File size: 5,995 Bytes
88b0b43
 
 
 
 
 
 
ca9fb7b
8112f92
318ace7
121a293
 
11f6ab1
 
ca9fb7b
88b0b43
 
 
 
 
 
 
 
 
 
dd1c3d4
88b0b43
dd1c3d4
88b0b43
 
dd1c3d4
88b0b43
 
 
 
 
 
ca9fb7b
 
 
 
c0634e7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0b3aecb
 
 
 
01e61ab
c751006
01e61ab
 
 
 
 
 
 
0b3aecb
ca9fb7b
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
---
language:
- th
pipeline_tag: summarization
tags:
- summarization
- pegasus_x
widget:

  - text: >-
      หลังจากมีการทดสอบนับหลายร้อยครั้งเพื่อตรวจสอบการปรากฏตัวของเขม่าดำหรือมลพิษอื่นๆ ตามแนวเทือกเขาแอนดีส ที่ตั้งขนานทอดยาวประมาณ 7,000 กิโลเมตรไปตามแนวชายฝั่งตะวันตกของละตินอเมริกา ล่าสุด ผู้เชี่ยวชาญด้านสิ่งแวดล้อมจากมหาวิทยาลัยซานติอาโก ในประเทศชิลี เผยว่า หิมะที่ปกคลุมเทือกเขาแอนดีสน่าจะมีความสะอาดพอๆกันกับหิมะในแถบอาร์กติกของประเทศแคนาดา,การวิจัยดังกล่าวเป็นส่วนหนึ่งของโครงการตรวจสอบการปรากฏตัวของเขม่าคาร์บอนสีดำหรือเขม่าดำ (black carbon) ในดินแดนที่เรียกว่า หิมะภาค (cryosphere) ซึ่งเป็นพื้นที่ส่วนหนึ่งของโลกที่มีทั้งหิมะและน้ำแข็งอยู่ใต้ผิวโลกและบนผิวโลก พื้นที่เย็นยะเยือกที่เต็มไปด้วยหิมะของเทือกเขาแอนดีสก็จัดว่าเป็นหิมะภาคแห่งหนึ่ง นักวิจัยพบว่าที่แห่งนี้มีเขม่าดำในปริมาณต่ำกว่า 14 นาโนกรัมต่อหิมะ 1 กรัม บ่งชี้ว่าอัตราส่วนเปรียบเทียบค่าการสะท้อนแสงของพื้นผิวกับปริมาณรังสีทั้งหมดที่ตกกระทบของหิมะมีน้อยกว่า 2% เปรียบเทียบแล้วก็นับว่าอยู่ในระดับเดียวกับหิมะในแถบรัฐอลาสกาของสหรัฐอเมริกา หรือพื้นที่อาร์กติกในแคนาดา,ทั้งนี้ เขม่าดำนั้นถูกปล่อยออกมาจากการเผาไหม้ของน้ำมัน ดีเซล เชื้อเพลิงจากไม้ และเชื้อเพลิงจากซากดึกดำบรรพ์หรือฟอสซิล (fossil) เป็นตัวที่ทำให้พื้นผิวหิมะหม่นดำและความสามารถในการสะท้อนแสงของพื้นผิวลดต่ำลง นั่นหมายความว่าหิมะจะละลายเร็วขึ้น ซึ่งการศึกษาความเข้มข้นของเขม่าดำอาจมีความสำคัญต่อการคาดการณ์อัตราการละลายของหิมะและน้ำแข็งในอนาคตได้.
    example_title: Test
---


This repository features a fine-tuned Pegasus X model designed for summarizing Thai text. The architecture of the model is based on the Pegasus X model.

# Library 

```
pip install transformers
```

# Example
```python
from transformers import PegasusXForConditionalGeneration, AutoTokenizer

model = PegasusXForConditionalGeneration.from_pretrained("satjawat/pegasus-x-thai-sum")
tokenizer = AutoTokenizer.from_pretrained("satjawat/pegasus-x-thai-sum")

new_input_string = "ข้อความ"
new_input_ids = tokenizer(new_input_string.lower(), return_tensors="pt").input_ids
summary_ids = model.generate(new_input_ids, max_length=50, num_beams=6, length_penalty=2.0, early_stopping=True)
summary = tokenizer.decode(summary_ids[0], skip_special_tokens=True)

print("Input:", new_input_string)
print("Generated Summary:", summary)
```

#  Training hyperparameters 
The following hyperparameters were used during training:
- accumulation_steps: 2
- num_epochs: 20
- num_beams: 6
- learning_rate: lr=5e-5
- optimizer: AdamW with betas=(0.9,0.999) and epsilon=1e-08
- activation_function: gelu
- add_bias_logits: True
- normalize_embedding: True
- add_final_layer_norm: False
- normalize_before: False

# Score
Evaluate the model with the test dataset of <a href="https://github.com/nakhunchumpolsathien/ThaiSum">ThaiSum</a>, consisting of a total of 11,000 articles, with the following scores:
- Rouge1: 0.490279		
- Rouge2: 0.289839
- Rougel: 0.489334

# Resource Funding
NSTDA Supercomputer center (ThaiSC) and the National e-Science Infrastructure Consortium for their support of computer facilities.

# Citation
If you use "satjawat/pegasus-x-thai-sum" in your project or publication, please cite the model as follows:

```
ปรีชานนท์ ชาติไทย และ สัจจวัจน์ ส่งเสริม. (2567),
การสรุปข้อความข่าวภาษาไทยด้วยโครงข่ายประสาทเทียม (Thai News Text Summarization Using Neural Network),
วิทยาศาสตรบัณฑิต (วทบ.):ขอนแก่น, มหาวิทยาลัยขอนแก่น
```