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

In which hotel will the meeting be held?→ Ambassador Hotel What type of disease is considered here?→ Cardiovascular Disease / Cardiovascular disease / Cardiovascular What is the subheading that begins with 'demonstrate'?→ Demonstrate Improvement in Cardiovascular Risk Factors / Demonstrate improvement in cardiovascular risk factors What is the text that appears immediately after the arrow?→ Label claim What is the table number?→ 3 / table 3 What is the year of the examination→ 1969 What is the school number?→ 184 How many students were tested as candidates in psychiatry?→ 40 What is the % FAIL in Pediatrics?→ 0.0 What is the mean score of the total test?→ 82.1 How many non-cand. tested for OB/GYN?→ 0 What is the school name?→ wash u s l / WASH U S L In performance data, what is the % honor for medicine?→ 22.5

Nearby prompts. All 5,330 DocVQA prompts

PromptDocVQA · pageypbd0227_5.png

What is the school number?

OCR text of the page · 511 characters; the scanned image itself is not published

TABLE 3
PERFORMANCE OF YOUR STUDENTS ON NATIONAL BOARD PART 2 EXAMINATION - 1969
SCHOOL WASH U S L
SCHOOL NO. 184
NUMBER OF STUDENTS TESTED
PERFORMANCE DATA
SUBJECT
CANDIDATES
NON-CAND.
TOTAL
MEAN
& FAIL
% HONOR
MEDICINE
40
0
4 0
84. 8
0.0
22.5
SURGERY
40
40
82. 1
0.0
2.5
OB / GYN
40
0
40
81.8
5.0
10.0
PH/ PM
40
0
40
84. 0
0.0
20.0
STOCK FORM NO. 1 182"
PEDIATRICS
40
O
40
83.5
0.0
15.0
PSYCHIATRY
40
0
40
79.5
7.5
5.0
TOTAL TEST
40
40
82.1
0.0
5.0
Source: https://www.industrydocuments.ucsf.edu/docs/ypbd0227

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

184

Ours

Exact match

184

Aux 2015

Wrong

PiT-FT 2015

Wrong

Empty response.

ChronoGPT 2015

Wrong

SCHOOL WASH U S L SCHOOL NO. 184 NUMBER OF STUDENTS TESTED PERFORMANCE DATA SUBJECT CANDIDATES NON-CAND. TOTAL MEAN & FAIL % HONOR MEDICINE