AI Paper Atlas

Research map / Reinforcement learning / Imitation learning

Boolean Satisfiability via Imitation Learning

Zewei Zhang, Huan Liu, YUANHAO YU, Jun Chen, Xiangyu Xu

ICLR 2026 · Reinforcement learning › Imitation learning · approach: Imitation combined with RL

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
add variable permutation augmentation (ablation)ImitSAT · Random 3-SAT 5-15MRPP / W1%: MRPP 1.00 -> 0.75; W1% 0.28 -> 0.64Improved
add staged curriculum (5-15 then 16-30) vs 16-30 only at matched 130k steps (ablation)ImitSAT · Random 3-SAT 5-15MRPP: 1.00 -> 0.75Improved
curriculum vs training from scratch at fully matched 430k steps (ablation)ImitSAT · Random 3-SAT 16-30MRPP: 0.80 (w/o stage*) vs 0.83 (w/ stage)No clear effect
replace CNF tokens with GNN embeddings (ablation)ImitSAT · Random 3-SAT 5-15MRPP: 0.75 -> 1.00Hurt
query model at first decision instead of after 3 VSIDS decisions (single call) (ablation)ImitSAT · Random 3-SAT 5-15MRPP / W1%: 1.00 -> 0.91; 0.27 -> 0.55Improved
spend 3 calls at first decision vs 1 per 3 VSIDS decisions (ablation)ImitSAT · Random 3-SAT 100 varsMRPP: 0.95 -> 0.76Improved
increase query budget beyond 3 calls (ablation)ImitSAT · Random 3-SAT 5-15MRPP / W1%No clear effect
Top-K masking of illegal literals instead of VSIDS fallback (ablation)ImitSAT · JNH and AIMMRPPHurt
train only on random 3-SAT, test on structured SATLIB families without tuningImitSAT · SATLIB familiesMRPP: JNH 1.00-0.85, PARITY 0.30Improved
retrain Graph-Q-SAT on same dataset (GQSAT*)Graph-Q-SAT · Random 3-SAT test setsMRPP: e.g. 5-15: 1.00 -> 1.38Hurt
use ImitSAT with 5 calls instead of 3ImitSAT · Random 3-SAT 100 varsMRPP: 0.76 -> 0.83Hurt

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