Cognitive reappraisal is a key strategy in emotion regulation, involving reinterpretation of emotionally charged stimuli to alter affective responses. Despite its central role in clinical and cognitive science, real-world reappraisal interventions remain cognitively demanding, abstract, and primarily verbal. This reliance on higher-order cognitive and linguistic processes is often impaired in individuals with trauma or depression, limiting the effectiveness of standard approaches. Here, we propose a novel, visually based augmentation of cognitive reappraisal by integrating large-scale text-to-image diffusion models into the emotional regulation process. Specifically, we introduce a system in which users reinterpret emotionally negative images via spoken reappraisals, which are transformed into supportive, emotionally congruent visualizations using stable diffusion models with a fine-tuned IP-adapter. This generative transformation visually instantiates users' reappraisals while maintaining structural similarity to the original stimuli, externalizing and reinforcing regulatory intent. To test this approach, we conducted a within-subject experiment (N = 20) using a modified cognitive emotion regulation (CER) task. Participants reappraised or described aversive images from the International Affective Picture System (IAPS), with or without AI-generated visual feedback. Results show that AI-assisted reappraisal significantly reduced negative affect compared to both non-AI and control conditions. Further analyses reveal that sentiment alignment between participant reappraisals and generated images correlates with affective relief, suggesting that multimodal coherence enhances regulatory efficacy. These findings demonstrate that generative visual input can support cogitive reappraisal and open new directions at the intersection of generative AI, affective computing, and therapeutic technology.

Large Language Models have shown impressive capabilities in coding tasks like code generation and code completion, as they have been trained on a large amount of code data. Also, since one of the core pretraining objectives is Next Token Prediction, these models tends to learn surface-level syntactic patterns in code. However, this does not guarantee code comprehension ability i.e. the ability to capture the semantics of the code. In our opinion, this is the reason why these models often underperform on tasks that require deeper semantic understanding, such as code debugging and code optimization. To address this, we propose fine-tuning these models specifically for code comprehension tasks using large-scale datasets, enabling them to develop a more robust understanding of code semantics. We evaluate three code models of varying sizes on a suite of code comprehension tasks designed to assess semantic understanding beyond surface-level syntactic pattern matching. In particular, we analyze performance on the Subjectivity Grading Task and observe that model performance improves after fine-tuning on relevant downstream tasks. The most significant improvement is seen in the QWQ-32B model, where accuracy increases from 70% to 83.47%. A similar or explainable trend is observed across other models, clearly indicating an enhancement in code comprehension ability. Among the models studied, the DPO-fine-tuned Codestral-22B achieves the highest micro-accuracy of 87.66% on the Subjectivity Grading Task.

Transforming natural language into SQL queries (NL2SQL) is crucial for data-driven business applications. Existing frameworks, trained on open-source datasets, struggle with complex business logic and lack domain-specific data for fine-tuning. Additionally, evaluation methods often require annotated data and executable database environments, which are scarce in real-world scenarios. To address these challenges, we propose SQLord, an enterprise-level NL2SQL framework. First, SQLord introduces a data reverse generation approach to convert raw SQL statements into annotated data for supervised fine-tuning (SFT). Second, it proposes a decomposition method for complex queries using an automated workflow generator. Additionally, SQLord features a comprehensive GPT-Judge evaluation framework, including Execution Evaluation (EXE), Query-SQL Evaluation (QSE), and SQL-SQL Evaluation (SSE), tailored to diverse scenarios. Offline tests significantly outperform state of the art baselines, and online accuracy consistently exceeds 90, highlighting SQLord's advantages and effectiveness in complex real world scenarios. SQLord has been successfully applied across multiple scenarios on the world's largest B2B e-commerce platform.

Algorithmic innovation in the pretraining of large language models has driven a massive reduction in the total compute required to reach a given level of capability. In this paper we empirically investigate the compute requirements for developing algorithmic innovations. We catalog 36 pre-training algorithmic innovations used in Llama 3 and DeepSeek-V3. For each innovation we estimate both the total FLOP used in development and the FLOP/s of the hardware utilized. Innovations using significant resources double in their requirements each year. We then use this dataset to investigate the effect of compute caps on innovation. Our analysis suggests that compute caps alone are unlikely to dramatically slow AI algorithmic progress. Even stringent compute caps -- such as capping total operations to the compute used to train GPT-2 or capping hardware capacity to 8 H100 GPUs -- could still have allowed for half of the cataloged innovations.

Graphical user interface (GUI) agents built on multimodal large language models (MLLMs) have recently demonstrated strong decision-making abilities in screen-based interaction tasks. However, they remain highly vulnerable to pop-up-based environmental injection attacks, where malicious visual elements divert model attention and lead to unsafe or incorrect actions. Existing defense methods either require costly retraining or perform poorly under inductive interference. In this work, we systematically study how such attacks alter the attention behavior of GUI agents and uncover a layer-wise attention divergence pattern between correct and incorrect outputs. Based on this insight, we propose \textbf{LaSM}, a \textit{Layer-wise Scaling Mechanism} that selectively amplifies attention and MLP modules in critical layers. LaSM improves the alignment between model saliency and task-relevant regions without additional training. Extensive experiments across 12 types of pop-up perturbations and 4 different model backbones show that LaSM consistently enhances the defense success rate. When combined with prompt-level alerts, LaSM achieves over 98\% robustness even under strong inductive attacks. Our findings reveal that attention misalignment is a core vulnerability in MLLM agents and can be effectively addressed through selective layer-wise modulation.

--Large Language Models (LLMs) excel in text classification, but their complexity hinders interpretability, making it difficult to understand the reasoning behind their predictions. Explainable AI (XAI) methods like LIME and SHAP offer local explanations by identifying influential words, but they rely on computationally expensive perturbations. These methods typically generate thousands of perturbed sentences and perform inferences on each, incurring a substantial computational burden, especially with LLMs. T o address this, we propose P erturbation-free L ocal Ex planation (PLEX), a novel method that leverages the contextual embeddings extracted from the LLM and a "Siamese network" style neural network trained to align with feature importance scores. This one-off training eliminates the need for subsequent perturbations, enabling efficient explanations for any new sentence. We demonstrate PLEX's effectiveness on four different classification tasks (sentiment, fake news, fake COVID-19 news and depression), showing more than 92% agreement with LIME and SHAP . Our evaluation using a "stress test" reveals that PLEX accurately identifies influential words, leading to a similar decline in classification accuracy as observed with LIME and SHAP when these words are removed. Notably, in some cases, PLEX demonstrates superior performance in capturing the impact of key features. PLEX dramatically accelerates explanation, reducing time and computational overhead by two and four orders of magnitude, respectively. This work offers a promising solution for explainable LLM-based text classification. ARGE language models (LLMs) have significantly advanced text classification, achieving state-of-the-art results in tasks like emotion recognition, sentiment analysis, topic categorization, and spam detection [1]. Powered by transformer architectures with millions or billions of parameters, they effectively capture complex linguistic patterns. However, the very complexity that enables their high performance also renders their internal workings opaque and difficult to interpret.

Human users increasingly rely on natural language interactions with large language models (LLMs) in order to receive help on a large variety of tasks and problems. However, the trustworthiness and perceived legitimacy of LLMs is undermined by the fact that their output is frequently stated in very confident terms, even when its accuracy is questionable. Therefore, there is a need to signal the confidence of the language model to a user in order to reap the benefits of human-machine collaboration and mitigate potential harms. Verbalized uncertainty is the expression of confidence with linguistic means, an approach that integrates perfectly into language-based interfaces. Nevertheless, most recent research in natural language processing (NLP) overlooks the nuances surrounding human uncertainty communication and the data biases that influence machine uncertainty communication. We argue for anthropomimetic uncertainty, meaning that intuitive and trustworthy uncertainty communication requires a degree of linguistic authenticity and personalization to the user, which could be achieved by emulating human communication. We present a thorough overview over the research in human uncertainty communication, survey ongoing research, and perform additional analyses to demonstrate so-far overlooked biases in verbalized uncertainty. We conclude by pointing out unique factors in human-machine communication of uncertainty and deconstruct anthropomimetic uncertainty into future research directions for NLP.

Large Language Models (LLMs) have demonstrated remarkable fluency across a range of natural language tasks, yet remain vulnerable to hallucinations - factual inaccuracies that undermine trust in real world deployment. We present AutoRAG-LoRA, a modular framework for Retrieval-Augmented Generation (RAG) that tackles hallucination in large language models through lightweight LoRA-based adapters and KL-regularized training. Our pipeline integrates automated prompt rewriting, hybrid retrieval, and low-rank adapter tuning to ground responses in retrieved evidence. A hallucination detection module, using both classifier-based and self-evaluation techniques, assigns confidence scores to generated outputs, triggering an optional feedback correction loop. This loop enforces factual alignment via contrastive KL loss and adapter fine tuning. We demonstrate that AutoRAG-LoRA significantly reduces the factual drift while preserving the efficiency and modularity of the model.

Recent Large Reasoning Models (LRMs) have achieved remarkable progress on task-specific benchmarks, yet their evaluation methods remain constrained by isolated problem-solving paradigms. Existing benchmarks predominantly assess single-question reasoning through sequential testing, resulting critical limitations: (1) vulnerability to data contamination and less challenging (e.g., DeepSeek-R1 achieves 97.0% on MATH500), forcing costly creation of new questions with large human efforts, (2) failure to evaluate models under multi-context pressure, a key requirement for real-world deployment. To bridge this gap, we present REST (Reasoning Evaluation through Simultaneous Testing), a stress-testing framework that exposes LRMs to multiple problems simultaneously. Beyond basic reasoning, REST evaluates several under-tested capabilities: contextual priority allocation, cross-problem interference resistance, and dynamic cognitive load management. Our evaluation reveals several striking findings: Even state-of-the-art (SOTA) models like DeepSeek-R1 exhibit substantial performance degradation under stress testing. Crucially, REST demonstrates stronger discriminative power than existing benchmarks, revealing pronounced performance differences among models that exhibit similar, near-ceiling performance under single-question evaluations. Some key insights emerge from our analysis: (1) the "overthinking trap" is a critical factor contributing to the performance degradation; (2) the models trained with "long2short" technique preserve more accuracy of their single-problem performance under REST, outperforming standard-trained counterparts. These results establish REST as a cost-efficient, future-proof evaluation paradigm that better reflects real-world reasoning demands while reducing reliance on continuous human annotation. Code and results are available at https://opendatalab.github.io/REST.

Fine-tuning methods such as Direct Preference Optimization (DPO) and Group Relative Policy Optimization (GRPO) have demonstrated success in training large language models (LLMs) for single-turn tasks. However, these methods fall short in multi-turn applications, such as diagnostic patient interviewing, where understanding how early conversational turns influence downstream completions and outcomes is essential. In medicine, a multi-turn perspective is critical for learning diagnostic schemas and better understanding conversation dynamics. To address this gap, I introduce Savage Conversation Forests (SCF), a reinforcement learning framework that leverages a branched conversation architecture to fine-tune LLMs for multi-turn dialogue. SCF generates multiple possible conversation continuations at each turn, enabling the model to learn how different early responses affect downstream interactions and diagnostic outcomes. In experiments simulating doctor-patient conversations, SCF with branching outperforms linear conversation architectures on diagnostic accuracy. I hypothesize that SCF's improvements stem from its ability to provide richer, interdependent training signals across conversation turns. These results suggest that a branched training architecture is an important strategy for fine tuning LLMs in complex multi-turn conversational tasks.