M4 の MacBook Air（メモリ 16GB）でのローカルLLM、Ollama ⇒ MLX-VLM と試してきて、次は MLX LM

#Gemma3n （mlx-community/gemma-3n-E4B-it-lm-4bit）を試した様子を動画にしてみた

出力スピードも良い感じ！

#Gemma

今まで M4 の Mac mini（メモリ 24GB）で試してきたローカルLLM を、M4 の MacBook Air（メモリ 16GB）でも試してみる

まずは #Ollama で #Gemma3n を使う形で動作確認

#Gemma

20250704 Todey's Image

Summer is in full swing in Japan, and ice cream tastes great during the hot days!
*Prompt generated by LM Studio + Gemma 3n

#Studying_AI #AIimage #AIphoto #AIart
#AIgravure #AIpinup #AIcovergirl #Amuse3.0
#LM_Studio #Gemma3n

🧠 #Gemma3n di Google in azione su task multimodali in locale sul mio smartphone.
✨ Negli esempi si vede come il modello interpreta il contenuto delle immagini, estraendo le informazioni richieste.

#AI #GenAI #GenerativeAI #IntelligenzaArtificiale #LLM

🧠 Ho provato #Gemma3n in locale, su un laptop attraverso #Ollama e anche su #Colab.
👉 Com'è andata? www.linkedin.com/posts/alessi...
___
✉️ 𝗦𝗲 𝘃𝘂𝗼𝗶 𝗿𝗶𝗺𝗮𝗻𝗲𝗿𝗲 𝗮𝗴𝗴𝗶𝗼𝗿𝗻𝗮𝘁𝗼/𝗮 𝘀𝘂 𝗾𝘂𝗲𝘀𝘁𝗲 𝘁𝗲𝗺𝗮𝘁𝗶𝗰𝗵𝗲, 𝗶𝘀𝗰𝗿𝗶𝘃𝗶𝘁𝗶 𝗮𝗹𝗹𝗮 𝗺𝗶𝗮 𝗻𝗲𝘄𝘀𝗹𝗲𝘁𝘁𝗲𝗿: bit.ly/newsletter-a...

#AI #GenerativeAI #IntelligenzaArtificiale #LLM

“gemma3n” is like finding out your quiet neighbor is a secret superhero. Overpowered, ridiculously efficient, and surprisingly from Google. Guess they had it in them all along. 😏 #TechSurprises #gemma3n

Introducing Gemma 3n: The developer guide- Google Developers Blog Learn how to build with Gemma 3n, a mobile-first architecture, MatFormer technology, Per-Layer Embeddings, and new audio and vision encoders.

developers.googleblog.com/en/introduci...

I have installed and used LM Studio.
I asked for specific examples of effective prompts, but nsfw was offended.

I just modified the normal prompt and
laughed when I got a pretty egregious composition.

#Studying_AI #google #Gemma3n #LM_Studio

📱 Google has unveiled Gemma 3n, a compact open AI model built for phones, tablets, and laptops! 🌍✨ With native multimodal input, a tiny memory footprint (just 2–3GB RAM), and broad support for developer tools, it's designed to bring on-device intelligence to the masses. 🧠💡
#AI #GoogleAI #Gemma3n

Google’s Gemma 3n AI can run locally on smartphones with 2GB RAM Google has released Gemma 3n, a powerful open-source AI model that can run locally on just 2GB RAM, making it ideal for smartphones and edge devices.

Google’s Gemma 3n runs with as little as 2GB RAM using a unique nested transformer design, making it ideal for mobile applications. #Google #Gemma3n

gadgetbond.com/google-gemma...

ITちゃんねる わずか2GBのメモリ使用量でスマホでの実行もOKな軽量オープンウェイトモデル「Gemma 3n」をGoogleが正式リリース、音声や動画の入力にも対応し日本語ももちろんOK #Gemma3n #ITニュース

わずか2GBのメモリ使用量でスマホでの実行もOKな軽量オープンウェイトモデル「Gemma 3n」をGoogleが正式リリース、音声や動画の入力にも対応し日本語ももちろんOK
#Gemma3n #ITニュース

Google Releases Gemma 3n: A Multimodal AI to Power a New Wave of On-Device AI Apps

#AI #Google #Gemma3n #Gemma3 #OpenSource #OnDeviceAI #EdgeAI #MachineLearning #Developer #MultimodalAI #LLM

winbuzzer.com/2025/06/27/g...

Google's Gemma 3n: Powerful AI on your device, just 2GB RAM! 📱

Runs large models efficiently & handles text, image, audio, & video.

Works offline too! What would you use this lightweight on-device AI for first? #Gemma3n #AI

Sign in - Google Accounts

Now chat with Gemma 3n in Google AI Studio
⚡ Faster. Smarter. Private.
Built for researchers, devs & AI explorers.

→ Optimized for fine-tuning & prompting
→ Works on-device & in the cloud
→ Small, open, cutting-edge

Try it now → aistudio.google.com
#Gemma3n #GoogleAI #AIStudio #OpenModels

Google AI Edge GalleryでオフラインAI！使い方と利点 | AI News Googleが発表したAI Edge Gallery。AndroidデバイスでオフラインAIモデルを実行可能に！プライバシーと低遅延を実現する開発者向けツール。

AIクリエーターの道 ニュース：GoogleのAI Edge Galleryが登場！データプライバシーを守りながらオフラインAIを実現。開発者必見！ #AIEdgeGallery #オフラインAI #Gemma3n

詳しくはこちら↓↓↓
gamefi.co.jp/2025/06/03/g...

🚀 Google introduces Gemma 3n, and it wants to live (and see and hear) on your phone! 🤖 I've written an analysis of this new on-device AI and what it means for developers. The end of total API dependency for AI? 🤔 Full post on: https://www.thedavestack.com/gemma-3n/

#Gemma3n #AI #Google

人とくるまのテクノロジー展2025*: 人とくるまのテクノロジー展2025が開催され、最新のテクノロジーが展示されました。ジェイテクトの歯車一体軸受「JIGB」やアルプスアルパインブースなどが注目を集めました。
#テクノロジー #AI #金融テック #人とくるまのテクノロジー展 #Google #Gemma3n

[Shoper Gamer] กูเกิลเปิดตัว "Gemma 3n" โมเดล AI ขนาดกะทัดรัด โดย โดย

กูเกิลเปิดตัว "Gemma 3n" โมเดล AI ขนาดกะทัดรัด

#ShoperGamer #Google #Ai #ModelAi #Gemma3n #Feed

Overview: HN chat on Google's Gemma 3n, a "mobile-first AI". Covers performance on devices, tech like Per-Layer Embeddings, potential uses, and accuracy vs other models. #Gemma3n 1/6

ファイナルフロンティア - IT関連ニュース 【 #ITニュース 】Google、効率的なモバイル向けAIモデル「Gemma 3n」プレビュー発表 #Gemma3n #CodeZine

【 #ITニュース 】Google、効率的なモバイル向けAIモデル「Gemma 3n」プレビュー発表
#Gemma3n #CodeZine

ITちゃんねる Google、効率的なモバイル向けAIモデル「Gemma 3n」プレビュー発表 #Gemma3n #ITニュース

Google、効率的なモバイル向けAIモデル「Gemma 3n」プレビュー発表
#Gemma3n #ITニュース

Google Gemma 3n: Powerful On-Device AI for Mobile & Edge Devices | HM Journal Explore Google's Gemma 3n, an efficient, mobile-first AI model designed for phones, laptops, and edge devices. Learn about its multimodal capabilities, low RAM usage, and impact on on-device intelligence.

Google has announced Gemma 3n, its latest AI model focused on efficient performance for mobile devices. This next-gen foundation aims to enable real-time, multimodal AI experiences directly on your phone. #Gemma3n #Google #ArtificialIntelligence
hmjournal.com/en/post/the-...

Gemma 3n: Noua Generație de Inteligență Artificială Mobilă - TECHNEWSRO % Gemma 3n: Noua Generație de Inteligență Artificială Mobilă TECHNEWSRO AI News

#Gemma3n: The Next Generation of Mobile #ArtificialIntelligence

Google unveils Gemma 3n which runs locally on your devices with less memory At Google I/O, the search giant has unveiled its latest small LLM, Gemma 3n. This model is designed to run on your personal devices and uses less memory.

At Google I/O, the search giant has unveiled its latest small LLM, Gemma 3n. This model is designed to run on your personal devices and uses less memory. #Google #Gemma3n #GoogleIO

#Gemma3n: Mobile-first #AI model with multimodal support and on-device privacy 🤖

#Google announces #Gemma3n preview - a mobile-optimized #AI model engineered for on-device performance. Key innovations include:

🧵👇 #ondeviceAI