Eval receipt
Example
A test of Vera, run before she worked on any lead.
Scored 3 of 3
Run 3 of 3- Saved before run 1
- 3 test leads, a pass mark of 3 of 3, and a $5.00 cap
- Model
- Claude
- Test set
- 3 sample leads with known answers
- Grader
- A second model with an answer key
- Cost
- $0.38 for 3 runs
Why did it change?
Run 3 scored 3 of 3 on October 6, after runs that scored 1 and 2.
Who approved it?
The founder approved the test plan and the $5.00 cap before run 1.
What did it cost?
$0.38 for all 3 runs, under the $5.00 cap.
How do we know?
The grader's results agree with all 3 runs, and the plan was saved before run 1.
Entries
5 entries on 1 day-
October 6
On October 6, Vera ran the test 3 times. Her score went from 1 of 3 to 3 of 3.
8:40 PM UTCThe founder saved the test plan before the first run.
Verified8:52 PM UTCVera ran the test and scored 1 of 3.
She missed 2 company names.
$0.12 Verified9:20 PM UTCVera ran the test again and scored 2 of 3.
She missed 1 company name.
$0.13 Verified9:51 PM UTCVera ran the test a third time and scored 3 of 3.
Open the code receipt that built Vera$0.13 Verified10:02 PM UTCThe grader checked all 3 runs against the answer key.
Open the work receipt for Vera's first live runWitnessed
Proof
Can't be changed without it showing.
Waiting for its Bitcoin stamp.