Keywords: Visual - language model Biomedical artificial intelligence Model scaling Instruction tuning Multimodal learning ABSTRACT Objective T o advance b iomedical v ison - language m odel capabilities through scaling up, fine - tuning, and instruction tuning, develop vision - language models with improved performance in handling long text, explore strategies to efficiently adopt vision langu a ge mode l s for diverse multi - modal biomedical tasks, and examine the zero - shot learning performance. Methods We developed two biomedical vision language models, BiomedGPT - Large and BiomedGPT - XLarge, based on a n encoder - decoder - based transformer architecture. We fine - tuned the two models on 23 benchmark datasets from 6 multi - modal biomedical tasks including one image - only task (image classification), three language - only tasks (text understanding, text summarization and question answering), and two vision - language tasks (visual question answering and image captioning) . We compared the developed scaled models with our previous BiomedGPT - Base model and existi ng prestigious models reported in the literature . W e instruction - tuned the two models using a large - scale multi - modal biomedical instruction - tuning dataset and assessed the zero - shot learning performance and alignment accuracy . Results and Conclusion The experimental results show that the new models developed in this study outperform our previous BiomedGPT - Base model on 17 of 2 3 benchmark datasets and achiev e state - of - the - art performance on 15 of 23 datasets when compared to previous models reported in the literature . The new models also demonstrated improved ability in handling long text, particularly on text summarization on MIMIC - III dataset and text understanding on SEER dataset, with a remarkable improvement of 4.6~11.4 I nstruction tuning on the scaled models resulted in significant enhancements in zero - shot learning ability and alignment accuracy in following complex instructions across multiple tasks, including image classification, visual question answering, and image captioning . This study develop s two vision - language models in the biomedical domain and examine s technologies to improve long text content in vision language models through scaling, fine - tuning, and instruction tuning .

The emergence of multi-codebook neutral audio codecs such as Residual Vector Quantization (RVQ) and Group Vector Quantization (GVQ) has significantly advanced Large-Language-Model (LLM) based Text-to-Speech (TTS) systems. These codecs are crucial in separating semantic and acoustic information while efficiently harnessing semantic priors. However, since semantic and acoustic information cannot be fully aligned, a significant drawback of these methods when applied to LLM-based TTS is that large language models may have limited access to comprehensive audio information. To address this limitation, we propose DistilCodec and UniTTS, which collectively offer the following advantages: 1) This method can distill a multi-codebook audio codec into a single-codebook audio codec with 32,768 codes while achieving a near 100\% utilization. 2) As DistilCodec does not employ a semantic alignment scheme, a large amount of high-quality unlabeled audio (such as audiobooks with sound effects, songs, etc.) can be incorporated during training, further expanding data diversity and broadening its applicability. 3) Leveraging the comprehensive audio information modeling of DistilCodec, we integrated three key tasks into UniTTS's pre-training framework: audio modality autoregression, text modality autoregression, and speech-text cross-modal autoregression. This allows UniTTS to accept interleaved text and speech/audio prompts while substantially preserving LLM's text capabilities. 4) UniTTS employs a three-stage training process: Pre-Training, Supervised Fine-Tuning (SFT), and Alignment. Source code and model checkpoints are publicly available at https://github.com/IDEA-Emdoor-Lab/UniTTS and https://github.com/IDEA-Emdoor-Lab/DistilCodec.

Multimodal models like CLIP have gained significant attention due to their remarkable zero-shot performance across various tasks. However, studies have revealed that CLIP can inadvertently learn spurious associations between target variables and confounding factors. To address this, we introduce \textsc{Locate-Then-Correct} (LTC), a contrastive framework that identifies spurious attention heads in Vision Transformers via mechanistic insights and mitigates them through targeted ablation. Furthermore, LTC identifies salient, task-relevant attention heads, enabling the integration of discriminative features through orthogonal projection to improve classification performance. We evaluate LTC on benchmarks with inherent background and gender biases, achieving over a $>50\%$ gain in worst-group accuracy compared to non-training post-hoc baselines. Additionally, we visualize the representation of selected heads and find that the presented interpretation corroborates our contrastive mechanism for identifying both spurious and salient attention heads. Code available at https://github.com/wj210/CLIP_LTC.

Large language models (LLMs) have achieved remarkable performance across a wide range of NLP tasks. However, their substantial inference cost poses a major barrier to real-world deployment, especially in latency-sensitive scenarios. To address this challenge, we propose \textbf{DASH}, an adaptive layer-skipping framework that dynamically selects computation paths conditioned on input characteristics. We model the skipping process as a Markov Decision Process (MDP), enabling fine-grained token-level decisions based on intermediate representations. To mitigate potential performance degradation caused by skipping, we introduce a lightweight compensation mechanism that injects differential rewards into the decision process. Furthermore, we design an asynchronous execution strategy that overlaps layer computation with policy evaluation to minimize runtime overhead. Experiments on multiple LLM architectures and NLP benchmarks show that our method achieves significant inference acceleration while maintaining competitive task performance, outperforming existing methods.

Generative error correction (GER) with large language models (LLMs) has emerged as an effective post-processing approach to improve automatic speech recognition (ASR) performance. However, it often struggles with rare or domain-specific words due to limited training data. Furthermore, existing LLM-based GER approaches primarily rely on textual information, neglecting phonetic cues, which leads to over-correction. To address these issues, we propose a novel LLM-based GER approach that targets rare words and incorporates phonetic information. First, we generate synthetic data to contain rare words for fine-tuning the GER model. Second, we integrate ASR's N-best hypotheses along with phonetic context to mitigate over-correction. Experimental results show that our method not only improves the correction of rare words but also reduces the WER and CER across both English and Japanese datasets.

Large reasoning models (LRMs) have demonstrated impressive performance across a range of reasoning tasks, yet little is known about their internal reasoning processes in multilingual settings. We begin with a critical question: {\it In which language do these models reason when solving problems presented in different languages?} Our findings reveal that, despite multilingual training, LRMs tend to default to reasoning in high-resource languages (e.g., English) at test time, regardless of the input language. When constrained to reason in the same language as the input, model performance declines, especially for low-resource languages. In contrast, reasoning in high-resource languages generally preserves performance. We conduct extensive evaluations across reasoning-intensive tasks (MMMLU, MATH-500) and non-reasoning benchmarks (CulturalBench, LMSYS-toxic), showing that the effect of language choice varies by task type: input-language reasoning degrades performance on reasoning tasks but benefits cultural tasks, while safety evaluations exhibit language-specific behavior. By exposing these linguistic biases in LRMs, our work highlights a critical step toward developing more equitable models that serve users across diverse linguistic backgrounds.

LLMs' decision-making process is opaque, prompting the need for explanation techniques like Chain-of-Thought. To investigate the relationship between answer and reasoning, we design a novel evaluation framework, MATCHA. In domains like education and healthcare, reasoning is key for model trustworthiness. MATCHA reveals that LLMs under input perturbations can give inconsistent or nonsensical reasoning. Additionally, we use LLM judges to assess reasoning robustness across models. Our results show that LLMs exhibit greater vulnerability to input perturbations for multi-step and commonsense tasks than compared to logical tasks. Also, we show non-trivial transfer rates of our successful examples to black-box models. Our evaluation framework helps to better understand LLM reasoning mechanisms and guides future models toward more robust and reasoning-driven architectures, enforcing answer-reasoning consistency.

Retrieval-augmented generation (RAG) grounds large language models (LLMs) in up-to-date external evidence, yet existing multi-hop RAG pipelines still issue redundant subqueries, explore too shallowly, or wander through overly long search chains. We introduce EVO-RAG, a curriculum-guided reinforcement learning framework that evolves a query-rewriting agent from broad early-stage exploration to concise late-stage refinement. EVO-RAG couples a seven-factor, step-level reward vector (covering relevance, redundancy, efficiency, and answer correctness) with a time-varying scheduler that reweights these signals as the episode unfolds. The agent is trained with Direct Preference Optimization over a multi-head reward model, enabling it to learn when to search, backtrack, answer, or refuse. Across four multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA, MuSiQue, and Bamboogle), EVO-RAG boosts Exact Match by up to 4.6 points over strong RAG baselines while trimming average retrieval depth by 15 %. Ablation studies confirm the complementary roles of curriculum staging and dynamic reward scheduling. EVO-RAG thus offers a general recipe for building reliable, cost-effective multi-hop RAG systems.

Large language models (LLMs) like GPT-4 show potential for scaling motivational interviewing (MI) in addiction care, but require systematic evaluation of therapeutic capabilities. We present a computational framework assessing user-perceived quality (UPQ) through expected and unexpected MI behaviors. Analyzing human therapist and GPT-4 MI sessions via human-AI collaboration, we developed predictive models integrating deep learning and explainable AI to identify 17 MI-consistent (MICO) and MI-inconsistent (MIIN) behavioral metrics. A customized chain-of-thought prompt improved GPT-4's MI performance, reducing inappropriate advice while enhancing reflections and empathy. Although GPT-4 remained marginally inferior to therapists overall, it demonstrated superior advice management capabilities. The model achieved measurable quality improvements through prompt engineering, yet showed limitations in addressing complex emotional nuances. This framework establishes a pathway for optimizing LLM-based therapeutic tools through targeted behavioral metric analysis and human-AI co-evaluation. Findings highlight both the scalability potential and current constraints of LLMs in clinical communication applications.

Enterprise customers are increasingly adopting Large Language Models (LLMs) for critical communication tasks, such as drafting emails, crafting sales pitches, and composing casual messages. Deploying such models across different regions requires them to understand diverse cultural and linguistic contexts and generate safe and respectful responses. For enterprise applications, it is crucial to mitigate reputational risks, maintain trust, and ensure compliance by effectively identifying and handling unsafe or offensive language. To address this, we introduce SweEval, a benchmark simulating real-world scenarios with variations in tone (positive or negative) and context (formal or informal). The prompts explicitly instruct the model to include specific swear words while completing the task. This benchmark evaluates whether LLMs comply with or resist such inappropriate instructions and assesses their alignment with ethical frameworks, cultural nuances, and language comprehension capabilities. In order to advance research in building ethically aligned AI systems for enterprise use and beyond, we release the dataset and code: https://github.com/amitbcp/multilingual_profanity.