Recent advancements in large language models (LLMs) demonstrate exceptional Chinese text processing capabilities, particularly in Chinese Spelling Correction (CSC). While LLMs outperform traditional BERT-based models in accuracy and robustness, challenges persist in reliability and generalization. This paper proposes CEC-Zero, a novel reinforcement learning (RL) framework enabling LLMs to self-correct through autonomous error strategy learning without external supervision. By integrating RL with LLMs' generative power, the method eliminates dependency on annotated data or auxiliary models. Experiments reveal RL-enhanced LLMs achieve industry-viable accuracy and superior cross-domain generalization, offering a scalable solution for reliability optimization in Chinese NLP applications. This breakthrough facilitates LLM deployment in practical Chinese text correction scenarios while establishing a new paradigm for self-improving language models.

Large Language Models (LLMs) represent a major step toward artificial general intelligence, significantly advancing our ability to interact with technology. While LLMs perform well on Natural Language Processing tasks -- such as translation, generation, code writing, and summarization -- questions remain about their output similarity, variability, and ethical implications. For instance, how similar are texts generated by the same model? How does this compare across different models? And which models best uphold ethical standards? To investigate, we used 5{,}000 prompts spanning diverse tasks like generation, explanation, and rewriting. This resulted in approximately 3 million texts from 12 LLMs, including proprietary and open-source systems from OpenAI, Google, Microsoft, Meta, and Mistral. Key findings include: (1) outputs from the same LLM are more similar to each other than to human-written texts; (2) models like WizardLM-2-8x22b generate highly similar outputs, while GPT-4 produces more varied responses; (3) LLM writing styles differ significantly, with Llama 3 and Mistral showing higher similarity, and GPT-4 standing out for distinctiveness; (4) differences in vocabulary and tone underscore the linguistic uniqueness of LLM-generated content; (5) some LLMs demonstrate greater gender balance and reduced bias. These results offer new insights into the behavior and diversity of LLM outputs, helping guide future development and ethical evaluation.

Predicting the future citation rates of academic papers is an important step toward the automation of research evaluation and the acceleration of scientific progress. We present $\textbf{ForeCite}$, a simple but powerful framework to append pre-trained causal language models with a linear head for average monthly citation rate prediction. Adapting transformers for regression tasks, ForeCite achieves a test correlation of $ρ= 0.826$ on a curated dataset of 900K+ biomedical papers published between 2000 and 2024, a 27-point improvement over the previous state-of-the-art. Comprehensive scaling-law analysis reveals consistent gains across model sizes and data volumes, while temporal holdout experiments confirm practical robustness. Gradient-based saliency heatmaps suggest a potentially undue reliance on titles and abstract texts. These results establish a new state-of-the-art in forecasting the long-term influence of academic research and lay the groundwork for the automated, high-fidelity evaluation of scientific contributions.

The emergence of telomere-to-telomere (T2T) genome assemblies has opened new avenues for comparative genomics, yet effective tokenization strategies for genomic sequences remain underexplored. In this pilot study, we apply Byte-Pair Encoding (BPE) to nine T2T primate genomes--including three human assemblies--by training independent BPE tokenizers with a fixed vocabulary of 512,000 tokens using our custom tool, dnaBPE. Our analysis reveals that only 11,569 tokens are shared across all assemblies, while nearly 991,854 tokens are unique to a single genome, indicating a rapid decline in shared vocabulary with increasing assembly comparisons. Moreover, phylogenetic trees derived from token overlap failed to recapitulate established primate relationships, a discrepancy attributed to the disproportionate influence of species-specific high-copy repetitive elements. These findings underscore the dual nature of BPE tokenization: while it effectively compresses repetitive sequences, its sensitivity to high-copy elements limits its utility as a universal tool for comparative genomics. We discuss potential hybrid strategies and repeat-masking approaches to refine genomic tokenization, emphasizing the need for domain-specific adaptations in the development of large-scale genomic language models. The dnaBPE tool used in this study is open-source and available at https://github.com/aglabx/dnaBPE .

As the digital landscape becomes more interconnected, the frequency and severity of zero-day attacks, have significantly increased, leading to an urgent need for innovative Intrusion Detection Systems (IDS). Machine Learning-based IDS that learn from the network traffic characteristics and can discern attack patterns from benign traffic offer an advanced solution to traditional signature-based IDS. However, they heavily rely on labeled datasets, and their ability to generalize when encountering unseen traffic patterns remains a challenge. This paper proposes a novel self-supervised contrastive learning approach based on transformer encoders, specifically tailored for generalizable intrusion detection on raw packet sequences. Our proposed learning scheme employs a packet-level data augmentation strategy combined with a transformer-based architecture to extract and generate meaningful representations of traffic flows. Unlike traditional methods reliant on handcrafted statistical features (NetFlow), our approach automatically learns comprehensive packet sequence representations, significantly enhancing performance in anomaly identification tasks and supervised learning for intrusion detection. Our transformer-based framework exhibits better performance in comparison to existing NetFlow self-supervised methods. Specifically, we achieve up to a 3% higher AUC in anomaly detection for intra-dataset evaluation and up to 20% higher AUC scores in inter-dataset evaluation. Moreover, our model provides a strong baseline for supervised intrusion detection with limited labeled data, exhibiting an improvement over self-supervised NetFlow models of up to 1.5% AUC when pretrained and evaluated on the same dataset. Additionally, we show the adaptability of our pretrained model when fine-tuned across different datasets, demonstrating strong performance even when lacking benign data from the target domain.

The application of AI in oncology has been limited by its reliance on large, annotated datasets and the need for retraining models for domain-specific diagnostic tasks. Taking heed of these limitations, we investigated in-context learning as a pragmatic alternative to model retraining by allowing models to adapt to new diagnostic tasks using only a few labeled examples at inference, without the need for retraining. Using four vision-language models (VLMs)-Paligemma, CLIP, ALIGN and GPT-4o, we evaluated the performance across three oncology datasets: MHIST, PatchCamelyon and HAM10000. To the best of our knowledge, this is the first study to compare the performance of multiple VLMs on different oncology classification tasks. Without any parameter updates, all models showed significant gains with few-shot prompting, with GPT-4o reaching an F1 score of 0.81 in binary classification and 0.60 in multi-class classification settings. While these results remain below the ceiling of fully fine-tuned systems, they highlight the potential of ICL to approximate task-specific behavior using only a handful of examples, reflecting how clinicians often reason from prior cases. Notably, open-source models like Paligemma and CLIP demonstrated competitive gains despite their smaller size, suggesting feasibility for deployment in computing constrained clinical environments. Overall, these findings highlight the potential of ICL as a practical solution in oncology, particularly for rare cancers and resource-limited contexts where fine-tuning is infeasible and annotated data is difficult to obtain.

As large language models (LLMs) improve in their capacity to serve as personal AI assistants, their ability to output uniquely tailored, personalized responses that align with the soft preferences of their users is essential for enhancing user satisfaction and retention. However, untrained lay users have poor prompt specification abilities and often struggle with conveying their latent preferences to AI assistants. To address this, we leverage activation steering to guide LLMs to align with interpretable preference dimensions during inference. In contrast to memory-based personalization methods that require longer user history, steering is extremely lightweight and can be easily controlled by the user via an linear strength factor. We embed steering into three different interactive chatbot interfaces and conduct a within-subjects user study (n=14) to investigate how end users prefer to personalize their conversations. The results demonstrate the effectiveness of preference-based steering for aligning real-world conversations with hidden user preferences, and highlight further insights on how diverse values around control, usability, and transparency lead users to prefer different interfaces.

Despite decades of research and practice in automated software testing, several fundamental concepts remain ill-defined and under-explored, yet offer enormous potential real-world impact. We show that these concepts raise exciting new challenges in the context of Large Language Models for software test generation. More specifically, we formally define and investigate the properties of hardening and catching tests. A hardening test is one that seeks to protect against future regressions, while a catching test is one that catches such a regression or a fault in new functionality introduced by a code change. Hardening tests can be generated at any time and may become catching tests when a future regression is caught. We also define and motivate the Catching 'Just-in-Time' (JiTTest) Challenge, in which tests are generated 'just-in-time' to catch new faults before they land into production. We show that any solution to Catching JiTTest generation can also be repurposed to catch latent faults in legacy code. We enumerate possible outcomes for hardening and catching tests and JiTTests, and discuss open research problems, deployment options, and initial results from our work on automated LLM-based hardening at Meta. This paper was written to accompany the keynote by the authors at the ACM International Conference on the Foundations of Software Engineering (FSE) 2025. Author order is alphabetical. The corresponding author is Mark Harman.

In the rapidly evolving field of data science, efficiently navigating the expansive body of academic literature is crucial for informed decision-making and innovation. This paper presents an enhanced Retrieval-Augmented Generation (RAG) application, an artificial intelligence (AI)-based system designed to assist data scientists in accessing precise and contextually relevant academic resources. The AI-powered application integrates advanced techniques, including the GeneRation Of BIbliographic Data (GROBID) technique for extracting bibliographic information, fine-tuned embedding models, semantic chunking, and an abstract-first retrieval method, to significantly improve the relevance and accuracy of the retrieved information. This implementation of AI specifically addresses the challenge of academic literature navigation. A comprehensive evaluation using the Retrieval-Augmented Generation Assessment System (RAGAS) framework demonstrates substantial improvements in key metrics, particularly Context Relevance, underscoring the system's effectiveness in reducing information overload and enhancing decision-making processes. Our findings highlight the potential of this enhanced Retrieval-Augmented Generation system to transform academic exploration within data science, ultimately advancing the workflow of research and innovation in the field.

Large Language Models (LLMs) are increasingly embedded in autonomous systems and public-facing environments, yet they remain susceptible to jailbreak vulnerabilities that may undermine their security and trustworthiness. Adversarial suffixes are considered to be the current state-of-the-art jailbreak, consistently outperforming simpler methods and frequently succeeding even in black-box settings. Existing defenses rely on access to the internal architecture of models limiting diverse deployment, increase memory and computation footprints dramatically, or can be bypassed with simple prompt engineering methods. We introduce $\textbf{Adversarial Suffix Filtering}$ (ASF), a lightweight novel model-agnostic defensive pipeline designed to protect LLMs against adversarial suffix attacks. ASF functions as an input preprocessor and sanitizer that detects and filters adversarially crafted suffixes in prompts, effectively neutralizing malicious injections. We demonstrate that ASF provides comprehensive defense capabilities across both black-box and white-box attack settings, reducing the attack efficacy of state-of-the-art adversarial suffix generation methods to below 4%, while only minimally affecting the target model's capabilities in non-adversarial scenarios.