Gresham commited on
Commit
731822f
·
1 Parent(s): c733ceb

docs(readme): refactor

Browse files
Files changed (1) hide show
  1. README.md +93 -0
README.md CHANGED
@@ -41,8 +41,101 @@ Github Repo: https://github.com/TreeAI-Lab/NumericBench
41
 
42
  ### Loading Data
43
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
44
  ### Data Format
45
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
46
  ### Evaluation
47
 
48
  # Dataset statistics
 
41
 
42
  ### Loading Data
43
 
44
+ ``` python
45
+ from huggingface_hub import hf_hub_download
46
+ import pandas as pd
47
+
48
+ dataset_name_list = ['arithmetic_operation/context_arithmetic_operation.json', 'arithmetic_operation/arithmetic_operation.json',
49
+ 'mixed_number_sting/mixed_number_string_500_per_sample.json',
50
+ 'num_list/num_list_500_per_sample_1000_length.json', 'num_list/num_list_500_per_sample_100_length.json',
51
+ 'sequence/sequence_500_sample_100_length.json',
52
+ 'stock/single-turn/stock_500_per_sample_150_length.json', 'stock/single-turn/stock_500_per_sample_300_length.json', 'stock/multi-turn/stock_multi_turn_100_per_sample_100_length.json',
53
+ 'weather/single-turn/weather_500_per_sample_200_length.json', 'weather/single-turn/weather_500_per_sample_400_length.json', 'weather/multi-turn/weather_multi_turn_100_per_sample_100_length.json'
54
+ ]
55
+
56
+ REPO_ID = "TreeAILab/NumericBench"
57
+
58
+ for dataset_name in dataset_name_list:
59
+ dataset = pd.read_json(
60
+ hf_hub_download(repo_id=REPO_ID, filename=dataset_name, repo_type="dataset")
61
+ )
62
+ ```
63
+
64
+ Alternatively, you can download the dataset from [this link](https://zenodo.org/records/14875784?token=eyJhbGciOiJIUzUxMiJ9.eyJpZCI6IjZhYjZlYzYwLWZkMTgtNGU1Ni1iM2I2LWUwNWVlMGIwZmYwZCIsImRhdGEiOnt9LCJyYW5kb20iOiIwNzQxNmU0NWY5ZDkxMzQ4ODVmYjdlZDgyOGJmNjVhYSJ9.m-_8Owb3TohJ76cGt2Mu4wrpsxMgC4E_aJG7Q07KHOTlKaxB4kipMY3eBZPaQEIkOv_iJEkRmlvZr23rLkTMBw).
65
+
66
  ### Data Format
67
 
68
+ Due to the excessive length of the content, "..." is used to indicate omission.
69
+
70
+ ##### single-turn
71
+
72
+ ``` json
73
+ {
74
+ "system_prompt": "...",
75
+ "system_prompt_cot": "...",
76
+ "description": "...",
77
+ "data": [
78
+ {
79
+ "idx": 0,
80
+ "question_index": 0,
81
+ "question": "What is the result of A + B? Please round the answer to two decimal places. ",
82
+ "struct_data": "{'A': 6.755, 'B': -1.225}",
83
+ "answer": 5.53,
84
+ "ability": "1-digit integer with 3 decimal places"
85
+ },
86
+ {
87
+ "idx": 1,
88
+ "question_index": 1,
89
+ "question": "What is the result of A - B? Please round the answer to two decimal places. ",
90
+ "struct_data": "{'A': 6.755, 'B': -1.225}",
91
+ "answer": 7.98,
92
+ "ability": "1-digit integer with 3 decimal places"
93
+ }
94
+ ]
95
+ }
96
+ ```
97
+
98
+ ##### multi-turn
99
+
100
+ ``` json
101
+ {
102
+ "system_prompt": "...",
103
+ "system_prompt_cot": "...",
104
+ "description": "...",
105
+ "data": [
106
+ {
107
+ "idx": 0,
108
+ "multi_turn_QA": [
109
+ {
110
+ "turn_index": 0,
111
+ "question_index": 6,
112
+ "struct_data": "...",
113
+ "question": "...",
114
+ "answer": "F",
115
+ "ability": "contextual retrieval"
116
+ },
117
+ {
118
+ "turn_index": 1,
119
+ "question_index": 6,
120
+ "struct_data": "...",
121
+ "question": "...",
122
+ "answer": "B",
123
+ "ability": "contextual retrieval"
124
+ },
125
+ {
126
+ "turn_index": 2,
127
+ "question_index": 6,
128
+ "struct_data": "...",
129
+ "question": "...",
130
+ "answer": "A",
131
+ "ability": "contextual retrieval"
132
+ }
133
+ ]
134
+ }
135
+ ]
136
+ }
137
+ ```
138
+
139
  ### Evaluation
140
 
141
  # Dataset statistics