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 · pageptjf0226_3.png

What is the underlined word in the first line on the page?

OCR text of the page · 2,943 characters; the scanned image itself is not published
see this story from CNN for more details
3. Regulators may order Tressa off market after trial failure
The FDA is considering removing AstraZeneca's Tressa from the market after a trial showed
that it was ineffective in prolonging the lives of lung cancer patients. AstraZeneca says it
will not pull the drug on its own but has decided to stop promoting it. Iressa was one of the
first in a new category of cancer drugs that promised to interfere with the development of a
tumor without the harsh side effects associated with chemotherapy. In trials, the drug
shrank tumors in 10 percent of patients who had failed to respond to other treatments. But
when the FDA gave its approval to Iressa, it required a follow-up study to demonstrate that
the drug prolonged patients' lives. That led to the results announced late last week and
leaves Iressa with a doubtful future.
see this story from The New York Times for more details (reg. req.)
4. Pharmos shares in meltdown after dexanabinol fails Phase III
Shares of Israel's Pharmos melted away in pre-market trading after the company
announced that its lead drug candidate flunked a late-stage trial for treating severe
traumatic brain injury. Researchers said that patients taking dexanabinol, which some
analysts believed had blockbuster potential, did not demonstrate any significant
improvement. Pharmos shares lost 77 percent of their value on the news. Pharmos has
been developing compounds developed from cannabis for treating central nervous system
disorders. "We are very disappointed and frankly somewhat perplexed that we did not
observe any neuroprotective effect of dexanabinol in TBI," said Pharmos CEO Haim Aviv. "In
the next few weeks, we will carefully analyze the results of this trial before making final
decisions about our future plans, but it is unlikely that we will continue to develop
dexanabinol for TBI."
- go to this article from the Associated Press for more information
ALSO: There was some good news for Pharmos, however. The company will be receiving a
$10 million milestone payment from Bausch & Lomb for the eye treatment Zylet, Article
5. Takeda suspends Phase III trial of diabetes drug
Japan's Takeda has stopped a Phase III trial of its experimental diabetes drug TAK-559
after researchers received abnormal liver enzyme test results from patients taking the drug.
Elevated liver enzyme results have stopped the FDA from approving other diabetes drugs.
TAK-559 was being developed as a successor to Takeda's Actos, which loses patent
protection in 2010.
- see this article from the Financial Times for more details
ALSO NOTED
TODAY'S SPOTLIGHT... Cardiome touts heart drug trial data Canada's Cardiome
Pharma says that a Phase III trial of RSD1235 shows it was effective in restoring normal
heart rhythm to 52 percent of patients suffering atrial fibrillation compared to 4 percent
PLA-TAK-00020531
Source: https://www.indup50 80boosts.ucsf.edu/docs/ptjf0226
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
story
Models
4 of 4 columns · click a model to add or remove it

Ours

Wrong

this

4 characters2 tokens

Aux 2015

Wrong

The underlined word in the first line of the page is Tressa.

64 characters18 tokens

PiT-FT 2015

Wrong

Empty response.

0 characters

ChronoGPT 2015

Wrong

Tressa

6 characters3 tokens