yilunzhao commited on
Commit
e91eb6a
·
1 Parent(s): ee775b9

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +7 -13
README.md CHANGED
@@ -8,6 +8,8 @@ dataset_info:
8
  dtype: string
9
  - name: tables
10
  dtype: sequence
 
 
11
  - name: topic
12
  dtype: string
13
  - name: python_solution
@@ -15,19 +17,16 @@ dataset_info:
15
  - name: ground_truth
16
  dtype: string
17
  ---
18
- ## Dataset Description
19
 
20
  **KnowledgeMath** is a knowledge-intensive dataset focused on mathematical reasoning within the domain of finance. It requires the model to comprehend specialized financial terminology and to interpret tabular data presented in the questions.
21
  **KnowledgeMath** includes **1200 QA examples** across 7 key areas in finance. These examples were collected from financial experts and feature detailed solution annotations in Python format.
22
 
23
- ## Dataset Information
24
-
25
  - Paper: https://arxiv.org/abs/2311.09797
26
  - Code: https://github.com/yale-nlp/KnowledgeMath
27
  - Leaderboard: will be released soon!
28
 
29
- ### Data Downloading and Usage
30
-
31
  All the data examples were divided into two subsets: *validation* and *test*.
32
 
33
  - **validation**: 200 examples used for model development, validation, or for those with limited computing resources.
@@ -39,11 +38,7 @@ You can download this dataset by the following command:
39
  from datasets import load_dataset
40
 
41
  dataset = load_dataset("yale-nlp/KnowledgeMath")
42
- ```
43
 
44
- Here are some examples of how to access the downloaded dataset:
45
-
46
- ```python
47
  # print the first example on the validation set
48
  print(dataset["validation"][0])
49
 
@@ -51,8 +46,6 @@ print(dataset["validation"][0])
51
  print(dataset["test"][0])
52
  ```
53
 
54
- ### Data Format
55
-
56
  The dataset is provided in json format and contains the following attributes:
57
 
58
  ```json
@@ -62,11 +55,12 @@ The dataset is provided in json format and contains the following attributes:
62
  "tables": [list] List of Markdown-format tables associated with the question,
63
  "python_solution": [string] Python-format and executable solution by financial experts. The code is written in a clear and executable format, with well-named variables and a detailed explanation,
64
  "ground_truth": [integer] Executed result of `python solution`, rounded to three decimal places,
65
- "topic": [string] The related financial area of the question
 
66
  }
67
  ```
68
 
69
- ### Automated Evaluation
70
 
71
  To automatically evaluate a model on **KnowledgeMath**, please refer to our GitHub repository [here](https://github.com/yale-nlp/KnowledgeMath).
72
 
 
8
  dtype: string
9
  - name: tables
10
  dtype: sequence
11
+ feature:
12
+ dtype: string
13
  - name: topic
14
  dtype: string
15
  - name: python_solution
 
17
  - name: ground_truth
18
  dtype: string
19
  ---
20
+ ## KnowledgeMath Benchmark Description
21
 
22
  **KnowledgeMath** is a knowledge-intensive dataset focused on mathematical reasoning within the domain of finance. It requires the model to comprehend specialized financial terminology and to interpret tabular data presented in the questions.
23
  **KnowledgeMath** includes **1200 QA examples** across 7 key areas in finance. These examples were collected from financial experts and feature detailed solution annotations in Python format.
24
 
 
 
25
  - Paper: https://arxiv.org/abs/2311.09797
26
  - Code: https://github.com/yale-nlp/KnowledgeMath
27
  - Leaderboard: will be released soon!
28
 
29
+ ## KnowledgeMath Dataset Information
 
30
  All the data examples were divided into two subsets: *validation* and *test*.
31
 
32
  - **validation**: 200 examples used for model development, validation, or for those with limited computing resources.
 
38
  from datasets import load_dataset
39
 
40
  dataset = load_dataset("yale-nlp/KnowledgeMath")
 
41
 
 
 
 
42
  # print the first example on the validation set
43
  print(dataset["validation"][0])
44
 
 
46
  print(dataset["test"][0])
47
  ```
48
 
 
 
49
  The dataset is provided in json format and contains the following attributes:
50
 
51
  ```json
 
55
  "tables": [list] List of Markdown-format tables associated with the question,
56
  "python_solution": [string] Python-format and executable solution by financial experts. The code is written in a clear and executable format, with well-named variables and a detailed explanation,
57
  "ground_truth": [integer] Executed result of `python solution`, rounded to three decimal places,
58
+ "topic": [string] The related financial area of the question,
59
+ "knowledge_terms": [list] List of knowledge terms in our constructed knowledge bank that is necessary to answer the given question. We will release this feature upon paper publication
60
  }
61
  ```
62
 
63
+ ## Automated Evaluation
64
 
65
  To automatically evaluate a model on **KnowledgeMath**, please refer to our GitHub repository [here](https://github.com/yale-nlp/KnowledgeMath).
66