Compute scaling for language model (LM) pretraining has outpaced the growth of human-written texts, leading to concerns that data will become the bottleneck to LM scaling. To continue scaling pretraining in this data-constrained regime, we propose that explicitly modeling and inferring the \emph{latent thoughts} that underlie the text generation process can significantly improve pretraining data efficiency. Intuitively, our approach views web text as the compressed final outcome of a verbose human thought process and that the latent thoughts contain important contextual knowledge and reasoning steps that are critical to data-efficient learning. We empirically demonstrate the effectiveness of our approach through data-constrained continued pretraining for math. We first show that synthetic data approaches to inferring latent thoughts significantly improve data efficiency over training on the same amount of raw data. Furthermore, we demonstrate latent thought inference without a strong teacher, where an LM \emph{bootstraps its own performance} by using an EM algorithm to iteratively improve the capability of the trained LM and the quality of thought-augmented pretraining data. We show that a 1B LM can bootstrap its performance across at least three iterations and significantly outperform baselines trained on raw data, with increasing gains from additional inference compute when performing the E-step. The gains from inference scaling and EM iterations suggest new opportunities for scaling data-constrained pretraining.

Chain-of-Thought reasoning has emerged as a pivotal methodology for enhancing model inference capabilities. Despite growing interest in Chain-of-Thought reasoning, its underlying mechanisms remain unclear. This paper explores the working mechanisms of Chain-of-Thought reasoning from the perspective of the dual relationship between in-context learning and pretrained priors. We first conduct a fine-grained lexical-level analysis of rationales to examine the model's reasoning behavior. Then, by incrementally introducing noisy exemplars, we examine how the model balances pretrained priors against erroneous in-context information. Finally, we investigate whether prompt engineering can induce slow thinking in large language models. Our extensive experiments reveal three key findings: (1) The model not only quickly learns the reasoning structure at the lexical level but also grasps deeper logical reasoning patterns, yet it heavily relies on pretrained priors.

This paper introduces a multi-agent framework for comprehensive highway scene understanding, designed around a mixture-of-experts strategy. In this framework, a large generic vision-language model (VLM), such as GPT-4o, is contextualized with domain knowledge to generates task-specific chain-of-thought (CoT) prompts. These fine-grained prompts are then used to guide a smaller, efficient VLM (e.g., Qwen2.5-VL-7B) in reasoning over short videos, along with complementary modalities as applicable. The framework simultaneously addresses multiple critical perception tasks, including weather classification, pavement wetness assessment, and traffic congestion detection, achieving robust multi-task reasoning while balancing accuracy and computational efficiency. To support empirical validation, we curated three specialized datasets aligned with these tasks. Notably, the pavement wetness dataset is multimodal, combining video streams with road weather sensor data, highlighting the benefits of multimodal reasoning. Experimental results demonstrate consistently strong performance across diverse traffic and environmental conditions. From a deployment perspective, the framework can be readily integrated with existing traffic camera systems and strategically applied to high-risk rural locations, such as sharp curves, flood-prone lowlands, or icy bridges. By continuously monitoring the targeted sites, the system enhances situational awareness and delivers timely alerts, even in resource-constrained environments.

However, prior studies have shown that these models often incur significant computational costs due to excessive reasoning, such as frequent switching between reasoning trajectories (e.g., underthinking) or redundant reasoning on simple questions (e.g., overthinking). In this work, we expose a novel threat: adversarial inputs can be crafted to exploit excessive reasoning behaviors and substantially increase computational overhead without compromising model utility. Therefore, we propose a novel loss framework consisting of three components: (1) Priority Cross-Entropy Loss, a modification of the standard cross-entropy objective that emphasizes key tokens by leveraging the autoregressive nature of LMs; (2) Excessive Reasoning Loss, which encourages the model to initiate additional reasoning paths during inference; and (3) Delayed Termination Loss, which is designed to extend the reasoning process and defer the generation of final outputs. We optimize and evaluate our attack for the GSM8K and ORCA datasets on DeepSeek-R1-Distill-LLaMA and DeepSeek-R1-Distill-Qwen. Empirical results demonstrate a 3x to 9x increase in reasoning length with comparable utility performance. Furthermore, our crafted adversarial inputs exhibit transferability, inducing computational overhead in o3-mini, o1-mini, DeepSeek-R1, and QWQ models.

Societies worldwide are rapidly entering a super-aged era, making elderly health a pressing concern. The aging population is increasing the burden on national economies and households. Dementia cases are rising significantly with this demographic shift. Recent research using voice-based models and large language models (LLM) offers new possibilities for dementia diagnosis and treatment. Our Chain-of-Thought (CoT) reasoning method combines speech and language models. The process starts with automatic speech recognition to convert speech to text. We add a linear layer to an LLM for Alzheimer's disease (AD) and non-AD classification, using supervised fine-tuning (SFT) with CoT reasoning and cues. This approach showed an 16.7% relative performance improvement compared to methods without CoT prompt reasoning. To the best of our knowledge, our proposed method achieved state-of-the-art performance in CoT approaches.

Fully comprehending scientific papers by machines reflects a high level of Artificial General Intelligence, requiring the ability to reason across fragmented and heterogeneous sources of information, presenting a complex and practically significant challenge. While Vision-Language Models (VLMs) have made remarkable strides in various tasks, particularly those involving reasoning with evidence source from single image or text page, their ability to use cross-source information for reasoning remains an open problem. This work presents MMCR, a high-difficulty benchmark designed to evaluate VLMs' capacity for reasoning with cross-source information from scientific papers. The benchmark comprises 276 high-quality questions, meticulously annotated by humans across 7 subjects and 10 task types. Experiments with 18 VLMs demonstrate that cross-source reasoning presents a substantial challenge for existing models. Notably, even the top-performing model, GPT-4o, achieved only 48.55% overall accuracy, with only 20% accuracy in multi-table comprehension tasks, while the second-best model, Qwen2.5-VL-72B, reached 39.86% overall accuracy. Furthermore, we investigated the impact of the Chain-of-Thought (CoT) technique on cross-source reasoning and observed a detrimental effect on small models, whereas larger models demonstrated substantially enhanced performance. These results highlight the pressing need to develop VLMs capable of effectively utilizing cross-source information for reasoning.

One of the most widely used methods to evaluate LLMs are Multiple Choice Question (MCQ) tests. MCQ benchmarks enable the testing of LLM knowledge on almost any topic at scale as the results can be processed automatically. To help the LLM answer, a few examples called few shots can be included in the prompt. Moreover, the LLM can be asked to answer the question directly with the selected option or to first provide the reasoning and then the selected answer, which is known as chain of thought. In addition to checking whether the selected answer is correct, the evaluation can look at the LLM-estimated probability of its response as an indication of the confidence of the LLM in the response. In this paper, we study how the LLM confidence in its answer depends on whether the model has been asked to answer directly or to provide the reasoning before answering. The results of the evaluation of questions on a wide range of topics in seven different models show that LLMs are more confident in their answers when they provide reasoning before the answer. This occurs regardless of whether the selected answer is correct. Our hypothesis is that this behavior is due to the reasoning that modifies the probability of the selected answer, as the LLM predicts the answer based on the input question and the reasoning that supports the selection made. Therefore, LLM estimated probabilities seem to have intrinsic limitations that should be understood in order to use them in evaluation procedures. Interestingly, the same behavior has been observed in humans, for whom explaining an answer increases confidence in its correctness.

Previous works have demonstrated the effectiveness of Chain-of-Thought (COT) prompts and verifiers in guiding Large Language Models (LLMs) through the space of reasoning. However, most such studies either use a fine-tuned verifier or rely on manually handcrafted few-shot examples. In contrast, in this paper, we focus on LLM-based self-verification of self-generated reasoning steps via COT prompts in a completely zero-shot regime. To explore this setting, we design a new zero-shot prompt, which we call COT STEP, to aid zero-shot decomposition of reasoning steps and design two new zero-shot prompts for LLM-based verifiers. We evaluate the verifiers' ability to classify the correctness of reasoning chains and explore different ways to use verifier scores in guiding reasoning for various mathematical and commonsense reasoning tasks with different LLMs.

Empathetic dialogue is crucial for natural human-computer interaction, allowing the dialogue system to respond in a more personalized and emotionally aware manner, improving user satisfaction and engagement. The emergence of large language models (LLMs) has revolutionized dialogue generation by harnessing their powerful capabilities and shown its potential in multimodal domains. Many studies have integrated speech with text-based LLMs to take speech question as input and output text response. However, the lack of spoken question-answering datasets that include speech style information to supervised fine-tuning (SFT) limits the performance of these systems. As a result, while these systems excel at understanding speech content, they often struggle to generate empathetic responses. In response, we propose a novel approach that circumvents the need for question-answering data, called Listen, Perceive, and Express (LPE). Our method employs a two-stage training process, initially guiding the LLM to listen the content and perceive the emotional aspects of speech. Subsequently, we utilize Chain-of-Thought (CoT) prompting to unlock the model's potential for expressing empathetic responses based on listened spoken content and perceived emotional cues. We employ experiments to prove the effectiveness of proposed method. To our knowledge, this is the first attempt to leverage CoT for speech-based dialogue.

We propose a variant of chain of thought (CoT) prompting called Program Trace Prompting that makes explanations more observable while preserving the power, generality and flexibility of CoT. In our approach, few-shot CoT demonstrations are wrapped in a formal syntax based on Python, and each prompt: identifies and names steps; defines the input/output behavior of steps; and replaces CoT explanations of in-context examples with chains of these formalized steps on the same examples. Program Trace Prompting is applicable to many tasks, achieving strong results on the 23 diverse tasks in the BIG-Bench Hard benchmark. More importantly, by instrumenting explanations in this way, we enable new types of analysis. In particular, we identify "non-local errors" (which correspond to incorrectly learning the reasoning method illustrated in the demonstrations) as an unaddressed issue in CoT learning, and we present methods for verifying the modularity of steps in a CoT explanation.