File size: 5,871 Bytes
603f6e2
 
 
 
 
 
 
 
 
5ce778f
9388c5f
5ce778f
1bb7cfe
 
99f8008
1bb7cfe
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
603f6e2
403af72
 
5ce778f
403af72
5ce778f
 
 
403af72
5ce778f
29e5c25
5ce778f
29e5c25
5ce778f
29e5c25
 
e2a3756
29e5c25
5ce778f
29e5c25
 
 
5ce778f
 
 
 
29e5c25
5ce778f
29e5c25
403af72
 
 
 
 
 
 
 
 
 
86100db
 
403af72
e2a3756
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
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
---
license: mit
task_categories:
- question-answering
language:
- en
tags:
- realtime
- news
configs:
  - config_name: default
    data_files:
      - split: relive
        path:
        - 20250211_qa_public.jsonl
        - 20250115_qa_public.jsonl
        - 20240705_qa_public.jsonl
        - 20240629_qa_public.jsonl
        - 20240620_qa_public.jsonl
        - 20240613_qa_public.jsonl
        - 20240606_qa_public.jsonl
        - 20240525_qa_public.jsonl
        - 20240519_qa_public.jsonl
        - 20240518_qa_public.jsonl
      - split: realtime
        path:
        - latest/20240119_qa_public.jsonl
        - past/2024/20240112_qa.jsonl
        - past/2024/20240105_qa.jsonl
        - past/2023/20231222_qa.jsonl
        - past/2023/20231215_qa.jsonl
        - past/2023/20231208_qa.jsonl
        - past/2023/20231201_qa.jsonl
        - past/2023/20231124_qa.jsonl
        - past/2023/20231117_qa.jsonl
        - past/2023/20231110_qa.jsonl
        - past/2023/20231103_qa.jsonl
        - past/2023/20231027_qa.jsonl
        - past/2023/20231020_qa.jsonl
        - past/2023/20231013_qa.jsonl
        - past/2023/20231006_qa.jsonl
        - past/2023/20230929_qa.jsonl
        - past/2023/20230922_qa.jsonl
        - past/2023/20230915_qa.jsonl
        - past/2023/20230908_qa.jsonl
        - past/2023/20230901_qa.jsonl
        - past/2023/20230825_qa.jsonl
        - past/2023/20230818_qa.jsonl
        - past/2023/20230811_qa.jsonl
        - past/2023/20230804_qa.jsonl
        - past/2023/20230728_qa.jsonl
        - past/2023/20230721_qa.jsonl
        - past/2023/20230714_qa.jsonl
        - past/2023/20230707_qa.jsonl
        - past/2023/20230630_qa.jsonl
        - past/2023/20230623_qa.jsonl
        - past/2023/20230616_qa.jsonl
        - past/2023/20230609_qa.jsonl
        - past/2023/20230602_qa.jsonl
        - past/2023/20230526_qa.jsonl
        - past/2023/20230519_qa.jsonl
        - past/2023/20230512_qa.jsonl
        - past/2023/20230505_qa.jsonl
        - past/2023/20230428_qa.jsonl
        - past/2023/20230421_qa.jsonl
        - past/2023/20230414_qa.jsonl
        - past/2023/20230407_qa.jsonl
        - past/2023/20230331_qa.jsonl
        - past/2023/20230324_qa.jsonl
        - past/2023/20230317_qa.jsonl
        - past/2023/20230310_qa.jsonl
        - past/2023/20230303_qa.jsonl
        - past/2023/20230224_qa.jsonl
        - past/2023/20230217_qa.jsonl
        - past/2023/20230210_qa.jsonl
        - past/2023/20230203_qa.jsonl
        - past/2023/20230127_qa.jsonl
        - past/2023/20230120_qa.jsonl
        - past/2023/20230113_qa.jsonl
        - past/2023/20230106_qa.jsonl
        - past/2022/20221230_qa.jsonl
        - past/2022/20221223_qa.jsonl
        - past/2022/20221216_qa.jsonl
        - past/2022/20221209_qa.jsonl
        - past/2022/20221202_qa.jsonl
        - past/2022/20221125_qa.jsonl
        - past/2022/20221118_qa.jsonl
        - past/2022/20221111_qa.jsonl
        - past/2022/20221104_qa.jsonl
        - past/2022/20221028_qa.jsonl
        - past/2022/20221021_qa.jsonl
        - past/2022/20221021_qa_public.jsonl
        - past/2022/20221014_qa.jsonl
        - past/2022/20220617-20221014_qa.jsonl
        - past/2022/20221007_qa.jsonl
        - past/2022/20220930_qa.jsonl
        - past/2022/20220923_qa.jsonl
        - past/2022/20220916_qa.jsonl
        - past/2022/20220909_qa.jsonl
        - past/2022/20220902_qa.jsonl
        - past/2022/20220826_qa.jsonl
        - past/2022/20220617-20220826_qa.jsonl
        - past/2022/20220819_qa.jsonl
        - past/2022/20220812_qa.jsonl
        - past/2022/20220805_qa.jsonl
        - past/2022/20220729_qa.jsonl
        - past/2022/20220722_qa.jsonl
        - past/2022/20220617-20220722_qa.jsonl
        - past/2022/20220715_qa.jsonl
        - past/2022/20220617-20220715_qa.jsonl
        - past/2022/20220708_qa.jsonl
        - past/2022/20220617-20220708_qa.jsonl
        - past/2022/20220701_qa.jsonl
        - past/2022/20220624_qa.jsonl
        - past/2022/20220617-20220624_qa.jsonl
        - past/2022/20220617_qa.jsonl
---
# relive-qa

Using RealtimeQA as a starting point for new articles + Q&A using a semi-automated format.

Also see:
- https://github.com/realtimeqa/realtimeqa_public (ended Jan 2024?)
- https://github.com/freshllms/freshqa (ongoing, last Dec 2024)

## Scraper process

Prerequisites: `pip install openai lxml cssselect requests xmltodict` and OpenAI API key

I've added these scripts:
- scrape.py : base script to load plain text from the latest WikiNews articles
- scrape_with_openai.py : pass scraped text to OpenAI's GPT-4o to generate questions and answers for each article
- scrape_morerecent_with_openai.py : scrape recent articles which WikiNews hasn't yet published from `Category:May 2024`, then use OpenAI's GPT-4o for Q&A

An LLM evaluated on this Q&A could read the WikiNews summary, the links collected from the story, and/or do its own web searches.

### Issues

- Prompt should discourage Q&A which are obvious, or stand out from alternative answers.
- Q&A should be based on new information in the article, and not general knowledge.
- Links and the article title could give away the answer as the subject of the article, rather than using reading comprehension.
- WikiNews articles may be niche / local stories, where facts are not known to an LLM unless it reads the specific article

## Original paper


Citation:

```
@inproceedings{
kasai2023realtime,
title={RealTime {QA}: What's the Answer Right Now?},
author={Jungo Kasai and Keisuke Sakaguchi and yoichi takahashi and Ronan Le Bras and Akari Asai and Xinyan Velocity Yu and Dragomir Radev and Noah A. Smith and Yejin Choi and Kentaro Inui},
booktitle={Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
year={2023},
eprint={2207.13332},
url={https://openreview.net/forum?id=HfKOIPCvsv}
}
```