Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Addition

Benchmarks

Task NameDataset NameSOTA ResultTrend
Addition10-digit addition 3 terms
Token Accuracy89.56
14
AdditionAddition
Accuracy99.5
7
AdditionAddition synthetic (val)
Skill 1 Score0.016
7
Token-level AdditionAddition T3 (val)
Token Accuracy89.56
6
Binary AdditionAddition (OOD lengths 20-60)
OOD Performance45
4
Arithmetic AdditionADDITIONBase-5 (test)
Accuracy23.5
4
Arithmetic AdditionADDITIONBase-3 (test)
Accuracy36.3
4
Arithmetic AdditionADDITIONBase-2 (test)
Accuracy95.4
4
AdditionAddition 200 prompts Qwen3-8B (test)
Accuracy100
3
Showing 9 of 9 rows