File size: 2,223 Bytes
28239b4
 
37cc7ef
 
 
 
 
 
 
 
 
 
 
 
 
28239b4
82a3181
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---

license: cc-by-nc-nd-4.0
language: [fa]
tags:
  - nlg
  - summarization
  - machine-translation
  - question-generation
  - persian
  - multilingual
pretty_name: Persian NLG Benchmark
task_categories:
  - text-generation
  - summarization
  - translation
---

# Persian NLG

## Dataset Summary

The **Persian NLG Benchmark** is a curated collection of Persian datasets designed to evaluate natural language generation (NLG) capabilities in a variety of tasks, including:

- **Summarization**: Using datasets like **SamSUM-fa** and **PnSummary**  
- **Machine Translation**: With parallel corpora such as **TEP**, **MIZAN**, and **EPOQUE**  
- **Question Generation**: Using the **PersianQA** dataset

This benchmark provides a comprehensive view of how Persian-capable models perform on generative tasks that are linguistically and semantically complex.

---

## Supported Tasks and Leaderboards

### Summarization
The summarization track tests how well models can generate concise and accurate summaries from Persian conversational or formal texts. Relevant columns include input text and human-written summaries.

### Machine Translation
The translation task evaluates models' performance in translating between Persian, English, and Arabic. The datasets cover multiple domains and include source-target sentence pairs. Evaluation can be done in both directions (e.g., fa→en, en→fa).

### Question Generation
The question generation task assesses the ability of models to generate meaningful questions based on a given passage and answer. This tests the contextual and semantic understanding of the models in Persian.

---

## Languages

The primary language is **Persian (fa)**, but the machine translation section includes **English (en)** and **Arabic (ar)** as well.

---

## License

This dataset is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0).

You must give appropriate credit, may not use it for commercial purposes, and may not distribute modified versions of the dataset.

For details, see the [license](LICENSE) file or https://creativecommons.org/licenses/by-nc-nd/4.0/