ProcessBench

3,400 human-annotated math solutions, mostly contest-level, where the model must name the earliest wrong step or report that the trace is clean.

unassessed

This page is a discovery lead. Nobody has yet assessed it against the catalogue contract, so it carries no disposition. Absence of evidence here is not evidence of staleness.
Categorymath
Subcategorystep-level error localisation in mathematical reasoning traces
Page statusactive
MetricF1 of error_acc and correct_acc
Directionhigher_is_better
Unit%
Dataset size3400
Dataset licenceApache-2.0
PublisherQwen Team, Alibaba Inc.

What it measures

ProcessBench tests whether a model can find mistakes in someone else's math, not whether it can solve the problem from scratch. Each item is a contest- or school-level problem plus a step-by-step solution whose first error (or lack of error) was labelled by human experts. The model must return the index of that earliest bad step, or -1 if every step is correct. The suite is aimed at process reward models and at general models prompted as critics.

Task format

English generation. OpenCompass wraps each step in <paragraph_i> tags and asks for a paragraph-by-paragraph critique with the final index in \\boxed{}. Official Qwen code uses the same earliest-error / -1 protocol. Subsets: gsm8k, math, olympiadbench, omnimath.

Models reporting this benchmark

No model card in ModelSpec reports this benchmark yet.

Data

This page as JSON · Edit on GitHub