Wall Street went into panic mode about two weeks ago after the Chinese startup DeepSeek released an artificial intelligence system that appeared to be radically more efficient than what its American competitors had built. The investors who had pumped trillions of dollars into tech stocks over the past few years worried whether the tens of billions of dollars that tech companies were spending on new data centers suddenly looked like comic overkill. But the biggest tech companies made clear in recent earnings reports that they believe there may be no such thing as overkill when it comes to new data centers.

Multimodal datasets are a critical component in recent breakthroughs such as CLIP, Stable Diffusion and GPT-4, yet their design does not receive the same research attention as model architectures or training algorithms. To address this shortcoming in the machine learning ecosystem, we introduce DataComp, a testbed for dataset experiments centered around a new candidate pool of 12.8 billion image-text pairs from Common Crawl. Participants in our benchmark design new filtering techniques or curate new data sources and then evaluate their new dataset by running our standardized CLIP training code and testing the resulting model on 38 downstream test sets. Our benchmark consists of multiple compute scales spanning four orders of magnitude, which enables the study of scaling trends and makes the benchmark accessible to researchers with varying resources. Our baseline experiments show that the DataComp workflow leads to better training sets.

We've seen numerous new features added to ChatGPT in recent months, including updated models, web search capabilities, and the ability to remember what you say to it--and the latest software upgrade added to the AI bot by OpenAI makes it more useful as a general-purpose digital assistant. Beginning in beta form, and available initially to paying subscribers--the feature will reach everyone eventually, OpenAI says--ChatGPT Tasks lets you request the AI chatbot perform actions regularly on an automated schedule, or remind you about something in the future. Here's everything you need to know about it. "In this early beta, you can create scheduled tasks that enable ChatGPT to run automated prompts and proactively reach out to you on a scheduled basis," explains OpenAI. Tasks are available on the web, in the mobile apps, and in the macOS desktop app; OpenAI says the feature will make it to the Windows desktop app soon.

At President Donald Trump's inauguration, Sam Altman, CEO of OpenAI, was relegated to the overflow room while other tech billionaires like Elon Musk and Mark Zuckerberg took prime spots on the dais under the Capitol rotunda. But days earlier, before flying to Washington, Altman was on the phone with Trump, preparing an announcement that would outflank Musk and put Altman's company at the center of the new administration's agenda for artificial intelligence. On the 25-minute call, Altman appealed to Trump's love of a big story and of a big deal. He told the president-elect that the tech industry would achieve artificial general intelligence -- the hypothetical moment when technology matches human intelligence -- during the Trump administration, according to three people familiar with the call. And to get there before Chinese competitors, OpenAI, Oracle and SoftBank had completed a 100 billion deal to build data centers across the country.

Refusal-Aware Instruction Tuning (RAIT) aims to enhance Large Language Models (LLMs) by improving their ability to refuse responses to questions beyond their knowledge, thereby reducing hallucinations and improving reliability. Effective RAIT must address two key challenges: firstly, effectively reject unknown questions to minimize hallucinations; secondly, avoid over-refusal to ensure questions that can be correctly answered are not rejected, thereby maintain the helpfulness of LLM outputs. In this paper, we address the two challenges by deriving insightful observations from the gradient-based perspective, and proposing the Gradient-driven Refusal Aware Instruction Tuning Framework GRAIT: (1) employs gradient-driven sample selection to effectively minimize hallucinations and (2) introduces an adaptive weighting mechanism during fine-tuning to reduce the risk of over-refusal, achieving the balance between accurate refusals and maintaining useful responses. Experimental evaluations on open-ended and multiple-choice question answering tasks demonstrate that GRAIT significantly outperforms existing RAIT methods in the overall performance. The source code and data will be available at https://github.com/opendatalab/GRAIT .

Hyperspectral imaging (HSI) captures spatial and spectral data, enabling analysis of features invisible to conventional systems. The technology is vital in fields such as weather monitoring, food quality control, counterfeit detection, healthcare diagnostics, and extending into defense, agriculture, and industrial automation at the same time. HSI has advanced with improvements in spectral resolution, miniaturization, and computational methods. This study provides an overview of the HSI, its applications, challenges in data fusion and the role of deep learning models in processing HSI data. We discuss how integration of multimodal HSI with AI, particularly with deep learning, improves classification accuracy and operational efficiency. Deep learning enhances HSI analysis in areas like feature extraction, change detection, denoising unmixing, dimensionality reduction, landcover mapping, data augmentation, spectral construction and super resolution. An emerging focus is the fusion of hyperspectral cameras with large language models (LLMs), referred as highbrain LLMs, enabling the development of advanced applications such as low visibility crash detection and face antispoofing. We also highlight key players in HSI industry, its compound annual growth rate and the growing industrial significance. The purpose is to offer insight to both technical and non-technical audience, covering HSI's images, trends, and future directions, while providing valuable information on HSI datasets and software libraries.

Mixture of Experts (MoE), with its distinctive sparse structure, enables the scaling of language models up to trillions of parameters without significantly increasing computational costs. However, the substantial parameter size presents a challenge for inference, as the expansion in GPU memory cannot keep pace with the growth in parameters. Although offloading techniques utilise memory from the CPU and disk and parallelise the I/O and computation for efficiency, the computation for each expert in MoE models is often less than the I/O, resulting in numerous bubbles in the pipeline. Therefore, we propose Klotski, an efficient MoE inference engine that significantly reduces pipeline bubbles through a novel expert-aware multi-batch pipeline paradigm. The proposed paradigm uses batch processing to extend the computation time of the current layer to overlap with the loading time of the next layer. Although this idea has been effectively applied to dense models, more batches may activate more experts in the MoE, leading to longer loading times and more bubbles. Thus, unlike traditional approaches, we balance computation and I/O time and minimise bubbles by orchestrating their inference orders based on their heterogeneous computation and I/O requirements and activation patterns under different batch numbers. Moreover, to adapt to different hardware environments and models, we design a constraint-sensitive I/O-compute planner and a correlation-aware expert prefetcher for a schedule that minimises pipeline bubbles. Experimental results demonstrate that Klotski achieves a superior throughput-latency trade-off compared to state-of-the-art techniques, with throughput improvements of up to 85.12x.

Semantic role labeling (SRL) is a central natural language processing (NLP) task aiming to understand the semantic roles within texts, facilitating a wide range of downstream applications. While SRL has garnered extensive and enduring research, there is currently a lack of a comprehensive survey that thoroughly organizes and synthesizes the field. This paper aims to review the entire research trajectory of the SRL community over the past two decades. We begin by providing a complete definition of SRL. To offer a comprehensive taxonomy, we categorize SRL methodologies into four key perspectives: model architectures, syntax feature modeling, application scenarios, and multi-modal extensions. Further, we discuss SRL benchmarks, evaluation metrics, and paradigm modeling approaches, while also exploring practical applications across various domains. Finally, we analyze future research directions in SRL, addressing the evolving role of SRL in the age of large language models (LLMs) and its potential impact on the broader NLP landscape. We maintain a public repository and consistently update related resources at: https://github.com/DreamH1gh/Awesome-SRL

Recent advancements in artificial intelligence have enabled processing of larger inputs, leading everyday software developers to increasingly rely on chat-based large language models (LLMs) like GPT-3.5 and GPT-4 to detect vulnerabilities across entire files, not just within functions. This new development practice requires researchers to urgently investigate whether commonly used LLMs can effectively analyze large file-sized inputs, in order to provide timely insights for software developers and engineers about the pros and cons of this emerging technological trend. Hence, the goal of this paper is to evaluate the effectiveness of several state-of-the-art chat-based LLMs, including the GPT models, in detecting in-file vulnerabilities. We conducted a costly investigation into how the performance of LLMs varies based on vulnerability type, input size, and vulnerability location within the file. To give enough statistical power to our study, we could only focus on the three most common (as well as dangerous) vulnerabilities: XSS, SQL injection, and path traversal. Our findings indicate that the effectiveness of LLMs in detecting these vulnerabilities is strongly influenced by both the location of the vulnerability and the overall size of the input. Specifically, regardless of the vulnerability type, LLMs tend to significantly (p < .05) underperform when detecting vulnerabilities located toward the end of larger files, a pattern we call the 'lost-in-the-end' effect. Finally, to further support software developers and practitioners, we also explored the optimal input size for these LLMs and presented a simple strategy for identifying it, which can be applied to other models and vulnerability types. Eventually, we show how adjusting the input size can lead to significant improvements in LLM-based vulnerability detection, with an average recall increase of over 37% across all models.

Historically, LLMs have been trained using either autoregressive (AR) or masked language modeling (MLM) objectives, with AR models gaining dominance in recent years. However, AR models are inherently incapable of masked infilling, which is the ability to predict masked tokens between past and future context. In contrast, MLM models suffer from intrinsic computational inefficiencies during both training and inference that hinder their scalability. This work introduces MARIA (Masked and Autoregressive Infilling Architecture), a novel approach that leverages the strengths of both paradigms to achieve state-of-the-art masked infilling performance. MARIA combines a pre-trained MLM and AR model by training a linear decoder that takes their concatenated hidden states as input. This minimal modification enables the AR model to perform infilling while retaining its inherent advantages in terms of faster inference with KV caching. Our results demonstrate that MARIA significantly outperforms existing methods, namely discrete diffusion models, on masked infilling tasks.