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

TaskBench

Benchmarks

Task NameDataset NameSOTA ResultTrend
Function SelectionTaskBench HuggingFace
Function Selection Accuracy77.1
45
Function SelectionTaskBench Multimedia
Function Selection Acc82.3
36
Function SelectionTaskBench DailyLife
Function Selection Accuracy96.8
36
Task PlanningTaskBench Daily Life
Node-F197.36
25
Task PlanningTaskBench Multimedia
Node F188.54
25
Tool Retrieval and Function SelectionTaskbench-HF
MRR0.75
18
Task PlanningTaskBench Multimedia v1 (test)
n-F188.54
14
Tool selectionTaskBench-MM
F1 Score20.7
12
Tool selectionTaskBench
F1 Score21.1
12
Tool Retrieval and Function SelectionTaskbench DL
Function Selection Accuracy58.9
9
Tool Retrieval and Function SelectionTaskbench-MM
Function Selection Accuracy27
9
End-to-end tool-graph planningTaskBench-23 2024 (N=501)
ToolF173.4
8
Multi-agent Planning and CompositionTaskBench Multimedia APIs (test)
Precision91.12
1
Multi-agent Planning and CompositionTaskBench Hugging Face APIs (test)
Precision74.97
1
Multi-agent Planning and CompositionTaskBench Dailylife APIs (test)
Precision95.01
1
Showing 15 of 15 rows