokenizer = FastVisionModel.from_pretrained( "AhmedZaky1/DIMI-Arabic-OCR", load_in_4bit=True, use_gradient_checkpointing="unsloth",) FastVisionModel.for_inference(model) # Prepare your imageimage = Image.open("/content/2.jpg") # Arabic instructioninstruction = "استخرج النص العربي والأرقام الموجودة في هذه الصورة بدقة عالية جدًا، مع الحفاظ الكامل على الترتيب الأصلي والتنسيق." # Prepare messagesmessages = [ {"role": "user", "content": [ {"type": "image", "image": image}, # Include image here {"type": "text", "text": instruction} ]}] # Apply chat templateinput_text = tokenizer.apply_chat_template( messages, add_generation_prompt=True,) # Tokenize with proper parameters to avoid truncationinputs = tokenizer( text=input_text, images=image, return_tensors="pt", padding=True, truncation=False, max_length=None, ).to("cuda") # Generatewith torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=2048, do_sample=False, temperature=None, top_p=None, pad_token_id=tokenizer.pad_token_id, eos_token_id=tokenizer.eos_token_id, ) # Decode the predictiongenerated_ids = outputs[0][inputs['input_ids'].shape[1]:]prediction = tokenizer.decode(generated_ids, skip_special_tokens=True).strip() print("Extracted Arabic Text:")print(prediction)| Metric | Description | Score (↓ better) |
|---|---|---|
| CER | Character Error Rate | 0.22 |
| WER | Word Error Rate | 0.40 |
Evaluation performed on a 2.6K image test set from combined Arabic OCR datasets (news + diacritics).
Fine-tuned on 26,000 Arabic text images combining:
The dataset covers modern standard Arabic with and without diacritics.
If you use this model, please cite:
bibtex
@misc{dimi-arabic-ocr-2025, author = {Ahmed Zaky}, title = {DIMI-Arabic-OCR: Fine-tuned Qwen2.5-VL for Arabic Text Recognition}, year = {2025}, publisher = {Hugging Face}, howpublished = {\url{https://huggingface.co/AhmedZaky1/DIMI-Arabic-OCR}}}Built with ❤️ by Ahmed Zaky
Advancing Arabic NLP through state-of-the-art embedding models
# IMPORTANT: Import unsloth first!import unslothfrom unsloth import FastVisionModelfrom PIL import Imageimport torch # Load the modelmodel, tokenizer = FastVisionModel.from_pretrained( "AhmedZaky1/DIMI-Arabic-OCR", load_in_4bit=True, use_gradient_checkpointing="unsloth",) FastVisionModel.for_inference(model) # Prepare your imageimage = Image.open("/content/2.jpg") # Arabic instructioninstruction = "استخرج النص العربي والأرقام الموجودة في هذه الصورة بدقة عالية جدًا، مع الحفاظ الكامل على الترتيب الأصلي والتنسيق." # Prepare messagesmessages = [ {"role": "user", "content": [ {"type": "image", "image": image}, # Include image here {"type": "text", "text": instruction} ]}] # Apply chat templateinput_text = tokenizer.apply_chat_template( messages, add_generation_prompt=True,) # Tokenize with proper parameters to avoid truncationinputs = tokenizer( text=input_text, images=image, return_tensors="pt", padding=True, truncation=False, max_length=None, ).to("cuda") # Generatewith torch.no_grad(): outputs = model.generate( **inputs, max_new_tokens=2048, do_sample=False, temperature=None, top_p=None, pad_token_id=tokenizer.pad_token_id, eos_token_id=tokenizer.eos_token_id, ) # Decode the predictiongenerated_ids = outputs[0][inputs['input_ids'].shape[1]:]prediction = tokenizer.decode(generated_ids, skip_special_tokens=True).strip() print("Extracted Arabic Text:")print(prediction)@misc{dimi-arabic-ocr-2025, author = {Ahmed Zaky}, title = {DIMI-Arabic-OCR: Fine-tuned Qwen2.5-VL for Arabic Text Recognition}, year = {2025}, publisher = {Hugging Face}, howpublished = {\url{https://huggingface.co/AhmedZaky1/DIMI-Arabic-OCR}}}