add-meta (#2)
Browse files- Add metadatas for documents and questions (64dc6970f21a86dff1bf70df56212383ca502a33)
- Update README.md (0d95ecc19681e46789bab9744e82e7ed25556e4b)
README.md
CHANGED
@@ -11,6 +11,102 @@ source_datasets:
|
|
11 |
- extended|allganize/RAG-Evaluation-Dataset-KO
|
12 |
task_categories:
|
13 |
- other
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
14 |
---
|
15 |
|
16 |
# Dataset Card for Reconstructed RAG Evaluation Dataset (KO)
|
@@ -75,7 +171,7 @@ The dataset is in Korean (`ko`).
|
|
75 |
* 원본 데이터에서 `allganize/RAG-Evaluation-Dataset-KO`, `documents.csv`와 유사한 구조로 데이터를 로드하는 코드.
|
76 |
* 분리된 pdf binary는 병합하여 로컬 경로에 pdf파일로 저장함.
|
77 |
|
78 |
-
```
|
79 |
import os
|
80 |
from typing import List
|
81 |
from itertools import groupby
|
@@ -147,4 +243,9 @@ def load_rag_eval_data(target_pdf_dir:str):
|
|
147 |
}
|
148 |
|
149 |
return qa_df, pdf_meta_dict
|
|
|
|
|
|
|
|
|
|
|
150 |
```
|
|
|
11 |
- extended|allganize/RAG-Evaluation-Dataset-KO
|
12 |
task_categories:
|
13 |
- other
|
14 |
+
dataset_info:
|
15 |
+
- config_name: default
|
16 |
+
features:
|
17 |
+
- name: __key__
|
18 |
+
dtype: string
|
19 |
+
- name: __url__
|
20 |
+
dtype: string
|
21 |
+
- name: json
|
22 |
+
struct:
|
23 |
+
- name: domain
|
24 |
+
dtype: string
|
25 |
+
- name: file_name
|
26 |
+
dtype: string
|
27 |
+
- name: page_number
|
28 |
+
dtype: int64
|
29 |
+
- name: pages
|
30 |
+
dtype: int64
|
31 |
+
- name: pid
|
32 |
+
dtype: string
|
33 |
+
- name: test_cases
|
34 |
+
list:
|
35 |
+
- name: context_type
|
36 |
+
dtype: string
|
37 |
+
- name: pid
|
38 |
+
dtype: string
|
39 |
+
- name: qid
|
40 |
+
dtype: string
|
41 |
+
- name: question
|
42 |
+
dtype: string
|
43 |
+
- name: target_answer
|
44 |
+
dtype: string
|
45 |
+
- name: target_page_no
|
46 |
+
dtype: string
|
47 |
+
- name: pdf
|
48 |
+
dtype: binary
|
49 |
+
splits:
|
50 |
+
- name: test
|
51 |
+
num_bytes: 460216362
|
52 |
+
num_examples: 1450
|
53 |
+
download_size: 429930358
|
54 |
+
dataset_size: 460216362
|
55 |
+
- config_name: documents_meta
|
56 |
+
features:
|
57 |
+
- name: pid
|
58 |
+
dtype: string
|
59 |
+
- name: domain
|
60 |
+
dtype: string
|
61 |
+
- name: file_name
|
62 |
+
dtype: string
|
63 |
+
- name: pages
|
64 |
+
dtype: int64
|
65 |
+
- name: qid_count
|
66 |
+
dtype: int64
|
67 |
+
- name: qid_list
|
68 |
+
sequence: string
|
69 |
+
splits:
|
70 |
+
- name: meta
|
71 |
+
num_bytes: 9762
|
72 |
+
num_examples: 64
|
73 |
+
download_size: 8728
|
74 |
+
dataset_size: 9762
|
75 |
+
- config_name: questions_meta
|
76 |
+
features:
|
77 |
+
- name: qid
|
78 |
+
dtype: string
|
79 |
+
- name: question
|
80 |
+
dtype: string
|
81 |
+
- name: pid
|
82 |
+
dtype: string
|
83 |
+
- name: file_name
|
84 |
+
dtype: string
|
85 |
+
- name: target_page_no
|
86 |
+
dtype: string
|
87 |
+
- name: context_type
|
88 |
+
dtype: string
|
89 |
+
- name: target_answer
|
90 |
+
dtype: string
|
91 |
+
splits:
|
92 |
+
- name: meta
|
93 |
+
num_bytes: 192716
|
94 |
+
num_examples: 300
|
95 |
+
download_size: 102039
|
96 |
+
dataset_size: 192716
|
97 |
+
configs:
|
98 |
+
- config_name: default
|
99 |
+
data_files:
|
100 |
+
- split: test
|
101 |
+
path: data/test-*
|
102 |
+
- config_name: documents_meta
|
103 |
+
data_files:
|
104 |
+
- split: meta
|
105 |
+
path: documents_meta/meta-*
|
106 |
+
- config_name: questions_meta
|
107 |
+
data_files:
|
108 |
+
- split: meta
|
109 |
+
path: questions_meta/meta-*
|
110 |
---
|
111 |
|
112 |
# Dataset Card for Reconstructed RAG Evaluation Dataset (KO)
|
|
|
171 |
* 원본 데이터에서 `allganize/RAG-Evaluation-Dataset-KO`, `documents.csv`와 유사한 구조로 데이터를 로드하는 코드.
|
172 |
* 분리된 pdf binary는 병합하여 로컬 경로에 pdf파일로 저장함.
|
173 |
|
174 |
+
```python
|
175 |
import os
|
176 |
from typing import List
|
177 |
from itertools import groupby
|
|
|
243 |
}
|
244 |
|
245 |
return qa_df, pdf_meta_dict
|
246 |
+
|
247 |
+
|
248 |
+
# Load Metadata
|
249 |
+
documents_meta = datasets.load_dataset("datalama/RAG-Evaluation-Dataset-KO", name="documents_meta") # 64개 문서에 대한 메타데이터
|
250 |
+
questions_meta = datasets.load_dataset("datalama/RAG-Evaluation-Dataset-KO", split="questions_meta") # 300개 질문에 대한 메타데이터
|
251 |
```
|
data/test-00000-of-00001.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cb1b2cf0110aeab7775a749fb36bd7114d5a3c123286b587542c02f962b9891c
|
3 |
+
size 429930358
|
documents_meta/meta-00000-of-00001.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:14ada05522ef03d62b12f7d9f88683d83bd4fd88c3f983e3947538366d2ce1f4
|
3 |
+
size 8728
|
questions_meta/meta-00000-of-00001.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:95e9a9c53db2276b2c5bd4066b0f5178ef2a62e51785a0a0de439378a10f13b1
|
3 |
+
size 102039
|