Recent advances in LLM agents have largely built on reasoning backbones like ReAct (Yao et al., 2023), which interleaves thought and action in complex environments.However, Re-Act often produces ungrounded or incoherent reasoning steps, leading to misalignment between the agent's actual state and goal.Our analysis finds that this stems from ReAct's inability to maintain consistent internal beliefs and goal alignment, causing compounding errors and hallucinations.To address this, we introduce ReflAct, a novel backbone that shifts reasoning from merely planning next actions to continuously reflecting on the agent's state relative to its goal.By explicitly grounding decisions in states and enforcing ongoing goal alignment, ReflAct dramatically improves strategic reliability.This design delivers substantial empirical gains: ReflAct surpasses Re-Act by 27.7% on average, achieving a 93.3% success rate in ALFWorld.Notably, ReflAct even outperforms ReAct with added enhancement modules (e.g., Reflexion, WKM), showing that strengthening the core reasoning backbone is key to reliable agent performance.Observation: You pick up the apple 1 from the countertop 1. Thought: Now I take an apple 1.Next, I need to go to a microwave 1 and heat it.Action: go to microwave 1 Observation: The microwave 1 is closed.Thought: The microwave 1 is closed, so I need to open it first.Action: open microwave 1 Observation: You open the microwave 1.The microwave 1 is open.In it, you see nothing.

As large language models (LLMs) are increasingly used across various applications, there is a growing need to control text generation to satisfy specific constraints or requirements. This raises a crucial question: Is it possible to guarantee strict constraint satisfaction in generated outputs while preserving the distribution of the original model as much as possible? We first define the ideal distribution — the one closest to the original model, which also always satisfies the expressed constraint — as the ultimate goal of guaranteed generation. We then state a fundamental limitation, namely that it is impossible to reach that goal through autoregressive training alone. This motivates the necessity of combining training-time and inference-time methods to enforce such guarantees. Based on this insight, we propose GUARD, a simple yet effective approach that combines an autoregressive proposal distribution with rejection sampling. Through GUARD’s theoretical properties, we show how controlling the KL divergence between a specific proposal and the target ideal distribution simultaneously optimizes inference speed and distributional closeness. To validate these theoretical concepts, we conduct extensive experiments on two text generation settings with hard-to-satisfy constraints: a lexical constraint scenario and a sentiment reversal scenario. These experiments show that GUARD achieves perfect constraint satisfaction while almost preserving the ideal distribution with highly improved inference efficiency. GUARD provides a principled approach to enforcing strict guarantees for LLMs without compromising their generative capabilities.

Minbeom Kim, Hwanhee Lee, Joonsuk Park, Hwaran Lee, Kyomin Jung. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

We introduce transductive program synthesis, a new formulation of the program synthesis task that explicitly leverages test inputs during synthesis. While prior approaches to program synthesis--whether based on natural language descriptions or input-output examples--typically aim to generalize from training examples, they often struggle with robustness, especially in real-world settings where training examples are limited and test inputs involve various edge cases. To address this, we propose a novel framework that improves robustness by treating synthesis as an active learning over a finite hypothesis class defined by programs' outputs. We use an LLM to predict outputs for selected test inputs and eliminate inconsistent hypotheses, where the inputs are chosen via a greedy maximin algorithm to minimize the number of LLM queries required. We evaluate our approach on four benchmarks: Playgol, MBPP+, 1D-ARC, and programmatic world modeling on MiniGrid. We demonstrate that our method significantly improves program synthesis in both accuracy and efficiency. We release our code at https://github.com/klee972/SYNTRA.

As video analysis using deep learning models becomes more widespread, the vulnerability of such models to adversarial attacks is becoming a pressing concern. In particular, Universal Adversarial Perturbation (UAP) poses a significant threat, as a single perturbation can mislead deep learning models on entire datasets. We propose a novel video UAP using image data and image model. This enables us to take advantage of the rich image data and image model-based studies available for video applications. However, there is a challenge that image models are limited in their ability to analyze the temporal aspects of videos, which is crucial for a successful video attack. To address this challenge, we introduce the Breaking Temporal Consistancy (BTC) method, which is the first attempt to incorporate temporal information into video attacks using image models. We aim to generate adversarial videos that have opposite patterns to the original. Specifically, BTC-UAP minimizes the feature similarity between neighboring frames in videos. Our approach is simple but effective at attacking unseen video models. Additionally, it is applicable to videos of varying lengths and invariant to temporal shifts. Our approach surpasses existing methods in terms of effectiveness on various datasets, including ImageNet, UCF-101, and Kinetics-400.

Thermal generation by power dissipation of the highly integrated System on Chip (SoC) device is irregularly distributed on the intra chip. It leads to thermal increment of the each thermally different region and effects on the propagation timing; consequently, the timing violation occurs due to the misestimated number of buffers. In this paper, the timing budgeting methodology which contains the estimated thermal distribution on a chip in the early stage of physical design by modeled the RC delay considering temperature and buffer insertion planning using by the proposed delay model are presented. Simulation results showed the reduction of the worst delay after buffer insertion up to 30% in contrast to the buffer insertion without considering temperature.

Three-dimensional integrated circuits (3-D IC) are available through the die-stacking and through-silicon-via (TSV) technologies. The 3-D IC using the TSV brings the performance improvement through the minimization of wire length and footprint area. However, the 3-D ICs have many challenges for power delivery network design due to larger supply currents and longer power delivery paths compared to 2-D ICs. The power delivery network in 3-D IC with flip chip package is largely composed of power/ground (P/G) bumps and P/G TSVs. Because the number of P/G bumps is limited and the size of P/G TSV is larger than that of standard cell, it is important to optimize the P/G bumps and P/G TSVs together while satisfying the IR-drop constraint. In this paper, we investigated an effect of the number of power bumps and power TSVs on the IR-drop in 3-D IC floorplan level and proposed the methodology that reduces the number of power bumps by 88.25% on average while the number of power TSVs and maximum IR-drop are comparable to the previous methodology.