AceGPT-7B-tuned-ARCD_QA

AceGPT-7B-tuned-ARCD_QA

Overview

AceGPT-7B-tuned-ARCD_QA is a fine-tuned version of the AceGPT model, which itself is based on LLaMA2 and fine-tuned for Arabic language tasks. This model has been further fine-tuned on the ARCD (Arabic Reading Comprehension Dataset) for question-answering tasks.

Model Details

• Base Model: AceGPT-7B (LLaMA2 fine-tuned for Arabic)
• Task: Question Answering
• Language: Arabic
• Training Data: ARCD (Arabic Reading Comprehension Dataset)
• Model Size: 7 billion parameters

Usage

This model is designed for Arabic question-answering tasks. It can be used to answer questions based on given contexts in Arabic.

Example usage with the Transformers library:

from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = "MSebaie/AceGPT-7B-tuned-ARCD_QA"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name)

context = "جمال أحمد حمزة خاشقجي (13 أكتوبر 1958، المدينة المنورة - 2 أكتوبر 2018)، صحفي وإعلامي سعودي، رأس عدّة مناصب لعدد من الصحف في السعودية، وتقلّد منصب مستشار، كما أنّه مدير عام قناة العرب الإخبارية سابقًا."
question = 'من هو جمال أحمد حمزة خاشقجي؟'

input_text = f"Context: {context}\n\nQuestion: {question}\n\n"
inputs = tokenizer(input_text, return_tensors="pt")

outputs = model.generate(**inputs, max_length=100)
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)

print(answer)