Mixture of expert (MoE) models are a promising approach to increasing model capacity without increasing inference cost, and are core components of many state-of-the-art language models. However, current MoE models typically use only few experts due to prohibitive training and inference cost. We propose Test-Time Model Merging (TTMM) which scales the MoE paradigm to an order of magnitude more experts and uses model merging to avoid almost any test-time overhead. We show that TTMM is an approximation of test-time training (TTT), which fine-tunes an expert model for each prediction task, i.e., prompt. TTT has recently been shown to significantly improve language models, but is computationally expensive. We find that performance of TTMM improves with more experts and approaches the performance of TTT. Moreover, we find that with a 1B parameter base model, TTMM is more than 100x faster than TTT at test-time by amortizing the cost of TTT at train-time. Thus, TTMM offers a promising cost-effective approach to scale test-time training.

Multimodal Retrieval Augmented Generation (MMRAG) is a powerful approach to questionanswering over multimodal documents. A key challenge with evaluating MMRAG is the paucity of high-quality datasets matching the question styles and modalities of interest. In light of this, we propose SMMQG, a synthetic data generation framework. SMMQG leverages interplay between a retriever, large language model (LLM) and large multimodal model (LMM) to generate question and answer pairs directly from multimodal documents, with the questions conforming to specified styles and modalities. We use SMMQG to generate an MMRAG dataset of 1024 questions Figure 1: An overview of SMMQG. Given userprovided over Wikipedia documents and evaluate stateof-the-art question style and modality requirements, SMmodels using it, revealing insights MQG selects question sources and produces questions into model performance that are attainable only and answers. The questions are grounded in the selected through style-and modality-specific evaluation question sources, and adhere to the question and modality data. Next, we measure the quality of data produced requirements.

Large language models (LLMs) have achieved remarkable success due to their exceptional generative capabilities. Despite their success, they also have inherent limitations such as a lack of up-to-date knowledge and hallucination. Retrieval-Augmented Generation (RAG) is a state-of-the-art technique to mitigate those limitations. In particular, given a question, RAG retrieves relevant knowledge from a knowledge database to augment the input of the LLM. For instance, the retrieved knowledge could be a set of top-k texts that are most semantically similar to the given question when the knowledge database contains millions of texts collected from Wikipedia. As a result, the LLM could utilize the retrieved knowledge as the context to generate an answer for the given question. Existing studies mainly focus on improving the accuracy or efficiency of RAG, leaving its security largely unexplored. We aim to bridge the gap in this work. Particularly, we propose PoisonedRAG , a set of knowledge poisoning attacks to RAG, where an attacker could inject a few poisoned texts into the knowledge database such that the LLM generates an attacker-chosen target answer for an attacker-chosen target question. We formulate knowledge poisoning attacks as an optimization problem, whose solution is a set of poisoned texts. Depending on the background knowledge (e.g., black-box and white-box settings) of an attacker on the RAG, we propose two solutions to solve the optimization problem, respectively. Our results on multiple benchmark datasets and LLMs show our attacks could achieve 90% attack success rates when injecting 5 poisoned texts for each target question into a database with millions of texts. We also evaluate recent defenses and our results show they are insufficient to defend against our attacks, highlighting the need for new defenses.

"Your face is your ticket," goes the motto of A.I. startup Wicket. "Your face is your credential," says Alcatraz AI, another vendor. Both these companies sell facial recognition technology to sports stadiums across the country. Citi Field, home of the Mets, contracted with Wicket in 2022 to add facial recognition ticket kiosks to all stadium gates. BMO Stadium, home of the Los Angeles Football Club, began using Alcatraz AI technology the year before.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. The Miami Dolphins have been revamping their offense under first-year head coach Mike McDaniel. On Tuesday, they made two deals that have many believing this could be a Super Bowl contending team. Tua Tagovailoa is one of the believers.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. WWE is ready for 2022. The pro wrestling company released its pay-per-view schedule for the next year with two more shows left on the docket for the year, Survivor Series in November and TLC: Tables Ladders & Chairs in December. MIAMI GARDENS, FL - APRIL 1: John Cena looks on before his match against Dwayne ''The Rock'' Johnson during WrestleMania XXVIII at Sun Life Stadium on April 1, 2012 in Miami Gardens, Florida.

A drone footage of the Niger ambush that killed four U.S. and five Nigerian soldiers that surfaced recently shows the service personnel desperately trying to escape and fighting for their lives after friendly Nigerien forces mistook them for the enemy. The video shows the harrowing hours of troops holding off their enemy and waiting for rescue. It shows how the soldiers set up a defensive location on the edge of a marsh and wrote letters to their loved ones thinking they were going to die. Pentagon released the video with explanatory narration and it contains more than 10 minutes of drone footage, animation and file tape that was not made public last week when the military released a portion of the final report on the October attack, the Guardian reported. In a failed attempt to target a local ISIS leader, 46 U.S. and Nigerien troops were involved in the initial mission in the West African nation.

WASHINGTON – Dramatic new drone video of the Niger ambush that killed four American soldiers shows U.S. forces desperately trying to escape and fighting for their lives after friendly Nigerien forces mistook them for the enemy. It describes how the fleeing troops set up a quick defensive location on the edge of a swamp and -- thinking they were soon to die -- wrote messages home to their loved ones. The video, released by the Pentagon with explanatory narration, includes more than 10 minutes of drone footage, file tape and animation that wasn't made public last week when the military released a portion of the final report on the October attack. The video depicts for the first time the harrowing hours as troops held off their enemy and waited for rescue. There were 46 U.S. and Nigerien troops out on the initial mission in the west African nation, going after but failing to find a high-value militant, then collecting intelligence at a site where the insurgent had been.

WASHINGTON – Dramatic new drone video of the Niger ambush that killed four American soldiers shows U.S. forces desperately trying to escape and fighting for their lives after friendly Nigerien forces mistook them for the enemy. It describes how the fleeing troops set up a quick defensive location on the edge of a swamp and -- thinking they were soon to die -- wrote messages home to their loved ones. The video, released by the Pentagon with explanatory narration, includes more than 10 minutes of drone footage, file tape and animation that wasn't made public last week when the military released a portion of the final report on the October attack. The video depicts for the first time the harrowing hours as troops held off their enemy and waited for rescue. There were 46 U.S. and Nigerien troops out on the initial mission in the West African nation, going after but failing to find a high-value militant, then collecting intelligence at a site where the insurgent had been.