tokenization_robustness
Collection
26 items
•
Updated
Datasets:
Dataset Viewer
question
stringlengths 23
95
| choices
listlengths 4
4
| answer
int64 0
3
| answer_label
stringclasses 4
values | split
stringclasses 2
values | subcategories
stringclasses 3
values | lang
stringclasses 1
value | second_lang
stringclasses 1
value | coding_lang
stringclasses 1
value | notes
stringlengths 0
62
| id
stringlengths 7
7
| set_id
float64 180
220
| variation_id
stringclasses 1
value |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Which branch of physics deals with motion?
|
[
"Thermodynamics",
"Quantum Mechanics",
"Optics",
"Classical Mechanics"
] | 3 |
D
|
dev
|
Unicode formatting
|
eng_Latn
|
Normal formatting
|
180-1.0
| 180 |
1.0
|
||
Which gas makes up most of Earth's atmosphere?
|
[
"Oxygen",
"Nitrogen",
"Carbon Dioxide",
"Argon"
] | 1 |
B
|
dev
|
Unicode formatting
|
eng_Latn
|
182-1.0
| 182 |
1.0
|
|||
Who wrote Romeo and Juliet?
|
[
"Charles Dickens",
"William Shakespeare",
"Mark Twain",
"Jane Austin"
] | 1 |
B
|
dev
|
Unicode formatting
|
eng_Latn
|
183-1.0
| 183 |
1.0
|
|||
What is the chemical symbol for gold?
|
[
"Ag",
"Al",
"Ar",
"Au"
] | 3 |
D
|
dev
|
Unicode formatting
|
eng_Latn
|
184-1.0
| 184 |
1.0
|
|||
Which planet is closest to the sun?
|
[
"Venus",
"Earth",
"Mercury",
"Mars"
] | 2 |
C
|
dev
|
Unicode formatting
|
eng_Latn
|
Fullwidth characters
|
186-1.0
| 186 |
1.0
|
||
What is the largest ocean?
|
[
"Atlantic",
"Indian",
"Pacific",
"Arctic"
] | 2 |
C
|
dev
|
Unicode formatting
|
eng_Latn
|
Small capitals
|
187-1.0
| 187 |
1.0
|
||
Who painted the Mona Lisa?
|
[
"Michelangelo",
"Lonardo da Vinci",
"Della Santa",
"Donato"
] | 1 |
B
|
test
|
Unicode formatting
|
eng_Latn
|
Mirrored text
|
192-1.0
| 192 |
1.0
|
||
What programming language uses indentation?
|
[
"Java",
"C++",
"Python",
"JavaScript"
] | 2 |
C
|
test
|
Unicode formatting
|
eng_Latn
|
Monospace unicode
|
193-1.0
| 193 |
1.0
|
||
What is the capital of Japan?
|
[
"Tokyo",
"Kyoto",
"Osaka",
"Hiroshima"
] | 0 |
A
|
dev
|
Unicode formatting
|
eng_Latn
|
Outlined letters
|
194-1.1
| 194 |
1.0
|
||
Which vitamin prevents scurvy?
|
[
"Vitamin D",
"Vitamin A",
"Vitamin C",
"Vitamin B"
] | 2 |
C
|
dev
|
Unicode formatting
|
eng_Latn
|
Bold script
|
195-1.1
| 195 |
1.0
|
||
What is the atomic number of carbon?
|
[
"6",
"4",
"8",
"12"
] | 0 |
A
|
dev
|
Unicode formatting
|
eng_Latn
|
Sans-serif bold
|
196-1.1
| 196 |
1.0
|
||
Which country has the largest population?
|
[
"India",
"USA",
"Brazil",
"China"
] | 3 |
D
|
dev
|
Unicode formatting
|
eng_Latn
|
Box drawing
|
197-1.1
| 197 |
1.0
|
||
What is the process of photosynthesis?
|
[
"Animals eat plants",
"Cells divide",
"Water evaporates",
"Plants make food"
] | 3 |
D
|
dev
|
Unicode formatting
|
eng_Latn
|
Overline text
|
198-1.1
| 198 |
1.0
|
||
What is the boiling point of water?
|
[
"90°C",
"110°C",
"80°C",
"100°C"
] | 3 |
D
|
dev
|
Unicode formatting
|
eng_Latn
|
Zalgo text
|
199-1.1
| 199 |
1.0
|
||
Who developed the theory of relativity?
|
[
"Isaac Newton",
"Albert Einstein",
"Galileo Galilei",
"Stephen Hawking"
] | 1 |
B
|
dev
|
Unicode formatting
|
eng_Latn
|
Fraktur text
|
200-1.1
| 200 |
1.0
|
||
Which planet has rings?
|
[
"Saturn",
"Jupiter",
"Uranus",
"Neptune"
] | 0 |
A
|
dev
|
Unicode formatting
|
eng_Latn
|
Italic unicode
|
201-1.1
| 201 |
1.0
|
||
Report card shows Math: A, English: B+, Calculus: A-. What grade was received in mathematics?
|
[
"A-",
"B+",
"B",
"A"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
Box drawing
|
203-1.0
| 203 |
1.0
|
||
Follow right and then left to find the right answer. Which way to go?
|
[
"left only",
"right only",
"up",
"right then left"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
→ Follow this direction to find the answer ←. Which way to go?
|
204-1.1
| 204 |
1.0
|
||
There is a way between A and B, A and C and C and D. What connects B and C?
|
[
"A",
"B",
"C",
"D"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
Text diagram
|
205-1.1
| 205 |
1.0
|
||
Root has Branch1 and Branch2. How many branches are shown?
|
[
"1",
"3",
"2",
"4"
] | 2 |
C
|
dev
|
Unusual formatting
|
eng_Latn
|
Tree structure
|
206-1.1
| 206 |
1.0
|
||
Start leads to Process leads to End. What comes after Start?
|
[
"End",
"Beginning",
"Middle",
"Process"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
Flow chart
|
207-1.1
| 207 |
1.0
|
||
H atoms surround C atom in tetrahedral arrangement. What is the chemical formula?
|
[
"water",
"carbon dioxide",
"oxygen",
"methane"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
Chemical structure
|
208-1.1
| 208 |
1.0
|
||
Three-sided polygon with straight edges. What shape is this?
|
[
"triangle",
"square",
"circle",
"rectangle"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
ASCII pyramid
|
209-1.1
| 209 |
1.0
|
||
Grid arrangement: first row has 1 and 2, second row has 3 and 4. What is the top-right element?
|
[
"2",
"1",
"3",
"4"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
Mathematical matrix
|
210-1.1
| 210 |
1.0
|
||
DNA codon ATG codes for amino acid Met which starts protein. What does ATG code for?
|
[
"methionine",
"both start and methionine",
"stop codon",
"start codon"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
Timeline format
|
211-1.1
| 211 |
1.0
|
||
Nested brackets: curly contains square contains round. Which bracket is innermost?
|
[
"square brackets",
"curly braces",
"angle brackets",
"parentheses"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
Bracket notation
|
212-1.1
| 212 |
1.0
|
||
Circuit shows a component labeled R between two connection points. What type of component is R?
|
[
"capacitor",
"inductor",
"resistor",
"battery"
] | 2 |
C
|
dev
|
Unusual formatting
|
eng_Latn
|
ASCII circuit
|
213-1.1
| 213 |
1.0
|
||
Outline: 1 Main topic, 1.1 Subtopic, 1.1.1 Detail. What level is "Detail"?
|
[
"first level",
"second level",
"third level",
"fourth level"
] | 2 |
C
|
dev
|
Unusual formatting
|
eng_Latn
|
Hierarchical list
|
214-1.1
| 214 |
1.0
|
||
DNA sequence ATGCTA. Which base pairs with A?
|
[
"G",
"C",
"T",
"U"
] | 2 |
C
|
dev
|
Unusual formatting
|
eng_Latn
|
DNA sequence
|
215-1.1
| 215 |
1.0
|
||
Bar chart shows A medium bar, B short bar, C long bar. Which has the highest value?
|
[
"C",
"A",
"B",
"equal"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
ASCII bar chart
|
216-1.1
| 216 |
1.0
|
||
What type of bonds are shown in the carbon dioxide molecule?
|
[
"double bonds",
"single bonds",
"triple bonds",
"ionic bonds"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
Molecular structure
|
217-1.1
| 217 |
1.0
|
||
States S1 to S2 to S3 with loop back to S1. How many states are shown?
|
[
"2",
"4",
"1",
"3"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
State diagram
|
218-1.1
| 218 |
1.0
|
||
Clock positions 12 at top, 3 at right, 6 at bottom, 9 at left. What time format is this?
|
[
"digital time",
"24-hour",
"military",
"analog clock"
] | 3 |
D
|
dev
|
Unusual formatting
|
eng_Latn
|
ASCII clock
|
219-1.1
| 219 |
1.0
|
||
DNA sequence ATG translates to amino acid Met. What is ATG called in genetics?
|
[
"start codon",
"methionine",
"both start and methionine",
"stop codon"
] | 0 |
A
|
dev
|
Unusual formatting
|
eng_Latn
|
Genetic code
|
220-1.1
| 220 |
1.0
|
||
What is Einstein's famous equation?
|
[
"f = m a",
"p v = n/t",
"E = m c^2",
"a^2 + b^2 = c^2"
] | 2 |
C
|
dev
|
Unicode formatting, Chemical formulas
|
eng_Latn
|
Superscript letters
|
188-1.0
| 188 |
1.0
|
||
What is the chemical formula for water?
|
[
"H_2 SO_4",
"NaCl",
"CO_2",
"H_2 O"
] | 3 |
D
|
dev
|
Unicode formatting, Chemical formulas
|
eng_Latn
|
Subscript letters
|
189-1.0
| 189 |
1.0
|
Dataset Card for Tokenization Robustness
A comprehensive evaluation dataset for testing robustness of different tokenization strategies.
Dataset Details
Dataset Description
This dataset evaluates how robust language models are to different tokenization strategies and edge cases. It includes questions with multiple choice answers designed to test various aspects of tokenization handling.
- Curated by: R3
- Funded by [optional]: [More Information Needed]
- Shared by [optional]: [More Information Needed]
- Language(s) (NLP): [More Information Needed]
- License: cc
Dataset Sources [optional]
- Repository: [More Information Needed]
- Paper [optional]: [More Information Needed]
- Demo [optional]: [More Information Needed]
Uses
Direct Use
[More Information Needed]
Out-of-Scope Use
[More Information Needed]
Dataset Structure
The dataset contains multiple-choice questions with associated metadata about tokenization types and categories.
Dataset Creation
Curation Rationale
[More Information Needed]
Source Data
Data Collection and Processing
[More Information Needed]
Who are the source data producers?
[More Information Needed]
Annotations [optional]
Annotation process
[More Information Needed]
Who are the annotators?
[More Information Needed]
Personal and Sensitive Information
[More Information Needed]
Bias, Risks, and Limitations
The dataset focuses primarily on English text and may not generalize to other languages or tokenization schemes not covered in the evaluation.
Recommendations
Users should be made aware of the risks, biases and limitations of the dataset. More information needed for further recommendations.
Citation [optional]
BibTeX:
[More Information Needed]
APA:
[More Information Needed]
Glossary [optional]
[More Information Needed]
More Information [optional]
[More Information Needed]
Dataset Card Authors [optional]
[More Information Needed]
Dataset Card Contact
[More Information Needed]
- Downloads last month
- 17
Size of downloaded dataset files:
25.6 kB
Size of the auto-converted Parquet files:
25.6 kB
Number of rows:
78