MIT spinoff Liquid debuts non-transformer AI models and they’re already state-of-the-art
The startup from MIT's CSAIL says its Liquid Foundation Models have smaller memory needs thanks to a post-transformer ...
ABC Columbia3 小时
Meta unveils cheaper VR headset, AI updates and shows off prototype for holographic AR glasses
Meta unveiled updates to the company's virtual reality headset and Ray Ban smart glasses on Wednesday as it tries to ...
Mark Zuckerberg Reveals Daughter August, 6, Wrote Her Own Book — and They Used AI to ...
Mark Zuckerberg reveals his daughter August, 6, wrote her own book and used AI to illustrate it. The Meta CEO shares his three daughters — August, Max and Aurelia — with wife Dr. Priscilla Chan.
CMS Wire11 小时
Meta Llama 3.2’s Vision Reasoning: Why Marketers Should Pay Attention
Meta’s Llama 3.2 adds vision reasoning. Use it to enhance your AI-driven CX with smarter image and text analysis.
Easiest Way to Fine Tune Llama 3.2 and Run it Locally
Meta's LLAMA-3.2 models represent a significant advancement in the field of language modeling, offering a range of sizes from ...
Meta Connect 大会 Wi-Fi 密码暗示，扎克伯格仍迷恋于“效率”
北京时间 9 月 30 日，2023 年，Meta CEO 马克・扎克伯格 (Mark Zuckerberg) 为应对公司困境将该年度定义为“效率之年”。他裁减了数千个工作岗位，消除了管理层级，并采取了其他激进措施为 Meta 节省资金。
Meta Platforms, Inc. (META) Partners with Arista Networks for $2 Billion AI Cluster Using ...
We recently compiled a list of the 20 AI News That Broke The Internet This Month. In this article, we are going to take a ...
tech-ish.com17 小时
Meta Awards $500K to Kenyan AI Projects: Digital Green and Jacaranda Health Lead the Way
Meta awards $500K to Kenyan AI projects, supporting innovative solutions in agriculture and healthcare through its Llama Impact Grants.
Manila Bulletin1 天
Meta's latest breakthroughs
At Meta Connect 2024, Meta unveiled a range of groundbreaking innovations in virtual reality (VR), augmented reality (AR), ...
研究称AI 越聪明越有可能“胡编乱造”
据媒体援引一项新研究表示，随着大型语言模型（LLM）变得越来越强大，它们似乎也越来越容易编造事实，而不是避免或拒绝回答它们无法回答的问题。这表明，这些更聪明的 AI 聊天机器人实际上变得不太可靠。 该研究发表在《自然》杂志上，研究人员研究了一些业界领先的商业 LLM：OpenAI 的 GPT 和 Meta 的 LLaMA，以及由研究小组 BigScience 创建的开源模型 ...
凤凰网1 天
研究发现：AI 越聪明就越有可能“胡编乱造”
IT之家 9 月 29 日消息，一项新研究发现，随着大型语言模型（LLM）变得越来越强大，它们似乎也越来越容易编造事实，而不是避免或拒绝回答它们无法回答的问题。这表明，这些更聪明的 AI 聊天机器人实际上变得不太可靠。
格隆汇 on MSN1 天
研究：AI大型语言模型越聪明就越有可能编造事实
格隆汇9月29日｜据IT之家，一项新研究发现，随着大型语言模型(LLM)变得越来越强大，它们似乎也越来越容易编造事实，而不是避免或拒绝回答它们无法回答的问题。该研究发表在《自然》杂志上，研究人员研究了一些业界领先的商业LLM：OpenAI的GPT和M ...