Spaces:
Sleeping
Sleeping
Upload 3 files
Browse filesinitial demo upload
- app.py +20 -0
- readme.md +33 -0
- requirements.txt +3 -0
app.py
ADDED
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
3 |
+
|
4 |
+
model = AutoModelForSeq2SeqLM.from_pretrained("drelhaj/FinAraT5")
|
5 |
+
tokenizer = AutoTokenizer.from_pretrained("drelhaj/FinAraT5")
|
6 |
+
|
7 |
+
def summarize(text):
|
8 |
+
inputs = tokenizer(text, return_tensors="pt", truncation=True)
|
9 |
+
outputs = model.generate(**inputs, max_length=64)
|
10 |
+
return tokenizer.decode(outputs[0], skip_special_tokens=True)
|
11 |
+
|
12 |
+
demo = gr.Interface(
|
13 |
+
fn=summarize,
|
14 |
+
inputs=gr.Textbox(lines=5, placeholder="أدخل النص المالي هنا مع بادئة مثل 'لخص:'", label="النص المالي"),
|
15 |
+
outputs=gr.Textbox(label="الملخص"),
|
16 |
+
title="FinAraT5: ملخصات نصوص مالية باللغة العربية",
|
17 |
+
description="أدخل نصًا ماليًا باللغة العربية مبدوءًا بـ 'لخص:' للحصول على ملخص باستخدام نموذج FinAraT5."
|
18 |
+
)
|
19 |
+
|
20 |
+
demo.launch()
|
readme.md
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# 🧠 FinAraT5 – Arabic Financial Text Summarisation Demo
|
2 |
+
|
3 |
+
This Space provides an interactive demo of **[FinAraT5](https://huggingface.co/drelhaj/FinAraT5)** — a T5-based model fine-tuned for Arabic financial text summarisation.
|
4 |
+
|
5 |
+
You can enter Arabic financial news or reports prefixed with **"لخص:"** to generate concise summaries.
|
6 |
+
|
7 |
+
---
|
8 |
+
|
9 |
+
## ✨ Example Input
|
10 |
+
|
11 |
+
لخص: أعلنت الشركة عن ارتفاع أرباحها بنسبة ١٥٪ في الربع الثاني من العام نتيجة لزيادة المبيعات في السوق الخليجية.
|
12 |
+
|
13 |
+
|
14 |
+
## 📤 Example Output
|
15 |
+
|
16 |
+
الشركة تحقق نمواً بنسبة ١٥٪ في أرباح الربع الثاني بفضل المبيعات الخليجية.
|
17 |
+
|
18 |
+
|
19 |
+
---
|
20 |
+
|
21 |
+
## 🏗️ Powered by
|
22 |
+
|
23 |
+
- [`drelhaj/FinAraT5`](https://huggingface.co/drelhaj/FinAraT5)
|
24 |
+
- [Transformers (🤗)](https://huggingface.co/transformers/)
|
25 |
+
- [Gradio](https://gradio.app)
|
26 |
+
|
27 |
+
---
|
28 |
+
|
29 |
+
## 📄 Citation
|
30 |
+
|
31 |
+
If you use this demo or model in your work, please cite the official paper:
|
32 |
+
|
33 |
+
> Nadhem Zmandar, Mo El-Haj, and Paul Rayson. 2023. FinAraT5: A text to text model for financial Arabic text understanding and generation. In Proceedings of the 4th Conference on Language, Data and Knowledge, pages 262–273, Vienna, Austria. NOVA CLUNL, Portugal https://aclanthology.org/2023.ldk-1.25/.
|
requirements.txt
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
transformers
|
2 |
+
torch
|
3 |
+
gradio
|