Dev claim · claim-698 · #153 of 154
“A recent study led by Lawrence Livermore National Laboratory climate scientist Ben Santer found that while the models ran hot, the ‘overestimation’ was ‘partly due to systematic deficiencies in some of the post-2000 external forcings used in the model simulations.’”
Claim tags (sorted stems; highlighted when a passage shares them)
- ben
- climat
- defici
- due
- extern
- forc
- found
- hot
- laboratori
- lawrenc
- led
- livermor
- model
- model
- nation
- overestim
- partli
- post
- ran
- recent
- santer
- scientist
- simul
- studi
- systemat
- use
The model’s verdict
The Transformer retrained from the notebook (the 2024 weights were never saved), shown three ways. The first is how the notebook evaluated it.
Notebook protocol
Dev claims predicted in batches of 16, in file order, from the 2024 retrieved evidence, as the notebook does.
- Supports
- 11.9%
- Refutes
- 3.0%
- Not enough info
- 80.8%
- Disputed
- 4.4%
One claim at a time
The same model run on a single claim, which is how the Try-it page runs it.
- Supports
- 27.1%
- Refutes
- 6.0%
- Not enough info
- 52.8%
- Disputed
- 14.0%
With gold evidence
Fed the human-annotated evidence instead of retrieved passages. This is the 'val accuracy' the training loop reports.
- Supports
- 46.8%
- Refutes
- 13.0%
- Not enough info
- 24.0%
- Disputed
- 16.2%
Retrieved vs gold evidence
Gold passages were picked by the dataset’s annotators. Retrieved passages are what the TF-IDF rule returned from all 1.19M passages. Switch between the submitted 2024 output and the re-runs.
2024 submission
The passages the team actually retrieved and submitted in 2024 (their saved output file). The file records only passage ids, so scores are recomputed with the submission rule.
- Path
- not recorded
- Found
- 0 of 1 gold
- P
- 0.00
- R
- 0.00
- F
- 0.00
#1evidence-350851 He ran again for the same post in 1938 and won.
- post (shared with the claim)
- ran (shared with the claim)
- cos
- 0.393
- overlap
- 1.000
- score
- 1.393
- shared
- 2
#2evidence-315092 United Nations Security Council Resolution 420, adopted on May 31, 1978, considered a report by the Secretary-General regarding the United Nations Disengagement Observer Force.
- nation (shared with the claim)
- cos
- 0.224
- overlap
- 1.000
- score
- 1.224
- shared
- 1
#3evidence-263594 National Laboratories
- nation (shared with the claim)
- laboratori (shared with the claim)
- cos
- 0.224
- overlap
- 1.000
- score
- 1.224
- shared
- 2
#4evidence-45901 There are 45 properties and districts listed on the National Register in the county, including one National Historic Landmark
- nation (shared with the claim)
- cos
- 0.224
- overlap
- 1.000
- score
- 1.224
- shared
- 1
#5evidence-821744 It was constructed in the National Park Service Rustic style of architecture, and was listed on the National Register of Historic Places in 1988.
- nation (shared with the claim)
- cos
- 0.224
- overlap
- 1.000
- score
- 1.224
- shared
- 1
#6evidence-646741 There are 9 properties and districts listed on the National Register in the county, including one National Historic Landmark.
- nation (shared with the claim)
- cos
- 0.224
- overlap
- 1.000
- score
- 1.224
- shared
- 1
Gold evidence (1)
evidence-660755 A 2007 study by David Douglass and coworkers, concluded that the 22 most commonly used global climate models used by the IPCC were unable to accurately predict accelerated warming in the troposphere although they did match actual surface warming, concluding "projections of future climate based on these models should be viewed with much caution".
- warm
- model (shared with the claim)
- troposph
- caution
- climat (shared with the claim)
- conclud
- use (shared with the claim)
Optional · uses your own API key
Second opinion from an LLM
The model reads this claim and the evidence passages shown on this page, picks one of the four verdicts and cites the passages it used. It is told to judge from the passages only. You review the answer; the call and your decision go to the AI audit log in this browser.
Without a key nothing is sent anywhere. The rest of the site does not need one.