Visible but Not Avoided: State-Dependent Re-Proposal of Rejected Candidates in Catalog-Based LLM Optimization Loops
LLM optimization loops often show the proposer the candidates it has already tried and rejected, but leave it to the proposer to avoid them.
ProofPaper ↗
Key points
- We ask whether avoidance measured in a static probe predicts avoidance in the states an optimizer's own trajectory produces.
- In catalog-based loops where three open LLMs improve a GUI agent by naming one of 40 edits, a re-proposal is identified exactly by the edit's name.
- Clean base-state probes underpredict re-proposal in the proposers' own loops (macro-averaged loop-minus-probe rate 0.32 [0.27, 0.37]), and proposals of already evaluated harness-edit pairs consume 28-63% of valid proposer iterations.
- For catalog-based loops with exact edit identities, rejection avoidance should be measured on the rejected sets the loop produces.
Sources (1)
- [1]Visible but Not Avoided: State-Dependent Re-Proposal of Rejected Candidates in Catalog-Based LLM Optimization LoopsarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 11, 04:18 PM
LLM optimization loops often show the proposer the candidates it has already tried and rejected, but leave it to the proposer to avoid them.
We ask whether avoidance measured in a static probe predicts avoidance in the states an optimizer's own trajectory produces.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 10, 2026UltraBERT: Beyond the Limit of BERT Pre-Training
- Oct 9, 2026Qwen/Qwen-Image-2.1-Turbo
- Oct 9, 2026Minions: A Dynamic and Distributed Virtual GPU Cloud Built on Idle Apple Silicon Devices
- Oct 9, 2026The Thinking Leakage in Reinforcement Learning for LLMs
- Oct 9, 2026Surfing the Activation Wave: Analyzing the Information and Geometry of LLM Hidden States for Speculative Decoding
- Oct 9, 2026Evaluating LLM Creativity through Long-Tail Performance