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

SCAN

Benchmarks

Task NameDataset NameSOTA ResultTrend
Reward ModelingSCAN HPD
Accuracy82.88
22
Instruction FollowingSCAN jump
Accuracy100
18
Semantic ParsingSCAN Around Right
Exact-match Accuracy100
16
Analogy generationSCAN (out-of-domain)
Accuracy15.3
15
Systematic GeneralizationSCAN Around Right (test)
Accuracy95.7
15
Systematic GeneralizationSCAN Around Right (val)
Accuracy99.8
15
Systematic GeneralizationSCAN Add Jump (test)
Accuracy99.8
15
Systematic GeneralizationSCAN Add Jump (val)
Accuracy99.6
15
Language-driven NavigationSCAN Simple v1.0
Accuracy1
12
Semantic ParsingSCAN MCD3
Exact Match Accuracy80.2
12
Semantic ParsingSCAN (MCD2)
Exact Match Accuracy80.8
12
Semantic ParsingSCAN (MCD1)
Exact-match Accuracy0.674
12
Semantic ParsingSCAN Jump
Exact-match Accuracy100
11
Command-to-action mappingSCAN (length)
Accuracy99.7
11
Language-driven NavigationSCAN around right v1.0
Accuracy1
8
Instruction FollowingSCAN around right
Accuracy99.51
7
Semantic ParsingSCAN (MCD)
Accuracy100
6
Semantic ParsingSCAN Template
Accuracy100
6
Semantic ParsingSCAN (Length)
Accuracy100
6
Semantic ParsingSCAN 0-shot lexical
Accuracy (0-shot)99
6
Semantic ParsingSCAN 1-shot lexical
Accuracy100
6
Semantic ParsingSCAN (IID)
Accuracy100
6
Around RightSCAN (val)
Accuracy97.7
6
Add JumpSCAN (val)
Accuracy96.9
6
Around RightSCAN (test)
Accuracy77.9
6
Showing 25 of 34 rows