Update README.md
Browse files
README.md
CHANGED
@@ -153,12 +153,36 @@ configs:
|
|
153 |
language:
|
154 |
- en
|
155 |
---
|
156 |
-
# Dataset Card for
|
157 |
|
158 |
<!-- Provide a quick summary of the dataset. -->
|
159 |
|
160 |
Introducing IsoBench, a benchmark dataset containing problems from four major areas: math, science, algorithms, and games. Each example is presented with multiple isomorphic representations of inputs, such as visual, textual, and mathematical presentations. Details of IsoBench can be found in the [paper](https://arxiv.org/abs/2404.01266) or [website](https://isobench.github.io).
|
161 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
162 |
## Dataset Details
|
163 |
|
164 |
## Uses
|
@@ -166,15 +190,11 @@ Introducing IsoBench, a benchmark dataset containing problems from four major ar
|
|
166 |
<!-- Address questions around how the dataset is intended to be used. -->
|
167 |
There are 4 major domains: math, algorithm, game, and science. Each domain has several subtasks. We will show how to load the data for each subtask.
|
168 |
|
169 |
-
###
|
170 |
-
|
171 |
-
<!-- This section describes suitable use cases for the dataset. -->
|
172 |
-
|
173 |
-
#### Mathematics
|
174 |
|
175 |
[More Information Needed]
|
176 |
|
177 |
-
|
178 |
There are three algorithmic tasks, with ascending complexity: graph connectivity, graph maximum flow, and graph isomorphism.
|
179 |
|
180 |
You can download the data by
|
@@ -190,17 +210,43 @@ Each task has 128 dev samples under the validation split.
|
|
190 |
|
191 |
|
192 |
|
193 |
-
|
194 |
|
195 |
[More Information Needed]
|
196 |
|
197 |
-
|
198 |
|
199 |
[More Information Needed]
|
200 |
|
201 |
|
202 |
## Data Fields
|
203 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
204 |
|
205 |
## Citation
|
206 |
|
@@ -223,6 +269,6 @@ Each task has 128 dev samples under the validation split.
|
|
223 |
Fu, Deqing, Ghazal Khalighinejad, Ollie Liu, Bhuwan Dhingra, Dani Yogatama, Robin Jia, and Willie Neiswanger. "IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations." arXiv preprint arXiv:2404.01266 (2024).
|
224 |
|
225 |
|
226 |
-
##
|
227 |
|
228 |
{deqingfu,zliu2898}@usc.edu, [email protected]
|
|
|
153 |
language:
|
154 |
- en
|
155 |
---
|
156 |
+
# Dataset Card for IsoBench
|
157 |
|
158 |
<!-- Provide a quick summary of the dataset. -->
|
159 |
|
160 |
Introducing IsoBench, a benchmark dataset containing problems from four major areas: math, science, algorithms, and games. Each example is presented with multiple isomorphic representations of inputs, such as visual, textual, and mathematical presentations. Details of IsoBench can be found in the [paper](https://arxiv.org/abs/2404.01266) or [website](https://isobench.github.io).
|
161 |
|
162 |
+
## Table of Contents
|
163 |
+
- [Dataset Details](#dataset-details)
|
164 |
+
- [Mathematics](#mathematics)
|
165 |
+
- [Algorithms](#algorithms)
|
166 |
+
- [Games](#games)
|
167 |
+
- [Science](#science)
|
168 |
+
- [Data Fields](#deta-fields)
|
169 |
+
- [Mathematics](#mathematics)
|
170 |
+
- [Convexity](#convexity)
|
171 |
+
- [Breakpoint](#breakpoint)
|
172 |
+
- [Parity](#parity)
|
173 |
+
- [Algorithms](#algorithms)
|
174 |
+
- [Connectivity](#connectivity)
|
175 |
+
- [Maxflow](#maxflow)
|
176 |
+
- [Isomorphism](#isomorphism)
|
177 |
+
- [Games](#games)
|
178 |
+
- [Winner Identification](#winner-identification)
|
179 |
+
- [Chess Puzzle](#chess-puzzle)
|
180 |
+
- [Science](#science)
|
181 |
+
- [Chemistry](#chemistry)
|
182 |
+
- [Physics](#physics)
|
183 |
+
- [Citation](#citation)
|
184 |
+
- [Contact](#contact)
|
185 |
+
|
186 |
## Dataset Details
|
187 |
|
188 |
## Uses
|
|
|
190 |
<!-- Address questions around how the dataset is intended to be used. -->
|
191 |
There are 4 major domains: math, algorithm, game, and science. Each domain has several subtasks. We will show how to load the data for each subtask.
|
192 |
|
193 |
+
### Mathematics
|
|
|
|
|
|
|
|
|
194 |
|
195 |
[More Information Needed]
|
196 |
|
197 |
+
### Algorithms
|
198 |
There are three algorithmic tasks, with ascending complexity: graph connectivity, graph maximum flow, and graph isomorphism.
|
199 |
|
200 |
You can download the data by
|
|
|
210 |
|
211 |
|
212 |
|
213 |
+
### Games
|
214 |
|
215 |
[More Information Needed]
|
216 |
|
217 |
+
### Science
|
218 |
|
219 |
[More Information Needed]
|
220 |
|
221 |
|
222 |
## Data Fields
|
223 |
|
224 |
+
### Mathematics
|
225 |
+
|
226 |
+
[More Information Needed]
|
227 |
+
|
228 |
+
### Algorithms
|
229 |
+
|
230 |
+
#### Connectivity
|
231 |
+
- `image`: a PIL Image feature
|
232 |
+
- `query_nodes_color`: a `string` feature
|
233 |
+
- `adjacency_matrix`: a `string` feature, a string of an 2d array representing the adjacency matrix of a graph
|
234 |
+
- `query_node_1`: a `unit32` feature
|
235 |
+
- `query_node_2`: a `unit32` feature
|
236 |
+
- `label`: a `bool` feature, with possible values including `True` (query nodes connected) and `False` (query nodes not connected)
|
237 |
+
- `id`: a `string` feature
|
238 |
+
|
239 |
+
#### Maxflow
|
240 |
+
|
241 |
+
#### Isomorphism
|
242 |
+
|
243 |
+
### Games
|
244 |
+
|
245 |
+
[More Information Needed]
|
246 |
+
|
247 |
+
### Science
|
248 |
+
|
249 |
+
[More Information Needed]
|
250 |
|
251 |
## Citation
|
252 |
|
|
|
269 |
Fu, Deqing, Ghazal Khalighinejad, Ollie Liu, Bhuwan Dhingra, Dani Yogatama, Robin Jia, and Willie Neiswanger. "IsoBench: Benchmarking Multimodal Foundation Models on Isomorphic Representations." arXiv preprint arXiv:2404.01266 (2024).
|
270 |
|
271 |
|
272 |
+
## Contact
|
273 |
|
274 |
{deqingfu,zliu2898}@usc.edu, [email protected]
|