List your company

Process-level verification on ProcessBench Aggregate (test)

56.5Avg F1Qwen2.5-Math-7B-PRM800K
25.40433.47741.5549.623May 20, 2025

Evaluation results

MethodLinks
Qwen2.5-Math-7B-PRM800Ktraining_data_source=high-quality manually annotated process-level data (PRM800K)2025.0556.5
SCOPEAST Normalization=true, Code Translation=true, training_data_source=automatically constructed datasets2025.0554.4
SCOPE (w/o AST Normalization)AST Normalization=false, Code Translation=true2025.0553.6
SCOPE (w/o Code Translation)AST Normalization=true, Code Translation=false2025.0550.6
SCOPE (Step Replacement)Step Compression Strategy=Step Replacement2025.0544.6
SCOPE (Step Skipping)Step Compression Strategy=Step Skipping2025.0544.6
Skywork-PRM-7Btraining_data_source=automatically constructed datasets2025.0542.1
Skywork-PRM-1.5Btraining_data_source=automatically constructed datasets2025.0536.4
Math-Shepherd-PRM-7Btraining_data_source=automatically constructed datasets2025.0531.5
EurusPRM-Stage2training_data_source=automatically constructed datasets2025.0531.3
EurusPRM-Stage1training_data_source=automatically constructed datasets2025.0531.2
RLHFlow-PRM-Mistral-8Btraining_data_source=automatically constructed datasets2025.0528.4
RLHFlow-PRM-Deepseek-8Btraining_data_source=automatically constructed datasets2025.0526.6