Multi-modal music generation, using multiple modalities like text, images, and video alongside musical scores and audio as guidance, is an emerging research area with broad applications. This paper reviews this field, categorizing music generation systems from the perspective of modalities. The review covers modality representation, multi-modal data alignment, and their utilization to guide music generation. Current datasets and evaluation methods are also discussed. Key challenges in this area include effective multi-modal integration, large-scale comprehensive datasets, and systematic evaluation methods. Finally, an outlook on future research directions is provided, focusing on creativity, efficiency, multi-modal alignment, and evaluation.

Pre-training large language models (LLMs) necessitates enormous diverse textual corpora, making effective data selection a key challenge for balancing computational resources and model performance. Current methodologies primarily emphasize data quality metrics and mixing proportions, yet they fail to adequately capture the underlying semantic connections between training samples and quality disparities within individual domains. We introduce ToReMi (Topic-based Reweighting for Model improvement), a novel two-stage framework that dynamically adjusts training sample weights according to their topical associations and observed learning patterns. Our comprehensive experiments reveal that ToReMi variants consistently achieve superior performance over conventional pre-training approaches, demonstrating accelerated perplexity reduction across multiple domains and enhanced capabilities on downstream evaluation tasks. Code is available at https://github.com/zxx000728/ToReMi.

In federated learning, fine-tuning pre-trained foundation models poses significant challenges, particularly regarding high communication cost and suboptimal model performance due to data heterogeneity between the clients. To address these issues, this paper introduces communication-efficient federated LoRA adaption (CE-LoRA), a method that employs a tri-factorization low-rank adaptation approach with personalized model parameter aggregation. We first presents a novel LoRA parameter factorization by introducing a small-size dense matrix, which can significantly reduce the communication cost and achieve comparable empirical performance than transferring the low-rank parameter matrix used by existing methods. Without violating data privacy, the server considers the client similarity in both training dataset and model parameter space, and learns personalized weights for model aggregation. Our experiments on various LLM and VLM fine-tuning tasks demonstrate that CE-LoRA not only significantly reduces communication overhead but also improves performance under not independently and identically distributed data conditions. In addition, CE-LoRA improves data privacy protection, effectively mitigating gradient-based data reconstruction attacks.

This paper presents the ZJUKLAB team's submission for SemEval-2025 Task 4: Unlearning Sensitive Content from Large Language Models. This task aims to selectively erase sensitive knowledge from large language models, avoiding both over-forgetting and under-forgetting issues. We propose an unlearning system that leverages Model Merging (specifically TIES-Merging), combining two specialized models into a more balanced unlearned model. Our system achieves competitive results, ranking second among 26 teams, with an online score of 0.944 for Task Aggregate and 0.487 for overall Aggregate. In this paper, we also conduct local experiments and perform a comprehensive analysis of the unlearning process, examining performance trajectories, loss dynamics, and weight perspectives, along with several supplementary experiments, to understand the effectiveness of our method. Furthermore, we analyze the shortcomings of our method and evaluation metrics, emphasizing that MIA scores and ROUGE-based metrics alone are insufficient to fully evaluate successful unlearning. Finally, we emphasize the need for more comprehensive evaluation methodologies and rethinking of unlearning objectives in future research. Code is available at https://github.com/zjunlp/unlearn/tree/main/semeval25.

Finals season looks different this year. Across college campuses, students are slogging their way through exams with all-nighters and lots of caffeine, just as they always have. Through the end of May, OpenAI is offering students two months of free access to ChatGPT Plus, which normally costs 20 a month. It's a compelling deal for students who want help cramming--or cheating--their way through finals: Rather than firing up the free version of ChatGPT to outsource essay writing or work through a practice chemistry exam, students are now able to access the company's most advanced models, as well as its "deep research" tool, which can quickly synthesize hundreds of digital sources into analytical reports. The OpenAI deal is just one of many such AI promotions going around campuses.

According to Bloomberg, AI startup Anthropic is about to release a voice mode for Claude. Currently, it's only possible to communicate with the Claude AI assistant via text, and adding voice mode would bring it up to parity with other advanced AIs like ChatGPT, Gemini, and Sesame. Claude's voice mode will initially only be available in English, with three different voice modes named "Airy," "Mellow," and "Buttery." According to Bloomberg, Claude's voice mode could be released as early as April, but it will initially be rolled out to a limited number of users. Anthropic hasn't yet commented on Claude's voice mode.

But AI is a tool; like any tool, it is only as good as the person wielding it. Now's the time to get the upper hand on AI and learn how to use tools like ChatGPT and automation platforms to work for you. The ChatGPT & Automation E-Degree from Eduonix Learning Solutions gives you the knowledge to stay on top for just 29.99 (MSRP 790) The course includes 12 modules and 25 hours of content you can move through at your own pace, and they never expire. You'll learn how to automate workflows, streamline repetitive tasks, and get AI to handle the boring stuff while you take credit for the results. It also dives into prompt engineering, real-world use cases, and customizing ChatGPT to fit your job, industry, or hustle.

As generative AI models produce increasingly realistic output, both academia and industry are focusing on the ability to detect whether an output was generated by an AI model or not. Many of the research efforts and policy discourse are centered around robust watermarking of AI outputs. While plenty of progress has been made, all watermarking and AI detection techniques face severe limitations. In this position paper, we argue that we are adopting the wrong approach, and should instead focus on watermarking via cryptographic signatures trustworthy content rather than AI generated ones. For audio-visual content, in particular, all real content is grounded in the physical world and captured via hardware sensors. This presents a unique opportunity to watermark at the hardware layer, and we lay out a socio-technical framework and draw parallels with HTTPS certification and Blu-Ray verification protocols. While acknowledging implementation challenges, we contend that hardware-based authentication offers a more tractable path forward, particularly from a policy perspective. As generative models approach perceptual indistinguishability, the research community should be wary of being overly optimistic with AI watermarking, and we argue that AI watermarking research efforts are better spent in the text and LLM space, which are ultimately not traceable to a physical sensor.

Large language models (LLMs) have demonstrated impressive performance across various domains. However, for clinical diagnosis, higher expectations are required for LLM's reliability and sensitivity: thinking like physicians and remaining sensitive to key medical information that affects diagnostic reasoning, as subtle variations can lead to different diagnosis results. Yet, existing works focus mainly on investigating the sensitivity of LLMs to irrelevant context and overlook the importance of key information. In this paper, we investigate the sensitivity of LLMs, i.e. GPT-3.5, GPT-4, Gemini, Claude3 and LLaMA2-7b, to key medical information by introducing different perturbation strategies. The evaluation results highlight the limitations of current LLMs in remaining sensitive to key medical information for diagnostic decision-making. The evolution of LLMs must focus on improving their reliability, enhancing their ability to be sensitive to key information, and effectively utilizing this information. These improvements will enhance human trust in LLMs and facilitate their practical application in real-world scenarios. Our code and dataset are available at https://github.com/chenwei23333/DiagnosisQA.

Data quality and diversity are key to the construction of effective instruction-tuning datasets. % With the increasing availability of open-source instruction-tuning datasets, it is advantageous to automatically select high-quality and diverse subsets from a vast amount of data. % Existing methods typically prioritize instance quality and use heuristic rules to maintain diversity. % However, this absence of a comprehensive view of the entire collection often leads to suboptimal results. % Moreover, heuristic rules generally focus on distance or clustering within the embedding space, which fails to accurately capture the intent of complex instructions in the semantic space. % To bridge this gap, we propose a unified method for quantifying the information content of datasets. This method models the semantic space by constructing a label graph and quantifies diversity based on the distribution of information within the graph. % Based on such a measurement, we further introduce an efficient sampling method that selects data samples iteratively to \textbf{M}aximize the \textbf{I}nformation \textbf{G}ain (MIG) in semantic space. % Experiments on various datasets and base models demonstrate that MIG consistently outperforms state-of-the-art methods. % Notably, the model fine-tuned with 5\% Tulu3 data sampled by MIG achieves comparable performance to the official SFT model trained on the full dataset, with improvements of +5.73\% on AlpacaEval and +6.89\% on Wildbench.