-
Notifications
You must be signed in to change notification settings - Fork 2
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge branch 'neurips_eval' into msaroufim/corr2cause
- Loading branch information
Showing
1 changed file
with
88 additions
and
4 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -1,6 +1,90 @@ | ||
| entries: [ | ||
| {description: "corr2cause:model=neurips/local", priority: 1} | ||
|
|
||
| {description: "sam_sum:model=neurips/local,max_train_instances=3", priority: 1} | ||
| {description: "ethics_utilitarianism:model=neurips/local", priority: 1} | ||
| entries: [ | ||
| ## Real | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=algebra,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=prealgebra,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=geometry,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=1,use_official_examples=True", priority: 2} | ||
| {description: "math:model=text_code,subject=precalculus,level=1,use_official_examples=True", priority: 2} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=algebra,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=prealgebra,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=geometry,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=2,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=precalculus,level=2,use_official_examples=True", priority: 4} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=algebra,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=prealgebra,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=geometry,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=3,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=precalculus,level=3,use_official_examples=True", priority: 3} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=algebra,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=prealgebra,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=geometry,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=4,use_official_examples=True", priority: 4} | ||
| {description: "math:model=text_code,subject=precalculus,level=4,use_official_examples=True", priority: 4} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=algebra,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=prealgebra,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=geometry,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=5,use_official_examples=True", priority: 3} | ||
| {description: "math:model=text_code,subject=precalculus,level=5,use_official_examples=True", priority: 3} | ||
|
|
||
| # With chain-of-thought prompting: | ||
| {description: "math:model=text_code,subject=number_theory,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=algebra,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=prealgebra,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=geometry,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=1,use_chain_of_thought=True", priority: 2} | ||
| {description: "math:model=text_code,subject=precalculus,level=1,use_chain_of_thought=True", priority: 2} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=algebra,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=prealgebra,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=geometry,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=2,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=precalculus,level=2,use_chain_of_thought=True", priority: 4} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=algebra,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=prealgebra,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=geometry,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=3,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=precalculus,level=3,use_chain_of_thought=True", priority: 3} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=algebra,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=prealgebra,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=geometry,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=4,use_chain_of_thought=True", priority: 4} | ||
| {description: "math:model=text_code,subject=precalculus,level=4,use_chain_of_thought=True", priority: 4} | ||
|
|
||
| {description: "math:model=text_code,subject=number_theory,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=intermediate_algebra,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=algebra,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=prealgebra,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=geometry,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=counting_and_probability,level=5,use_chain_of_thought=True", priority: 3} | ||
| {description: "math:model=text_code,subject=precalculus,level=5,use_chain_of_thought=True", priority: 3} | ||
|
|
||
| {description: "sam_sum:model=neurips/local,max_train_instances=3", priority: 1} | ||
| {description: "ethics_utilitarianism:model=neurips/local", priority: 1} | ||
| {description: "corr2cause:model=neurips/local", priority: 1} | ||
|
|
||
| ] |