Open Source LLM

Liquid AI發表非Transformer的AI模型

去年才成立的Liquid AI週一(9/30) 發表 了3個Liquid基礎模型(Liquid Foundation Models,LFMs),分別是LFM-1.3B、LFM-3.1B 及LFM-40.3B。Liquid AI利用非Transformer架構來建置模型,並宣稱旗下模型在基準測試上,已凌駕那些規模相同的Transformer模型


Open Source LLM

AMD unveils its first small language model, AMD-135M — AI performance enhanced by speculative decoding

(Image credit: AMD)

As AMD flexes its muscles in the AI game, it is not only introducing new hardware but is betting on software too, trying to hit new market segments not already dominated by Nvidia.

Thus, AMD has unveiled its first small language model, AMD-135M, which belongs to the Llama family and is aimed at private business deployments. It is unclear whether the new model has to do anything with the company's recent acquisition of Silo AI (as the deal has

Continue reading...

Open Source LLM

AMD Releases AMD-135M: An Open-Source Small Language Model

AMD today announced "AMD-135M" as their first small language model they are publicly releasing. AMD-135M is open-source with the training code, dataset, and weights all being open-source to help in the development of other SLMs and LLMs.

AMD-135M features speculative decoding and was trained from scratch using AMD Instinct MI250 accelerators with 670 billion tokens. Training using four MI250 nodes took six days. There is
Continue reading...

Open Source LLM|Intel

Intel Boasts Strong Lunar Lake AI Performance With Meta LLaMA 3.2

The Intel Lunar Lake CPU platform is demonstrating impressive performance in AI Inference with the latest Meta LLaMA 3.2 models. Intel Lunar Lake "Core Ultra 9 288V" CPU shows latency of just 28.5ms for 32 and 31.4ms for 1024 input tokens Meta has just launched its LLaMA 3.2, now providing more models for AI tasks. The original LLaMA (Large Language Model Meta AI) was released last year in February, which seeks to fulfill

Open Source LLM

Google的AI筆記助理NotebookLM現可輸入YouTube影片及聲音檔案

Google本週 更新了AI筆記助理NotebookLM ,新版支援YouTube影片及聲音檔案的輸入,還能與他人共享語音摘要(Audio Overviews)。

NotebookLM是Google在 去年7月發表 的產品,這是一個由大型語言模型助攻的筆記軟體,根據Google的説法,它與傳統AI聊天機器人最大的差別


Open Source LLM|Intel

(PR) AMD Instinct MI300X Accelerators Available on Oracle Cloud Infrastructure

AMD today announced that Oracle Cloud Infrastructure (OCI) has chosen AMD Instinct MI300X accelerators with ROCm open software to power its newest OCI Compute Supercluster instance called BM.GPU.MI300X.8. For AI models that can comprise hundreds of billions of parameters, the OCI Supercluster with AMD MI300X supports up to 16,384 GPUs in a single cluster by harnessing the same ultrafast network fabric technology used by other accelerators on OCI. Designed to run demanding

Open Source LLM

Exclusive: OpenAI to remove non-profit control and give Sam Altman equity, sources say - Reuters

Exclusive: OpenAI to remove non-profit control and give Sam Altman equity, sources say Reuters

Open Source LLM

NVIDIA’s Blackwell To Power OpenAI’s Cutting-Edge o1 LLM Model, Credits To The 50x Uplift In Inferencing Capabilities

NVIDIA's Blackwell platform is set to power OpenAI's newest o1 LLM model, as Jensen reveals massive uplifts in the architecture's inferencing capabilities. NVIDIA's Blackwell Architecture Is Going To Play a Massive Role In Reshaping The AI Markets, Sees Tremendous Demand We are seemingly away from the generative AI hype now, and it looks like the markets are heading toward developing LLM models that can generate "reasoning-focused" responses, which is one of the most crucial steps toward achieving artificial general intelligence. Recently

Open Source LLM

Intel's LPMD "Low Power Mode Daemon" Now Identifies As The "Energy Optimizer"

The Intel LPMD open-source project is a user-space daemon for optimizing active idle power handling on Linux and can be useful particularly for modern Intel Core hybrid processors. LPMD is short for the "Low Power Mode Daemon" while with today's v0.0.7 release it's now re-identified itself as the "Energy Optimizer" instead.

The Low Power Mode Daemon name is quite straight-forward and especially with the "LPMD" mentions in the binary names and throughout the project.
Continue reading...

Open Source LLM

OpenAI o1預覽版推向企業、教育用户,一週上限50則訊息

OpenAI 宣佈 最新能複雜思考的AI模型OpenAI o1、o1-mini,將推向企業與教育用户,OpenAI o1預覽版供用户一週最多輸入50則訊息。

上週 公佈 預覽版的OpenAI o1家族是一組有進階理解力的模型,比之前模型能理解更複雜的任務,解決更困