Datasets:

Formats:
json
Languages:
English
Size:
< 1K
ArXiv:
Libraries:
Datasets
pandas
License:
nielsr HF Staff commited on
Commit
1416202
·
verified ·
1 Parent(s): 185d9ef

Add project page link and clarify description

Browse files

This PR adds a link to the project page and clarifies the dataset description.

Files changed (1) hide show
  1. README.md +24 -12
README.md CHANGED
@@ -1,12 +1,15 @@
1
  ---
2
- license: apache-2.0
 
3
  language:
4
  - en
 
5
  size_categories:
6
  - n<1K
7
  task_categories:
8
  - question-answering
9
  - multiple-choice
 
10
  configs:
11
  - config_name: benchmark
12
  data_files:
@@ -14,15 +17,13 @@ configs:
14
  path: dataset.json
15
  tags:
16
  - geospatial
17
- annotations_creators:
18
- - expert-generated
19
- paperswithcode_id: mapeval-textual
20
  ---
21
 
22
-
23
  # MapEval-Textual
24
 
25
- [MapEval](https://arxiv.org/abs/2501.00316)-Textual is created using [MapQaTor](https://arxiv.org/abs/2412.21015).
 
 
26
 
27
  ## Usage
28
 
@@ -37,18 +38,29 @@ for item in ds["test"]:
37
  # Start with a clear task description
38
  prompt = (
39
  "You are a highly intelligent assistant. "
40
- "Based on the given context, answer the multiple-choice question by selecting the correct option.\n\n"
41
- "Context:\n" + item["context"] + "\n\n"
42
- "Question:\n" + item["question"] + "\n\n"
43
- "Options:\n"
 
 
 
 
 
 
 
 
 
44
  )
45
 
46
  # List the options more clearly
47
  for i, option in enumerate(item["options"], start=1):
48
- prompt += f"{i}. {option}\n"
 
49
 
50
  # Add a concluding sentence to encourage selection of the answer
51
- prompt += "\nSelect the best option by choosing its number."
 
52
 
53
  # Use the prompt as needed
54
  print(prompt) # Replace with your processing logic
 
1
  ---
2
+ annotations_creators:
3
+ - expert-generated
4
  language:
5
  - en
6
+ license: apache-2.0
7
  size_categories:
8
  - n<1K
9
  task_categories:
10
  - question-answering
11
  - multiple-choice
12
+ paperswithcode_id: mapeval-textual
13
  configs:
14
  - config_name: benchmark
15
  data_files:
 
17
  path: dataset.json
18
  tags:
19
  - geospatial
 
 
 
20
  ---
21
 
 
22
  # MapEval-Textual
23
 
24
+ MapEval-Textual is a benchmark dataset for evaluating the geospatial reasoning capabilities of foundation models. It consists of 700 multiple-choice questions related to spatial relationships, navigation, travel planning, and map interactions across various cities and countries. The dataset is designed to test long-context reasoning abilities.
25
+
26
+ [MapEval](https://huggingface.co/papers/2501.00316) | [MapQaTor](https://arxiv.org/abs/2412.21015) | [Project Website](https://mapeval.github.io/)
27
 
28
  ## Usage
29
 
 
38
  # Start with a clear task description
39
  prompt = (
40
  "You are a highly intelligent assistant. "
41
+ "Based on the given context, answer the multiple-choice question by selecting the correct option.
42
+
43
+ "
44
+ "Context:
45
+ " + item["context"] + "
46
+
47
+ "
48
+ "Question:
49
+ " + item["question"] + "
50
+
51
+ "
52
+ "Options:
53
+ "
54
  )
55
 
56
  # List the options more clearly
57
  for i, option in enumerate(item["options"], start=1):
58
+ prompt += f"{i}. {option}
59
+ "
60
 
61
  # Add a concluding sentence to encourage selection of the answer
62
+ prompt += "
63
+ Select the best option by choosing its number."
64
 
65
  # Use the prompt as needed
66
  print(prompt) # Replace with your processing logic