Meta is reportedly in talks to acquire South Korean chip startup FuriosaAI, with an announcement possible as soon as this month. FuriosaAI builds chips that speed up model serving, including text models like Llama 3. The move highlights Meta's push to cut reliance on Nvidia as its AI budget climbs to $65 billion this year.
Mark Zuckerberg used YouTube's fight against piracy to justify Meta's use of copyrighted datasets during a deposition in an ongoing AI copyright lawsuit. Newly unsealed court filings allege Meta trained its Llama models on pirated e-book repositories like LibGen and Z-Library despite internal legal warnings.
Sakana AI researchers just developed CycleQD, an evolutionary framework that merges specialized small models into multi-skilled AI swarms without expensive fine-tuning. By using model merging and SVD for crossover and mutation, it outpaced traditional fine-tuning on Llama 3-8B tests, offering a cheaper path to balanced AI agents.
Mark Zuckerberg just bet Meta's future on open-source AI, turning the Llama model series into the backbone of its apps and smart glasses. After the metaverse slump, downloads topped 600 million, though investors are still waiting to see how billions in computing spend will pay off.
Meta just revealed the winners of its inaugural Llama Impact Innovation Awards, awarding organizations across Africa, the Middle East, Asia Pacific, and Latin America up to $35,000 in funding and matching ad credits. The recognized projects leverage Llama models for local challenges, including bilingual medical diagnosis, vaccine education, and low-cost vision aids.
MultiOn just introduced Agent Q, a research breakthrough bringing planning and autonomous self-healing capabilities to AI agents. It delivers a 340% performance improvement over LLaMA 3's zero-shot baseline, designed to keep agents on track even when workflows break down.
Meta just confirmed it won't launch its upcoming multimodal Llama model—or future multimodal versions—in the EU, blaming an unpredictable regulatory environment under GDPR. European firms won't get access even under an open license, though a larger text-only Llama 3 model will still release there.
Meta just rolled out Meta AI in India in English across WhatsApp, Facebook, Messenger, and Instagram. Built on Llama 3, the free assistant lets users brainstorm in group chats, pull up details directly from Facebook feeds, and generate or animate images by typing "imagine".
Elon Musk and Meta's Yann LeCun just clashed in an unfiltered spat on X over scientific publishing and AI hype. After LeCun slammed Musk's apocalyptic claims and wild deadlines, Musk questioned LeCun's recent research credentials. Why it matters: The clash exposes the widening rift between open academic science and secretive, hype-driven commercial labs.
Researchers at EPFL and Yale School of Medicine just released Llama-3[8B]-MeditronV1.0, fine-tuned on Meta's Llama 3 in 24 hours to help clinicians in low-resource and humanitarian settings. The open-source model beats all open models in its size on MedQA and MedMCQA, with weights and data openly available to ensure equitable access.
Meta just launched Llama 3, delivering 8B and 70B open-access models trained on over 15 trillion tokens. The smaller 8B variant can easily run and fine-tune on consumer GPUs down to 4-bit precision, backed by an expanded 128k vocabulary. Why it matters: Meta is resetting the benchmark for open AI, giving developers frontier-level capabilities without proprietary API lock-in.
NVIDIA just rolled out full-stack optimizations for Meta's brand-new Llama 3 across its data center GPUs, edge devices, and PCs. In early benchmarks, a single H200 GPU churned out roughly 3,000 tokens per second on the 70B model, enough to handle 300 users simultaneously. The 8B model is also landing on RTX PCs and Jetson Orin modules.
Meta just upgraded its Meta AI assistant with Llama 3 and launched a standalone web experience at meta.ai. The assistant is woven directly into the search bars of WhatsApp, Instagram, Facebook, and Messenger, rolling out across more than a dozen English-speaking countries with real-time text-to-image generation as you type.
Meta just confirmed that its next-generation open-source model, Llama 3, will begin rolling out within the next month. The company plans to release multiple versions across the year and integrate them throughout its product lineup. Llama 3 aims to relax previous over-cautious restrictions, with reports expecting parameter sizes to reach roughly 140B.
Mark Zuckerberg just confirmed that Meta is actively training Llama 3, aiming for state-of-the-art coding and reasoning capabilities to pursue AGI. To power the push, Meta is stockpiling over 340,000 Nvidia H100 GPUs by the end of 2024, building a computing fortress that rivals almost any competitor in the industry.
Meta is reportedly gearing up to train Llama 3 and Llama 4, with plans to bring Llama 3 up to GPT-4's performance while keeping it freely accessible under its open license. The detail surfaced from OpenAI researcher Jason Wei at a Meta event. If true, Meta is doubling down on commoditizing frontier models to rival closed labs.
