RSS Reader

The Linux Foundation along with industry stakeholders like Intel, Red Hat, Hugging Face, MariaDB, Cloudera, and others have launched the Open Platform for Enterprise AI (OPEA) as a new enterprise AI collaborative effort.

In today's announcement of the Open Platform for Enterprise AI it's explained as:

"This pioneering initiative unites industry leaders, including Anyscale, Cloudera, Datastax, Domino Data Lab, Hugging Face, Intel, KX, MariaDB Foundation, Minio, Qdrant, Red Hat, SAS, VMware (acquired

Open Source LLM

OpenAI triples Japanese language processing speed

Nikkei Asia | April 16, 2024, 3 p.m. | Original

TOKYO -- U.S. artificial intelligence company OpenAI is preparing an improved version of its core large language model with triple the Japanese language processing power, aiming to compete with similar services being worked on by companies in Japan.

"Japan is an important market" for OpenAI, Chief Operating Officer Brad Lightcap told reporters Monday as the company opened its Japanese subsidiary in Tokyo. Lightcap also demonstrated a custom large language model optimized for Japanese language at a news conference.

Open Source LLM

Khronos Releases OpenXR 1.1 For Cross-Platform AR/VR Development

Phoronix | April 16, 2024, 5:55 a.m. | Original

The Khronos Group on Monday released OpenXR 1.1 as the latest version of this industry standard for open, cross-platform development for virtual reality (VR), augmented reality (AR), and mixed reality (MR) devices.

OpenXR 1.1 promotes a number of extensions into the core specification to help with reducing fragmentation and simplifying development. Now part of OpenXR 1.1 core are local floor, stereo with foveated rendering, grip surface, XrUuid, and xrLocateSpaces.

OpenXR 1.1 also introduces a

Open Source LLM

國科會釋出繁中語言模型TAIDE-LX-7B了！

ITHome | April 15, 2024, 3:04 a.m. | Original

2022年末ChatGPT爆紅，掀起大型語言模型（LLM）應用浪潮，為能夠在地化運用，國科會去年初就開始推動可信任生成式AI發展先期計畫（Trustworthy AI Dialogue Engine，簡稱TAIDE），來打造本土化LLM。經過1年多努力，國科會主委吳政忠今日（4/1

Open Source LLM

AI趨勢周報第250期：Google一篇論文揭示LLM如何處理無限長文字輸入

ITHome | April 14, 2024, 12:04 a.m. | Original

重點新聞(0405～0411)

LLM Google 文長

Google一篇論文揭示LLM如何處理無限長文字輸入

Google最近揭露一種Transformer大型語言模型（LLM）的新擴展方法，可利用有限的記憶體和運算資源，來處理無限長的文字輸入。Google在《Leave No Context Behind》論文中

Open Source LLM

LPython 0.21 Released For Alpha-Stage Python AOT Compiler

Phoronix | April 12, 2024, 8:59 a.m. | Original

As the first new release since last September, LPython 0.21 has been released for this alpha-stage Python ahead-of-time compiler written in C++. LPython remains focused on providing "the best possible performance" especially for numerical use-cases, cross-platform compatibility, and hopes to be able to eventually transform Python code over to C++ and Fortran or other languages.

LPython in its early stage is only compiling a subset of Python code but continues picking up more features. LPython

Open Source LLM

Mistral AI釋出開源模型Mixtral 8x22B

ITHome | April 11, 2024, 2 a.m. | Original

去年4月才於法國成立的AI業者Mistral AI週三（4/10）釋出了新的開源模型Mixtral 8x22B，它採用稀疏混合專家（Sparse Mixture of Experts，SMoE）架構，支援1,760億個參數，以及6.5萬個Token的脈絡長度，已藉由Mistral AI的官方X帳號、 Together API

Open Source LLM

Meta證實5月將釋出Llama 3首個版本

ITHome | April 11, 2024, 12:01 a.m. | Original

面臨四面八方競爭者推出新式AI模型，Meta昨（9）日證實，將在下個月開源釋出下世代大型語言模型Llama 3。

Meta是在《The Information》本週稍早報導後證實此事。這家英國新聞網站引述不具名Meta員工消息報導，Meta下週將推出Llama 3二款較小型

Open Source LLM

Lutris 0.5.17 Game Manager Brings Bug Fixes, Library Syncing & New Runners

Phoronix | April 10, 2024, 8:21 p.m. | Original

A new release of Lutris is now available, the open-source game manager that's popular with Linux gamers and enthusiasts for managing games from Steam, GOG, a number of retro game consoles and emulators, and other sources from one convenient UI.

Lutris 0.5.17 introduces new runners for Vita3k (a PS Vita emulator) and Supermodel (a Sega Model 3 Arcade Emulator). Version 0.5.17 also brings experimental support for UMU to allow running games with Proton and

Open Source LLM