Paul Parkinson

New video on building a Q&A chatbot 💬 https://nvda.ws/3XaWchY ✅ 𝐄𝐦𝐛𝐞𝐝𝐝𝐢𝐧𝐠 𝐂𝐫𝐞𝐚𝐭𝐢𝐨𝐧: Use NVIDIA NIM microservices to convert text into top-quality embeddings. ✅ 𝐕𝐞𝐜𝐭𝐨𝐫 𝐃𝐚𝐭𝐚𝐛𝐚𝐬𝐞: Store and retrieve embeddings efficiently with GPU-accelerated Milvus. ✅ 𝐈𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐰𝐢𝐭𝐡 𝐀𝐈 𝐚𝐭 𝐌𝐞𝐭𝐚 𝐋𝐥𝐚𝐦𝐚 3: Utilize the NIM API Llama 3 model for accurate responses to user queries. ✅ 𝐎𝐫𝐜𝐡𝐞𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐰𝐢𝐭𝐡 𝐋𝐥𝐚𝐦𝐚𝐈𝐧𝐝𝐞𝐱: Seamlessly integrate and manage all components for a smooth Q&A experience.

3

Raspberry Pi 5 is getting a first party AI accelerator add on, Hailo 8L accelerator which can do 13 trillion operations at int8 , enough to run quantized LLM's fast, processing videos in real time What would you build if you get access to it?

33

2 Comments

I knew the week was too quiet. So these are not outrageous specs to reach this level of Graph performance, but also not inexpensive. Software: NetworkX 3.4.1, cuGraph/nx-cugraph 24.10 CPU: Intel Xeon Gold 6128 CPU @ 3.40GHz 45GB RAM (that is a really weird RAM count) GPU: NVIDIA Quadro RTX 8000 50GB RAM You can pick up a refurbished HP Z4 G5 workstation from NewEgg for about $9,000...it comes with a normal amount of RAM (64GB). https://lnkd.in/gZ_WyVsN

2

2 Comments

Procedural Ruins Tutorial Julian Bragagna | In this lesson, learn how to create these alien-looking ruins. Start by learning how to build a procedural tool that transforms a rough blockout into a detailed, high-poly mesh that can be used inside Unreal. These kinds of tools are designed to give you fast iteration times, making it easier to experiment and refine your environments efficiently. https://lnkd.in/ge8PpuGp #Houdini #ProceduralGeneration #GameDevelopment #3DModeling #DigitalArt #VFX #CGI #EnvironmentArt #LevelDesign #GameArt #ProceduralTools #3DAnimation #VisualEffects #GameDesign #Tutorial #SideFX #ProceduralRuins #UnrealArtists #HoudiniArtists

25

AI Democratization at scale ! Watched the CES NVIDIA keynote. Here's the breakdown of the announcements: 🤖 1️⃣ Project DIGITS & GB10 Superchip A 200B-parameter AI supercomputer in your palm. This isn't science fiction - it's happening now. The implications for edge AI are staggering. 2️⃣ Autonomous Mobility Revolution Cosmos World Foundation Models + three-computer solution = billions of virtual testing miles. Aurora and Continental are already deploying NVIDIA DRIVE-powered driverless trucks at scale. 3️⃣ DRIVE Thor Built on the Blackwell architecture, setting new benchmarks for autonomous vehicle safety and security. 4️⃣ Next-Gen Video AI Blueprint Chain-of-thought reasoning and dynamic task planning are transforming video search and summarization. The possibilities for content creators are endless. 5️⃣ 'Mega' Omniverse for Manufacturing Digital twin technology simulating entire robot fleets. This could revolutionize the $50T manufacturing and logistics industries. 6️⃣ Nemotron AI Models NVIDIA NIM microservices bringing enterprise-grade AI to any accelerated system through Llama Nemotron language models and Cosmos Nemotron vision models. 7️⃣ Omniverse Sensor RTX Early access announced! Foretellix's implementation shows how accurate sensor simulation is transforming autonomous system development. 8️⃣ RTX AI PC Evolution NIM microservices and AI Blueprints democratizing access to generative AI. Digital humans and content creation workflows are becoming plug-and-play. 🎯 The Showstopper: RTX 5070 promising RTX 4090 performance at just $549. High-end AI capabilities are becoming more accessible than ever. The convergence of more powerful, more accessible AI computing with sophisticated simulation capabilities suggests we're entering a new era of technological innovation. More competition on the way but what possibilities do we see with digital twin technology? #AI #Data #Innovation #gaming #CES2025 #NVIDIA

14

It is lifecycle of any thing which came to existence on earth evolves, as only thing is constant is change. ofcourse as day by day XR Application and Experiences has started contributing to multiple streamlines and seems to be much better solution than any other possible alternatives to that experience, it is just that people have starting adapting to it, so without a doubt we can say that we are going to see another revolutionary boom very soon in the industry. #ar #vr #xr #mr #virtualreality #future #analysis #revolution

2

1 Comment

A Preliminary Review of Cursor + claude-3.5-sonnet Cursor is a fork of the open source Microsoft VSCode editor. The fork enables Anysphere developers to better integrate their AI features than is possible with a plugin or module. The Cursor Pro subscription includes use of claude-3.5-sonnet and gpt-4o amongst other LLMs. I've used the OpenAI models for the last year, so when I started using sonnet I had a solid base for comparison and have never really been tempted to use the OpenAI models again. Sonnet is better at code generation. Part of being better means Sonnet is also worse. Yea! It's faster which gives the sense of it being more aggressive which in turn quickly leads it deep into unlit alleys. Without a base of best practices these spelunking sprees will repeat and they are catastrophic. Think of this as myself, who has never ridden horseys, suddenly being cast as a jockey in the Kentucky Derby. It's not going to end well. Documentation is essential for controlling Sonnet or any other model for maintaining context and getting the beasts to go where you need. Twice I've deleted all the Sonnet generated code. Despite each of the failed attempts being based in documentation. What's different with the third attempt is repeatedly referencing the documentation which includes: ### Development Discipline 1. Core Principles   - Build only what's in the spec   - No "while we're at it" features   - If it's not documented, it's not approved   - Keep implementations simple   - Resist the urge to optimize prematurely 2. Feature Request Process ... 3. Phase 1 Focus ... 4. Development Checklist ... Instructing the model to "Review the document and pay particular attention to Development Discipline and the Authentication development stage," produces the following response. LLM: I've reviewed the system design document. Here are my key observations focusing on the Development Discipline and Authentication sections: Development Discipline is well structured with clear principles Strong emphasis on building only what's specified Clear "no feature creep" policy Documentation-driven development Focus on simplicity over premature optimization Authentication section (under Core Implementation) appears somewhat basic and could benefit from more detail. Cursor with a sternly coerced Sonnet is fun because it's fast and productive. It leaves little time for dipping into Derbys past and imagining myself mishandling Secretariat.

1

Quick test with Ultraleap 2 + Unreal Engine 5 + Touchdesigner via OSC. https://lnkd.in/entshVEs #xr #touchdesigner #ultraleap2 #unrealengine5 #realtime #dynamicvisuals #handtracking #gestures #test #immersiveprototype #lumen #immersivemedia

8

Is there a "one-click" RAG application for C2E similar to NVIDIA's Chat with RTX (https://lnkd.in/ejrehxZ3)? I'm searching for a simple Retrieval-Augmented Generation (RAG) application that can run in the Commercial Cloud Enterprise (C2E) environment. It would allow non-programmers to easily create and use RAG applications and fine-tune Large Language Models (LLMs). Here's what I'm envisioning: • User-friendly interface for selecting documents and an LLM • Ability to create and manage multiple RAG applications • Built-in Role-Based Access Control (RBAC) • Option to fine-tune LLMs with the same documents • Dashboard showing all created RAG applications, and LLM fine-tunes (with RBAC enforced) My primary goal is to find a tool to democratize RAG and fine-tune LLM capabilities within a secure C2E environment. This tool should be designed with the typical user in mind, making it accessible even to those without programming skills. Has anyone come across or developed something like this? If not, is there a need for such a tool in the intelligence community? #RAG #C2E #AI #MachineLearning #NLP #CloudComputing #IntelligenceCommunity #LLM #FineTuning #RBAC #GovTech

18

29 Comments

Anthropic’s newest and now most intelligent model, Claude 3.5 Sonnet, is generally available on #VertexAI! Learn what makes Claude 3.5 Sonnet and Vertex AI the perfect match → goo.gle/claude

28

7 Comments

NVIDIA appears to have developed a groundbreaking new GPU architecture that achieves a 50x improvement in memory performance while reducing energy consumption by a factor of 10! 🤯

472

16 Comments

CTRL Documentable now has local search for all generated #Unreal C++ and #Blueprint properties, functions, and nodes. (we'll be releasing the first version of this plugin within the week) #buildinpublic #unrealengine #ue5 #documentation #opensource #programming #development #gamedev #ctrlframework #indiedev

15

Shocking Claude 3.5 Sonnet Examples, Artifacts Explained, What is AntThinking? --- 💰 Sell White Label AI Agents https://app.stammer.ai/ Timestamps 00:00 Introduction to Claude 3.5 SONNET 00:17 Exploring the Artifacts Feature 02:00 Creating Interactive Applications 04:05 Future of Programming with Claude 05:31 Understanding Ant Thinking 06:34 Deep Dive into Ant Thinking 09:54 Conclusion and Future Prospects

11

2 Comments