<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title>Bing: MATLAB Script Icon</title><link>http://www.bing.com:80/search?q=MATLAB+Script+Icon</link><description>Search results</description><image><url>http://www.bing.com:80/s/a/rsslogo.gif</url><title>MATLAB Script Icon</title><link>http://www.bing.com:80/search?q=MATLAB+Script+Icon</link></image><copyright>Copyright © 2026 Microsoft. All rights reserved. These XML results may not be used, reproduced or transmitted in any manner or for any purpose other than rendering Bing results within an RSS aggregator for your personal, non-commercial use. Any other use of these results requires express written permission from Microsoft Corporation. By accessing this web page or using these results in any manner whatsoever, you agree to be bound by the foregoing restrictions.</copyright><item><title>Clever: A Curated Benchmark for Formally Verified Code Generation</title><link>https://openreview.net/attachment?id=pqNFDA2TFm&amp;name=pdf</link><description>We introduce CLEVER, the first curated benchmark for evaluating the generation of specifications and formally verified code in Lean. The benchmark comprises of 161 programming problems; it evaluates both formal speci-fication generation and implementation synthesis from natural language, requiring formal correctness proofs for both.</description><pubDate>Thu, 28 May 2026 13:29:00 GMT</pubDate></item><item><title>Evaluating the Robustness of Neural Networks: An Extreme Value...</title><link>https://openreview.net/forum?id=BkUHlMZ0b</link><description>Our analysis yields a novel robustness metric called CLEVER, which is short for Cross Lipschitz Extreme Value for nEtwork Robustness. The proposed CLEVER score is attack-agnostic and is computationally feasible for large neural networks.</description><pubDate>Sun, 31 May 2026 13:18:00 GMT</pubDate></item><item><title>CLEVER: A Curated Benchmark for Formally Verified Code Generation</title><link>https://openreview.net/forum?id=pqNFDA2TFm</link><description>TL;DR: We introduce CLEVER, a hand-curated benchmark for verified code generation in Lean. It requires full formal specs and proofs. No few-shot method solves all stages, making it a strong testbed for synthesis and formal reasoning.</description><pubDate>Fri, 29 May 2026 03:55:00 GMT</pubDate></item><item><title>The Clever Hans Mirage: A Comprehensive Survey on Spurious...</title><link>https://openreview.net/forum?id=kIuqPmS1b1</link><description>Back in the early 20th century, a horse named Hans appeared to perform arithmetic and other intellectual tasks during exhibitions in Germany, while it actually relied solely on involuntary cues in...</description><pubDate>Sun, 31 May 2026 03:38:00 GMT</pubDate></item><item><title>On the Planning Abilities of Large Language Models : A Critical ...</title><link>https://openreview.net/pdf?id=X6dEqXIsEW</link><description>While, as we mentioned earlier, there can be thorny “clever hans” issues about humans prompting LLMs, an automated verifier mechanically backprompting the LLM doesn’t suffer from these. We tested this setup on a subset of the failed instances in the one-shot natural language prompt configuration using GPT-4, given its larger context window.</description><pubDate>Fri, 29 May 2026 22:25:00 GMT</pubDate></item><item><title>Djork-Arné Clevert - OpenReview</title><link>https://openreview.net/profile?id=~Djork-Arn%C3%A9_Clevert2</link><description>Promoting openness in scientific communication and the peer-review process</description><pubDate>Mon, 25 May 2026 23:34:00 GMT</pubDate></item><item><title>RL's Razor: Why Online Reinforcement Learning Forgets Less</title><link>https://openreview.net/forum?id=7HNRYT4V44</link><description>The ParityMNIST experiment is genuinely clever. By constructing an oracle SFT distribution that provably minimizes KL, the authors demonstrate that RL's advantage comes from implicit KL minimization rather than something inherent to the RL objective itself. When SFT uses this oracle distribution, it matches or exceeds RL's performance.</description><pubDate>Sun, 31 May 2026 23:19:00 GMT</pubDate></item><item><title>Sparse but Critical: A Token-Level Analysis of Distributional...</title><link>https://openreview.net/forum?id=8vWIXno8LW</link><description>Main claims of the paper are supported, but the execution of experiments might be significantly strenghtened further (see weaknesses). S2. Experiments with cross-sampling and advantage reweighting are interesting and clever and seem to be a promising analysis toolkit; however, see W3 and W4.</description><pubDate>Sat, 30 May 2026 04:51:00 GMT</pubDate></item><item><title>STAIR: Improving Safety Alignment with Introspective Reasoning</title><link>https://openreview.net/forum?id=aHzPGyUhZa</link><description>One common approach is training models to refuse unsafe queries, but this strategy can be vulnerable to clever prompts, often referred to as jailbreak attacks, which can trick the AI into providing harmful responses. Our method, STAIR (SafeTy Alignment with Introspective Reasoning), guides models to think more carefully before responding.</description><pubDate>Thu, 28 May 2026 18:15:00 GMT</pubDate></item><item><title>Counterfactual Debiasing for Fact Verification</title><link>https://openreview.net/pdf?id=BddNTCq65yq</link><description>579 In this paper, we have proposed a novel counter- factual framework CLEVER for debiasing fact- checking models. Unlike existing works, CLEVER is augmentation-free and mitigates biases on infer- ence stage. In CLEVER, the claim-evidence fusion model and the claim-only model are independently trained to capture the corresponding information.</description><pubDate>Tue, 26 May 2026 05:03:00 GMT</pubDate></item></channel></rss>