AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

The Two-Hump Problem: Bridging the Difficulty Gap in Mathematical Reinforcement Learning

Lucas Fagan, Michele Tarquini, Ali Shehper, Maksymilian Manko, Angus Gruen, Coco Huang, Giorgi Butbaia, Davide Passaro, Sergei Gukov

ICML 2026 · Reinforcement learning › Visual and representation RL · approach: In-context and adaptive RL · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use substitution supermoves with the Dual-Ring Transformer instead of elementary AC-moves with ResNetPPO agent · Miller-Schupp benchmark (1190 presentations)presentations solved: 457 -> 591Improved
add AC-19 and AC-1M training data (ablation)PPO-Sub-DRT · Miller-Schupp benchmark (1190 presentations)presentations solved: 591 -> 610Improved
use cyclic relative positional encodings instead of canonical form with absolute positional encodings (ablation)PPO with substitutions · Miller-Schupp benchmark (1190 presentations)presentations solved: 562.6 -> 588.2Improved
add AC-19 data to canonical-form PPO (ablation)PPO-Sub-Canon · Miller-Schupp benchmark (1190 presentations)presentations solved: 562.6 -> 575.5Improved
use substitution moves and canonical form in greedy searchgreedy search · Miller-Schupp benchmark (1190 presentations)nodes needed: 1M -> 614 nodesImproved
increase greedy search node budget from 1M to 10M (ablation)GS-Sub · Miller-Schupp benchmark (1190 presentations)presentations solved: 640 -> 640No clear effect
increase greedy search node budget from 10K to 1M (ablation)GS-Sub · Miller-Schupp benchmark (1190 presentations)presentations solved: 604 -> 640Improved
use PPO policy as heuristic in beam searchPPO · Miller-Schupp benchmark (1190 presentations)mean substitution path length: 13.77 -> 12.39Improved
use automorphic numbers as reward signal for greedy searchgreedy search · Miller-Schupp benchmark (1190 presentations)presentations solvedNo clear effect
train generator with automorphism game rather than random automorphisms (ablation)AC-19 data generation · 200 AC-trivial AC-19 presentations (10K node budget)difficulty of generated presentationsImproved
train on all 1190 benchmark presentations instead of only the 640 AC-trivial ones (ablation)PPO · Miller-Schupp benchmark (1190 presentations)presentations solvedHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy