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 · pagehjxn0226_57.png

What is the title of the second table?

OCR text of the page · 1,848 characters; the scanned image itself is not published
ser pol
Cause of revision
Rate
%% %% distribut, of failure causes
Recurrent prosthesis dislocation
95/8.888
1.1
23,9
Aseptic loosening of the stem
87/8.888
1,0
21,9
Aseptic loosening of the cup
58/8.888
0,7
14,6
Periprosthetic bone fracture
42/8.888
0,5
10,6
Septic loosening
25/8.888
0,3
6,3
Global aseptic loosening
24/8.88
0,3
6,0
Breakage of stem
12/8.888
0, 1
3.0
Primary instability
8/8.888
0 , 1
2,0
Pain without loosening
7/8.888
0,1
Poly wear
7/8.88
1,8
0, 1
1,8
Heterotopic bone
4/8.888
0,05
1,0
Breakage of cup
4/8.888
0,05
10
Breakage of head
4/8.888
0,05
1,0
Other
3/8.888
0,03
0,8
Unknown
18/8.888
0,2
4,5
Total
398/8.888
4.5
100,0
Prosthesis CER XIR
Cause of revision
Rate
%. distribut. of failure causes
Aseptic loosening of the stem
23/4.962
0,5
19,0
Recurrent prosthesis dislocation
20/4.962
0,4
16,5
Aseptic loosening of the cup
18/4.962
0,4
14,9
Periprosthetic bone fracture
11/4.962
0,2
9, 1
Primary instability
11/4.962
0,2
9,1
Septic loosening
11/4.962
0 ,2
9 , 1
Pain without loosening
3/4.962
0, 1
2,5
Breakage of prosthesis
3/4.962
0,1
2,5
Global aseptic loosening
2/4.962
0,04
1,7
Heterotopic bone
2/4.962
0,04
Poly wear
2/4.962
0,04
Other
4/4.962
0, 1
1.7
3,3
Unknown
1/4.962
0,2
9 ,1
Total
21/4962
2,4
100,0
Prosthesis met met
Cause of revision
Rate
%% distribut, of failure causes
Aseptic loosening of the cup.
71/4.637
26,0
Aseptic loosening of the stem
43/4.637
1.5
0,9
15,8
Septic loosening
28/4.637
0,6
10,3
Global aseptic loosening
27/4.637
0,6
9,9
Recurrent prosthesis dislocation
23/4.637
0,5
8,4
Breakage of prosthesis
19/4.637
0,4
7,0
Periprosthetic bone fracture
16/4.637
0,3
5,9
Pain without loosening
8/4.637
0,2
2,9
Primary instability
4/4.637
0,1
1,5
2/4.637
0,04
7/4.637
0,7
Other
0,2
2,6
Unknown
25/4.637
0,5
9.2
Total
273/4.637
5,9
100,0
57
Source: https://www.industrydocuments.ucsf.edu/docs/hjxn0226
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
Prosthesis CER-XLP
Models
4 of 4 columns · click a model to add or remove it

Ours

Partial · ANLS 0.83

Prosthesis CER XIR

18 characters7 tokens

Aux 2015

Wrong

The second table is the cause of revision table.

The OCR text has several errors:

  • "ser pol" → "ser" (likely "ser" or "ser" - likely "ser" or "ser" - but "ser" is common in some contexts)
  • "Cause of
205 characters64 tokens

PiT-FT 2015

Wrong

<|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<> <|<>

64 characters64 tokens

ChronoGPT 2015

Wrong

ChronoGPT

Causes of Revision

Rate

Aseptic loosening of the stem

23/4.962 0,5 19,0 Recurrent prosthesis dislocation 20/4.962 0,4 14,6 Periprosthetic

155 characters64 tokens