Tab. 4 provides detailed statistics of the two datasets. B.2 Clinical Predictive T asks We focus on two common clinical predictive tasks: readmission prediction and mortality prediction. In the case of the eICU dataset, the predictions are made 12 hours after admission. The overall prevalence for these tasks is 15% for readmission and 4% for mortality. For the MIMIC-IV dataset, the predictions are made at the time of discharge.

A language arts teacher has shared the bizarre conspiracy theories her sixth grade students believe in and what fostered that beliefs. The teacher, who goes by the name Ms Alexanderr, said was amazed by her students' ideas and wanted to compile a list of the top five most she felt were the most bizarre. While the teacher said she wasn't surprised by one conspiracy theory that birds aren't real, she was shocked and couldn't understand others. Among them was the theory that Bill Nye the science guy is a Russian spy while another claimed Michael Jackson was still alive. The pop-star conspiracy was particularly perplexing, because her students were born after he died in 2009.

The current public datasets for speech recognition (ASR) tend not to focus specifically on the fairness aspect, such as performance across different demographic groups. This paper introduces a novel dataset, Fair-Speech, a publicly released corpus to help researchers evaluate their ASR models for accuracy across a diverse set of self-reported demographic information, such as age, gender, ethnicity, geographic variation and whether the participants consider themselves native English speakers. Our dataset includes approximately 26.5K utterances in recorded speech by 593 people in the United States, who were paid to record and submit audios of themselves saying voice commands. We also provide ASR baselines, including on models trained on transcribed and untranscribed social media videos and open source models.

Large Language Models (LLMs) have demonstrated great potential as generalist assistants, showcasing powerful task understanding and problem-solving capabilities. To deploy LLMs as AI assistants, it is crucial that these models exhibit desirable behavioral traits, such as non-toxicity and resilience against jailbreak attempts. Current methods for detoxification or preventing jailbreaking usually involve Supervised Fine-Tuning (SFT) or Reinforcement Learning from Human Feedback (RLHF), which requires finetuning billions of parameters through gradient descent with substantial computation cost. Furthermore, models modified through SFT and RLHF may deviate from the pretrained models, potentially leading to a degradation in foundational LLM capabilities. In this paper, we observe that surprisingly, directly editing a small subset of parameters can effectively modulate specific behaviors of LLMs, such as detoxification and resistance to jailbreaking. Specifically, for a behavior that we aim to avoid, we employ a linear classifier, which we term the behavior probe, to classify binary behavior labels within the hidden state space of the LLM. Using this probe, we introduce an algorithm to identify a critical subset of LLM parameters that significantly influence this targeted behavior. Then we directly edit these selected parameters by shifting them towards the behavior probe. Such a direct parameter editing method necessitates only inference-level computational resources. Experiments demonstrate that in the representative detoxification task, our approach achieves reductions of up to 90.0\% in toxicity on the RealToxicityPrompts dataset and 49.2\% on ToxiGen, while maintaining the LLM's general capabilities in areas such as common sense, question answering, and mathematics. Our code is available at https://github.com/lucywang720/model-surgery.

Google's Gemini chatbot, which was formerly called Bard, has the capability to whip up AI-generated illustrations based on a user's text description. You can ask it to create pictures of happy couples, for instance, or people in period clothing walking modern streets. As the BBC notes, however, some users are criticizing Google for depicting specific white figures or historically white groups of people as racially diverse individuals. Now, Google has issued a statement, saying that it's aware Gemini "is offering inaccuracies in some historical image generation depictions" and that it's going to fix things immediately. We're aware that Gemini is offering inaccuracies in some historical image generation depictions.

For more than 15 years, Wikipedia discussed what to call the third child of Ernest Hemingway, a doctor who was born and wrote books as Gregory, later lived as Gloria after undergoing gender-affirming surgery, and, when arrested for public disorderliness late in life, used a third name, Vanessa. Last year, editors on the site finally settled the question: The Gregory Hemingway article was deleted, and its contents were moved to a new one for Gloria Hemingway. This would be her name going forward, and she/her would be her pronouns. Wikipedia's billions of facts, rendered as dry prose in millions of articles, help us understand the world. They are largely the brain behind Siri and Alexa.

In 2020, the Wikipedia community was engulfed in scandal when it came out that a US teen had written 27,000 entries in a language they didn't speak. The episode was a reminder that the online encyclopedia is not a perfect source of information. Sometimes people will attempt to edit Wikipedia entries out of malice, but frequently factual errors come from some well-intentioned individual making a mistake. That's a problem the Wikimedia Foundation recently partnered with Facebook parent company Meta to address. The two set their sights on citations.

In the summer of 1956, 10 scientists met at Dartmouth College and invented artificial intelligence. Researchers from fields like mathematics, engineering, psychology, economics, and political science got together to find out whether they could describe learning and human thinking so precisely that it could be replicated with a machine. Hardly a decade later, these same scientists contributed to dramatic breakthroughs in robotics, natural language processing, and computer vision. Although a lot of time has passed since then, robotics, natural language processing, and computer vision remain some of the hottest research areas to this day. One could say that we're focused on teaching AI to move like a human, speak like a human and see like a human.

Graph-to-text generation has benefited from pre-trained language models (PLMs) in achieving better performance than structured graph encoders. However, they fail to fully utilize the structure information of the input graph. In this paper, we aim to further improve the performance of the pre-trained language model by proposing a structured graph-to-text model with a two-step fine-tuning mechanism which first fine-tunes the model on Wikipedia before adapting to the graph-to-text generation. In addition to using the traditional token and position embeddings to encode the knowledge graph (KG), we propose a novel tree-level embedding method to capture the inter-dependency structures of the input graph. This new approach has significantly improved the performance of all text generation metrics for the English WebNLG 2017 dataset.