Recently, Meta officially launched an open-source AI virtual try-on framework named Leffa. This framework aims to enhance the user dressing experience by generating new images, allowing users to switch between different clothing and poses based on a reference image. Compared to previous methods, Leffa excels in detail retention and reducing image distortion. The launch of Leffa brings new possibilities for online shopping and virtual try-ons. Users only need to upload a reference image, and the system can generate an entirely new outfit effect based on that image.

Meta recently announced that its Ray-Ban smart glasses will introduce three new features: Real-Time AI, Real-Time Translation, and Shazam. Currently, the Real-Time AI and Real-Time Translation features are available only to members of Meta’s early access program, while the Shazam feature is open to all users in the United States and Canada. The Real-Time AI and Real-Time Translation capabilities were first previewed earlier this year at the Meta Connect 2024 conference. Real-Time AI allows users to interact with their surroundings while the glasses continuously observe the environment.

Wall Street is closely watching Meta Platforms (META) for its aggressive push into artificial intelligence (AI) and its dominant position in digital advertising. The tech giant's early investments in AI infrastructure are seen as a strong engine for long-term growth, with some analysts even setting lofty price targets between $650 and $800 per share. Over the past year, Meta's revenue grew by 23% year-over-year, reaching $156.23 billion. Additionally, its operational efficiency remains impressive, boasting a return on equity of 36%.

Large Language Models (LLMs) have made significant progress in the field of Natural Language Processing (NLP), shining in applications such as text generation, summarization, and question answering. However, the reliance of LLMs on token-level processing (predicting one word at a time) presents some challenges. This method contrasts with human communication, which typically operates at higher levels of abstraction, such as sentences or ideas. Token-level modeling also struggles in tasks that require understanding long contexts and may produce inconsistent outputs.

Meta has recently launched a new neural watermark system called VideoSeal, designed to help identify AI-generated videos, even if the videos have been edited, it can effectively trace their origin. This watermark is invisible to viewers but can be verified through detection methods to confirm the video's authenticity. Meta has released the entire system as open source software under the MIT license, including the VideoSeal watermark model, research papers, and training and inference code.

In recent years, large language models (LLMs) have made significant progress in the field of natural language processing (NLP), widely applicable in scenarios such as text generation, summarization, and question answering. However, these models rely on a token-level processing method that predicts word by word, which struggles with contextual understanding and often leads to inconsistent outputs. Moreover, when scaling LLMs to multilingual and multimodal applications, the computational costs and data requirements tend to be relatively high. To address these issues, Meta AI has proposed a novel approach.

Meta Platforms recently submitted a letter to California Attorney General Rob Bonta, urging him to prevent OpenAI's plan to transition from a non-profit organization to a for-profit company. According to The Wall Street Journal, the letter was sent by Meta on Thursday, stating that allowing OpenAI to make this conversion would set a concerning precedent, potentially enabling startups to enjoy non-profit status while ultimately shifting to a for-profit model. Meta emphasized in the letter.

Today, Deepfake is ubiquitous. With the rise of generative AI, the amount of false content online has exploded. According to identity verification platform Sumsub, the number of Deepfakes globally has increased fourfold from 2023 to 2024. By 2024, Deepfake will account for 7% of all fraud incidents, involving everything from identity impersonation and account theft to complex social engineering attacks. To effectively combat Deepfake, Meta...

With the proliferation of deepfake technology, the amount of false content online has surged. According to data from the ID verification platform Sumsub, the number of deepfakes quadrupled from 2023 to 2024, accounting for 7% of all fraud in 2024. This includes various forms such as identity impersonation, account takeover, and complex social engineering attacks. To tackle this challenge, Meta has recently launched a new watermark tool - Meta Video Seal, aimed at helping with AI-generated content.

Meta is investing $10 billion to build a large AI data center in Richland Parish, Louisiana. This data center will be the largest of Meta's 20 data centers, with construction expected to begin this month and continue until 2030. It is reported that it will be optimized to support Meta's AI workloads, aiding in the operation of Meta's technologies including Facebook, Messenger, Instagram, WhatsApp, and Threads.

Meta recently announced the launch of the latest member of its Llama series - Llama3.370B. Meta's VP of Generative AI, Ahmad Al-Dahle, announced this on the X platform, noting that Llama3.370B has achieved significant improvements in performance compared to Meta's previous largest Llama model, Llama3.1405B, while significantly reducing costs. Al-Dahle stated that by leveraging the latest technology after training, Llama3.370B...