Spatial-Gym benchmark shows the best tested model solves only 16% of pathfinding tasks versus 98% for humans, with step-by-step and backtracking formats producing mixed effects across model strengths.
John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
A harmonized RSSC-oriented taxonomy standardizes research software definitions, maps prior studies, and demonstrates that security signals from OpenSSF Scorecard vary meaningfully across taxonomy clusters on the RSE corpus.
citing papers explorer
-
Mind the Gap Between Spatial Reasoning and Acting! Step-by-Step Evaluation of Agents With Spatial-Gym
Spatial-Gym benchmark shows the best tested model solves only 16% of pathfinding tasks versus 98% for humans, with step-by-step and backtracking formats producing mixed effects across model strengths.
-
Operationalizing Research Software for Supply Chain Security
A harmonized RSSC-oriented taxonomy standardizes research software definitions, maps prior studies, and demonstrates that security signals from OpenSSF Scorecard vary meaningfully across taxonomy clusters on the RSE corpus.