PROCESSBENCH
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
PROCESSBENCH
88GSM8K Accuracy
44
ProcessBench (test)
96.9GSM8K Accuracy
35
ProcessBench
65.7GSM8K Accuracy
28
PROCESSBENCH
82.9GSM8K Accuracy
28
ProcessBench
69.85Accuracy
20
PROCESSBENCH
87.3GSM8K Accuracy
20
ProcessBench Without Standard Answers
71.9Precise Accuracy
18
ProcessBench With Standard Answers
78.9Precise Accuracy
18
ProcessBench 1.0 (test)
87.3GSM8K Score
14
ProcessBench Overall
72.3F1 Score
13
ProcessBench Omni-MATH
63.1TNR
13
ProcessBench OlympiadBench
61.3TNR
13
ProcessBench Math
69.5TNR
13
ProcessBench GSM8K v1 (val)
68.6True Negative Rate
13
ProcessBench Aggregate (test)
56.5Avg F1
13
ProcessBench GSM8K MATH Olympiad Bench Omni Math
0.242GSM8K Error Rate
12
ProcessBench
83.2F1 Score
10
ProcessBench (OlympiaBench) 1.0 (test)
79.8Accuracy
10
ProcessBench MATH 1.0 (test)
88.4Accuracy
10
ProcessBench GSM8K 1.0 (test)
96Accuracy
10
ProcessBench (test)
0.24Worst-case Size Distortion (QwenPRM)
9
ProcessBench (test)
82.1MATH Accuracy
8
ProcessBench Average
36.8Mean F1
7
ProcessBench Omni
25.6F1 Score
7
ProcessBench Olympiad
28.7F1 Score
7