Output Explorer

Every prompt in the paper, and what each model wrote back.

Answer a question about a scanned page from its OCR text. Scored by exact match and ANLS against the accepted answers.

13 of 5,330 prompts

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagehqvd0227_19.png

what is the # servings/ day of Kool aid ?

OCR text of the page · 1,063 characters; the scanned image itself is not published
ADMISSION PRENATAL STATUS - TEENAGERS,
U. T. M. B. '64-'66
PATTERN OF FOOD INTAKE
Range
Major
To Not Eating
Food . Groups
# Servings / Day
ADMISSION PRENATAL STATUS - TEENAGERS,
0-30%
1. Bread / cereals etc.
4. 8
0
- white bread
2, 1
U.T. M. B. '64-'66
10-34
2. Milk products
2.4
PATTERN OF NONESSENTIAL FOOD INTAKE
10
- whole milk
. . 1, 3
5-42
3. Meat / legumes
2 .5
Miscellaneous
. 5
- chicken
0. 2
5
- hamburger
0 .2
Not Eating
Food Items
# Servings / Day
18 .
12
- lunch meat
0.5
30
1. Potato chips.
0.2
- pinto beans
0 . 3
. 3
12
4. Eggs
1.0
37.
. 2. Saltines
0.2
38
5. Citrus
0.8
1 1
3. Soda pop
0 .7
20-40
6. Fruits / vegetables
2.4
15-20
7. Fats / oils
0 . 3
34
4. Kool aid
0 .5
2
8. Water
4.8
23
5. Candy
0:4
5
6. Sugar
1 .3
lable 3
Admission Prenatal Status - Teenagers
UTM8 1964-66
Percent Adequacy Nutrient Intake
Thiamine
Riboflavin
120
110
Calories
Protein
- Calcium
iron
Vitamin-A
Vitamin-C
Niacin
100
SO
80
TO
60
Ist group
2nd group
Inicke
2155 Col. 73GM. 780MG. 12.7MG. 5070 1.U. 93M.G.
12MG. 2.0 MG. 15MG.
dustrydocuments.ucsf.edu/docs/hqvd0227
System prompt · identical for every setup
Answer the question using only the OCR text from a single document page. Return only the answer, with no explanation. Preserve the answer wording from the OCR text when possible.
Expected answer
0.5
Models
4 of 4 columns · click a model to add or remove it

Ours

Exact match

0.5

3 characters4 tokens

Aux 2015

Wrong

The question asks for the number of servings of Kool aid per day for Admission Prenatal Status - Teenagers.

Looking at the provided text, we see:

  • "UTM8 1964-66" is likely a typo for "UTM8 1964
200 characters64 tokens

PiT-FT 2015

Wrong

<|assistant|<|assistant|<|assistant|<|assistant|<|assistant|<|assistant|<assistant|<assistant|<assistant|<assistant|<assistant|<assistant|<assistant|<assistant|<ass

164 characters64 tokens

ChronoGPT 2015

Wrong
# Servings / Day
# Servings / Day
# Servings / Day

The same 17-character fragment repeats 10 times until the token limit. Showing the first three.

182 characters64 tokens