AI Paper Atlas

Research map / Large language models / Agents, code and math

Towards Feedback-to-Plan Decisions for Self-Evolving LLM Agents in CUDA Kernel Generation

Yee Hin Chong, Jiaming Wu, Youhui Zhang, Peng Qu

ICML 2026 · Large language models › Agents, code and math · approach: Long-horizon LLM agents · 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
add explicit planning without feedback (P+NF) (ablation)OpenEvolve CUDA agent · PolyBench-ACCgeneration-level success rateHurt
add aligned feedback to explicit planning (P+F) (ablation)explicit planning · PolyBench-ACCgeneration-level success rateImproved
replace planner output with content-free DummyPlan (ablation)explicit planning · PolyBench-ACCgeneration-level success rateHurt
randomize feedback across programs (P+RF) (ablation)explicit planning · PolyBench-ACCgeneration-level success rateHurt
add summarization of feedback (P+S vs P+F) (ablation)explicit planning, weak models · PolyBench-ACCoverall successImproved
add summarization of feedback (P+S vs P+F) (ablation)explicit planning, strong models · PolyBench-ACCoverall successNo clear effect
add explicit planning over summaries alone (P+S vs NP+S) (ablation)summarized feedback · PolyBench-ACCexecution successImproved
inject strong-model plans into weak model (ablation)DeepSeek-V3.2, Qwen3-Coder-30B · PolyBench-ACCexecution successImproved
use same-family plan distillation (ablation)plan transfer · PolyBench-ACCexecution successImproved
raise attempts k beyond 5 (ablation)pass@k evaluation · PolyBench-ACC 3DCONV, DeepSeek-V3.2pass rateNo clear effect
add multi-tool feedback interaction (debugger, analyzer, profiler jointly)planning · PolyBench-ACCsynergy sigma_FP: 0.233 -> 0.430Improved

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