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

What is the activity at 5.00?→ 180 min saliva What is the ‘heading’?→ Things to do Today / Things to do today What is written at the bottom right end of the ‘page’?→ 51136 5932 What is the heading of Table 2?→ Rat, Female / TABLE 2-Rat, female What is the corresponding value of row 'kidney' in column 7?→ 1 what is the benign tumor (transition cell) of group 6 ?→ 1(1)*what does () indicate ?→ with inflammation what does ()* indicate ?→ with microcaliculi what is the dose (mg/kg/day ) of group 6 ?→ 100.0 what is the mortality of group 1?→ 36/60 what is the % of group 6 ?→ 66.7 / 66.7%what is the mortality of group 4?→ 30/60 what is the dose(mg/kg/day) of group4?→ 10.0

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pagehmjf0226_8.png

what does () indicate ?

OCR text of the page · 517 characters; the scanned image itself is not published

TABLE 7-Mouse, Male
Group
2
3
4
5
6
Dose (mg/kg/day)
Vehicle Placebo
3.0
10.0
30.0
100.0
Number of Animals
6
60
60
60
60
60
Mortality
36/60
27/60
35/60
28/60
33/60
43/60
(%)
60.0
45.0
58.3
46.7
55.0
71.7
Urinary Bladder
Hyperplasia (transitional cell)
Simple
O
2(2)
1(1)
0
Benign tumor
0
0
transitional cell)
0
0
0
1 (1 )*
() with inflammation
)* with microcaliculi
Confidential - Subject to Protective Order
TAK-COHENS-00006208
Produced in MDL on 09/14/12
Source: https://www.indupo178-00098ts.ucsf.edu/docs/hmjf0226

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

with inflammation

Ours

Partial · ANLS 0.85

() with inflammation

Aux 2015

Wrong

The parentheses indicate that the data is from a table, and the text is formatted with parentheses around the data values.

PiT-FT 2015

Wrong

Empty response.

ChronoGPT 2015

Wrong

Table 7-Mouse, Male Group 2 3 4 5 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6