Our work makes one of the first attempts at highlighting the potential of post hoc explanations as valuable tools for enhancing the effectiveness of LLMs.

The empirical gains can be striking.

Large Language Models (LLMs) have demonstrated remarkable capabilities in performing complex tasks. Moreover, recent research has shown that incorporating human-annotated rationales (e.g., Chain-of-Thought prompting) during in-context learning can significantly enhance the performance of these models, particularly on tasks that require reasoning capabilities. However, incorporating such rationales poses challenges in terms of scalability as this requires a high degree of human involvement. In this work, we present a novel framework, Amplifying Model Performance by Leveraging In-Context Learning with Post Hoc Explanations (AMPLIFY), which addresses the aforementioned challenges by automating the process of rationale generation.

Large language models (LLMs) are increasingly being applied to clinical care, a domain where both accuracy and transparent reasoning are critical for safe and trustworthy deployment. Chain-of-thought (CoT) prompting, which elicits step-by-step reasoning, has demonstrated improvements in performance and interpretability across a wide range of tasks. However, its effectiveness in clinical contexts remains largely unexplored, particularly in the context of electronic health records (EHRs), the primary source of clinical documentation, which are often lengthy, fragmented, and noisy. In this work, we present the first large-scale systematic study of CoT for clinical text understanding. We assess 95 advanced LLMs on 87 real-world clinical text tasks, covering 9 languages and 8 task types. Contrary to prior findings in other domains, we observe that 86.3\% of models suffer consistent performance degradation in the CoT setting. More capable models remain relatively robust, while weaker ones suffer substantial declines. To better characterize these effects, we perform fine-grained analyses of reasoning length, medical concept alignment, and error profiles, leveraging both LLM-as-a-judge evaluation and clinical expert evaluation. Our results uncover systematic patterns in when and why CoT fails in clinical contexts, which highlight a critical paradox: CoT enhances interpretability but may undermine reliability in clinical text tasks. This work provides an empirical basis for clinical reasoning strategies of LLMs, highlighting the need for transparent and trustworthy approaches.

Mathematical Word Problems (MWPs) are among the most challenging tasks in natural language processing because they require both linguistic understanding and multi-step numerical reasoning. While Chain-of-Thought (CoT) prompting has shown promise, its linear structure often propagates errors, limiting overall effectiveness. To address this limitation, we present the a systematic study of Tree-of-Thought (ToT) reasoning for Bengali MWPs using the SOMADHAN dataset. Owing to computational and token-cost constraints, we evaluate a curated set of 100 representative problems across multiple large language models (LLMs), including GPT-OSS and LLaMA variants, under standard prompting, CoT, and ToT strategies. Our results show that CoT improves baseline accuracy from 78% (standard prompting) to 83% on average, while ToT further increases performance by up to 5 percentage points, achieving 88% accuracy with GPT-OSS-120B. These improvements highlight that ToT is particularly effective in medium-to-large-scale models but may offer less advantage for smaller ones. Overall, our findings establish ToT as a robust framework for solving mathematical problems in low-resource languages such as Bengali. More broadly, this study shows that structured reasoning methods like ToT can provide more reliable and globally consistent outcomes than CoT, paving the way for better reasoning strategies in multilingual NLP.

Prompt design plays a critical role in the reasoning performance of large language models (LLMs), yet the impact of prompt specificity - how detailed or vague a prompt is - remains understudied. This paper introduces DETAIL, a framework for evaluating LLM performance across varying levels of prompt specificity. We generate multi-level prompts using GPT-4, quantify specificity via perplexity, and assess correctness using GPT-based semantic equivalence. Experiments on 30 novel reasoning tasks across GPT-4 and O3-mini reveal that specificity improves accuracy, especially for smaller models and procedural tasks. Our results highlight the need for adaptive prompting strategies and provide tools and data to support further research.

Recent years have witnessed significant progress in Unified Multimodal Models, yet a fundamental question remains: Does understanding truly inform generation? To investigate this, we introduce UniSandbox, a decoupled evaluation framework paired with controlled, synthetic datasets to avoid data leakage and enable detailed analysis. Our findings reveal a significant understanding-generation gap, which is mainly reflected in two key dimensions: reasoning generation and knowledge transfer. Specifically, for reasoning generation tasks, we observe that explicit Chain-of-Thought (CoT) in the understanding module effectively bridges the gap, and further demonstrate that a self-training approach can successfully internalize this ability, enabling implicit reasoning during generation. Additionally, for knowledge transfer tasks, we find that CoT assists the generative process by helping retrieve newly learned knowledge, and also discover that query-based architectures inherently exhibit latent CoT-like properties that affect this transfer. UniSandbox provides preliminary insights for designing future unified architectures and training strategies that truly bridge the gap between understanding and generation. Code and data are available at https://github.com/PKU-YuanGroup/UniSandBox

We study how different Chain-of-Thought (CoT) designs affect the acquisition of the generalizable visual reasoning ability in vision-language models (VLMs). While CoT data, especially long or visual CoT such as "think with image", has been widely used to supervise intermediate reasoning, it remains unclear why specific CoT designs help and which ones truly support generalizable reasoning. To systematically evaluate this, we focus on a controlled maze-solving benchmark where reasoning rules are fully visual, difficulty can be tuned by grid size, and all the intermediate steps can be automatically generated. Using Qwen2.5-VL-7B under a standard SFT-then-RL pipeline, we compare three representative CoT formats: Language CoT, Grounding CoT (with spatial coordinate trajectories), and Visual CoT (with image manipulations). Our experiments reveal that visual and longer CoT mainly accelerate convergence but do not lift the final performance ceiling; concise CoT containing only essential grounding steps outperforms longer traces; and, strikingly, CoT retaining only the minimal grounding results generalizes best across different maze sizes. We further validate these insights on other vision-centric tasks. These findings highlight a "short is long" effect and provide practical guidance for constructing more generalizable SFT datasets for visual reasoning.

Large Language Models (LLMs) are increasingly vulnerable to adversarial attacks that can subtly manipulate their outputs. While various defense mechanisms have been proposed, many operate as black boxes, lacking transparency in their decision-making. This paper introduces ExplainableGuard, an interpretable adversarial defense framework leveraging the chain-of-thought (CoT) reasoning capabilities of DeepSeek-Reasoner. Our approach not only detects and neutralizes adversarial perturbations in text but also provides step-by-step explanations for each defense action. We demonstrate how tailored CoT prompts guide the LLM to perform a multi-faceted analysis (character, word, structural, and semantic) and generate a purified output along with a human-readable justification. Preliminary results on the GLUE Benchmark and IMDB Movie Reviews dataset show promising defense efficacy. Additionally, a human evaluation study reveals that ExplainableGuard's explanations outperform ablated variants in clarity, specificity, and actionability, with a 72.5% deployability-trust rating, underscoring its potential for more trustworthy LLM deployments.

This study aims to systematically evaluate the performance of large language models (LLMs) in abstract visual reasoning problems. We examined four LLM models (GPT-4.1-Mini, Claude-3.5-Haiku, Gemini-1.5-Flash, Llama-3.3-70b) utilizing four different reasoning architectures (single-shot, embedding-controlled repetition, self-reflection, and multi-agent) on the RAVEN-FAIR dataset. Visual responses generated through a three-stage process (JSON extraction, LLM reasoning, and Tool Function) were evaluated using SSIM and LPIPS metrics; Chain-of-Thought scores and error types (semantic hallucination, numeric misperception) were analyzed. Results demonstrate that GPT-4.1-Mini consistently achieved the highest overall accuracy across all architectures, indicating a strong reasoning capability. While the multi-agent architecture occasionally altered semantic and numeric balance across models, these effects were not uniformly beneficial. Instead, each model exhibited distinct sensitivity patterns to architectural design, underscoring that reasoning effectiveness remains model-specific. Variations in response coverage further emerged as a confounding factor that complicates direct cross-architecture comparison. To estimate the upper-bound performance of each configuration, we report the best of five independent runs, representing a best-case scenario rather than an averaged outcome. This multi-run strategy aligns with recent recommendations, which emphasize that single-run evaluations are fragile and may lead to unreliable conclusions.