Prior works on detoxification are scattered in the sense that they do not cover all aspects of detoxification needed in a real-world scenario. Notably, prior works restrict the task of developing detoxification models to only a seen subset of platforms, leaving the question of how the models would perform on unseen platforms unexplored. Additionally, these works do not address non-detoxifiability, a phenomenon whereby the toxic text cannot be detoxified without altering the meaning. We propose GreenLLaMA, the first comprehensive end-to-end detoxification framework, which attempts to alleviate the aforementioned limitations. We first introduce a cross-platform pseudo-parallel corpus applying multi-step data processing and generation strategies leveraging ChatGPT. We then train a suite of detoxification models with our cross-platform corpus. We show that our detoxification models outperform the SoTA model trained with human-annotated parallel corpus. We further introduce explanation to promote transparency and trustworthiness. GreenLLaMA additionally offers a unique paraphrase detector especially dedicated for the detoxification task to tackle the non-detoxifiable cases. Through experimental analysis, we demonstrate the effectiveness of our cross-platform corpus and the robustness of GreenLLaMA against adversarial toxicity.

Dense retrievers compress source documents into (possibly lossy) vector representations, yet there is little analysis of what information is lost versus preserved, and how it affects downstream tasks. We conduct the first analysis of the information captured by dense retrievers compared to the language models they are based on (e.g., BERT versus Contriever). We use 25 MultiBert checkpoints as randomized initialisations to train MultiContrievers, a set of 25 contriever models. We test whether specific pieces of information -- such as gender and occupation -- can be extracted from contriever vectors of wikipedia-like documents. We measure this extractability via information theoretic probing. We then examine the relationship of extractability to performance and gender bias, as well as the sensitivity of these results to many random initialisations and data shuffles. We find that (1) contriever models have significantly increased extractability, but extractability usually correlates poorly with benchmark performance 2) gender bias is present, but is not caused by the contriever representations 3) there is high sensitivity to both random initialisation and to data shuffle, suggesting that future retrieval research should test across a wider spread of both.

Advances in large language models (LLMs) provide new opportunities in healthcare for improved patient care, clinical decision-making, and enhancement of physician and administrator workflows. However, the potential of these models importantly depends on their ability to generalize effectively across clinical environments and populations, a challenge often underestimated in early development. To better understand reasons for these challenges and inform mitigation approaches, we evaluated ClinicLLM, an LLM trained on [HOSPITAL]'s clinical notes, analyzing its performance on 30-day all-cause readmission prediction focusing on variability across hospitals and patient characteristics. We found poorer generalization particularly in hospitals with fewer samples, among patients with government and unspecified insurance, the elderly, and those with high comorbidities. To understand reasons for lack of generalization, we investigated sample sizes for fine-tuning, note content (number of words per note), patient characteristics (comorbidity level, age, insurance type, borough), and health system aspects (hospital, all-cause 30-day readmission, and mortality rates). We used descriptive statistics and supervised classification to identify features. We found that, along with sample size, patient age, number of comorbidities, and the number of words in notes are all important factors related to generalization. Finally, we compared local fine-tuning (hospital specific), instance-based augmented fine-tuning and cluster-based fine-tuning for improving generalization. Among these, local fine-tuning proved most effective, increasing AUC by 0.25% to 11.74% (most helpful in settings with limited data). Overall, this study provides new insights for enhancing the deployment of large language models in the societally important domain of healthcare, and improving their performance for broader populations.

With the rising popularity of the internet and the widespread use of networks and information systems via the cloud and data centers, the privacy and security of individuals and organizations have become extremely crucial. In this perspective, encryption consolidates effective technologies that can effectively fulfill these requirements by protecting public information exchanges. To achieve these aims, the researchers used a wide assortment of encryption algorithms to accommodate the varied requirements of this field, as well as focusing on complex mathematical issues during their work to substantially complicate the encrypted communication mechanism. as much as possible to preserve personal information while significantly reducing the possibility of attacks. Depending on how complex and distinct the requirements established by these various applications are, the potential of trying to break them continues to occur, and systems for evaluating and verifying the cryptographic algorithms implemented continue to be necessary. The best approach to analyzing an encryption algorithm is to identify a practical and efficient technique to break it or to learn ways to detect and repair weak aspects in algorithms, which is known as cryptanalysis. Experts in cryptanalysis have discovered several methods for breaking the cipher, such as discovering a critical vulnerability in mathematical equations to derive the secret key or determining the plaintext from the ciphertext. There are various attacks against secure cryptographic algorithms in the literature, and the strategies and mathematical solutions widely employed empower cryptanalysts to demonstrate their findings, identify weaknesses, and diagnose maintenance failures in algorithms.

Recent years have witnessed the promise of coupling machine learning methods and physical domain-specific insight for solving scientific problems based on partial differential equations (PDEs). However, being data-intensive, these methods still require a large amount of PDE data. This reintroduces the need for expensive numerical PDE solutions, partially undermining the original goal of avoiding these expensive simulations. In this work, seeking data efficiency, we design unsupervised pretraining and in-context learning methods for PDE operator learning. To reduce the need for training data with simulated solutions, we pretrain neural operators on unlabeled PDE data using reconstruction-based proxy tasks. To improve out-of-distribution performance, we further assist neural operators in flexibly leveraging in-context learning methods, without incurring extra training costs or designs. Extensive empirical evaluations on a diverse set of PDEs demonstrate that our method is highly data-efficient, more generalizable, and even outperforms conventional vision-pretrained models.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. BOGOTA, Colombia (AP) -- Colombia's government on Friday announced plans for a deep-water expedition to explore the mythical galleon San José, sunk in the 18th century in the country's northern Caribbean and believed to contain cargo valued at billions of dollars. It is the first phase of a scientific research into deep waters that aims at collecting information to determine which pieces are suitable and possible to extract. The wreckage is 600 meters deep in the sea.

Radio host Tommy Sotomayor reacts to artificial intelligence images rewriting history, on'Jesse Watters Primetime.' Sen. Tom Cotton, R-Ark., slammed Google's AI chatbot Gemini as "preposterously woke" on Friday for its refusal to produce any images of White people. The company paused the chatbot's image generation on Thursday after social media users pointed out that the system was creating inaccurate historical images that sometimes replaced White people, like the Founding Fathers, with images of Black, Native American and Asian people. "Google deserves condemnation for creating a racist, preposterously woke, Hamas-sympathizing AI system," Cotton said in a statement on X, formerly Twitter. "Republican lawmakers will remember this the next time Google comes asking for antitrust help."

Privacy experts are sounding the alarm on a new satellite capable of spying on your every move that is set to launch in 2025. The satellite, created by startup company Albedo, is so high quality it can zoom in on people or license plates from space, raising concerns among expert that it will create a'big brother is always watching' scenario. Albedo claims the satellite won't have facial recognition software but doesn't mention that it will refrain from imaging people or protecting people's privacy. Albedo signed two separate million-dollar contracts with the U.S. Air Force and the National Air and Space Intelligence Center to help the government monitor potential threats to U.S. national security. Albedo claims the satellite won't have facial recognition software but doesn't mention that it will refrain from imaging people or protecting people's privacy.

In the arid tech sphere of semiconductor manufacturing, one hardback book-sized processor stands out: Nvidia's H-100. On Friday, the Santa Clara, California, company surpassed 2tn in valuation. Where it goes next will be down to a chip named after "Amazing Grace" Hopper, a US navy rear admiral who became instrumental in the development of design and implementation of programming languages. Nvidia supplies approximately 80% of the global market in chips used in AI applications. The company's H-100 chips – the H is for Hopper – are now so valuable they have to be transported by armored car, the Wall Street Journal reported on Friday, and demand is so great that some customers are waiting as long as six months to receive it.

According to a report from Lloyd's Register Foundation,a at present, cybercrime is one of the biggest concerns of Internet users worldwide, with disinformationb ranking highest among such risks (57% of Internet users across all parts of the world, socioeconomic groups, and all ages). For years, there has been a discussion in the security community about whether disinformation should be considered a cyber threat.10 However, recent worldwide phenomena (for example, an increase in the frequency and sophistication of cyberattacks, the 2016 U.S. election interference, the Russian invasion in Ukraine, the COVID-19 pandemic, and so forth) have made disinformation one of the most potent cybersecurity threats for businesses, governments, the media, and society as a whole. In addition, recent breakthroughs in AI have further enabled the creation of highly realistic fake content at scale. As such, we argue that disinformation should be rightfully considered a cyber threat, and therefore developing effective countermeasures is critically necessary.