- 15
- 8 229
Mohammad Reza Bonyadi
Приєднався 30 вер 2011
An automatic drive for racing.
Applied AI News #3
Hello and welcome to the Applied AI News #3. In this issue:
- Industry News and Standards: "intention economy" warning, ChatGPT recommendations impacting consumer choice, AI digital employees redefining work, Copilot Vision 'seeing' web pages, and Microsoft pre-built AI agents.
- Academic and Science: Generative agents for AI-driven innovation, LLMs alone insufficient, needs user habits & assistants, inference-time compute beats scaling model size, LLMs and theory of mind, and more
- Hands-on: LLM fine-tuned on text messages for digital "me", Small Language Models are cost effective solutions, AI Agent Workflow Design Patterns overview, LLMs alone may not scale well for multi-agent systems
Subscribe and share if you enjoyed it!
Full AI news series: www.linkedin.com/newsletters/applied-ai-news-7277932155001880576/
Spotify: open.spotify.com/show/6lNEkSvndIRjBnoXMYITqL
- Industry News and Standards: "intention economy" warning, ChatGPT recommendations impacting consumer choice, AI digital employees redefining work, Copilot Vision 'seeing' web pages, and Microsoft pre-built AI agents.
- Academic and Science: Generative agents for AI-driven innovation, LLMs alone insufficient, needs user habits & assistants, inference-time compute beats scaling model size, LLMs and theory of mind, and more
- Hands-on: LLM fine-tuned on text messages for digital "me", Small Language Models are cost effective solutions, AI Agent Workflow Design Patterns overview, LLMs alone may not scale well for multi-agent systems
Subscribe and share if you enjoyed it!
Full AI news series: www.linkedin.com/newsletters/applied-ai-news-7277932155001880576/
Spotify: open.spotify.com/show/6lNEkSvndIRjBnoXMYITqL
Переглядів: 805
Відео
Applied AI News #2
Переглядів 408День тому
Hello and welcome to the Applied AI News. Today we’ll explore the rapidly evolving world of AI, focusing on breakthroughs that are impacting how we work and interact with technology. We'll get into Microsoft's new model, Phi-4, and how it challenges the notion of ever-increasing model size. We’ll explore the shifting landscape of business applications toward an AI-first approach. We’ll also exa...
Applied AI News #1
Переглядів 55714 днів тому
Hello and welcome to Applied AI News. Today, we'll explore some of the latest advancements in AI, covering new industry tools and academic breakthroughs. We'll examine how Meta is changing speech AI, NVIDIA is envisioning AI coworkers, and Google is aiming at enterprise AI integration. We will get into the technical details of GUI interaction, language model efficiency, concept models, and new ...
digital RA
Переглядів 872 місяці тому
Provide a paragraph describing an idea, the Digital RA is "hired" for you with the right skills (overridable by you), performs a paper search for you, estimates the relevance between papers and your idea, picks a subset, and writes a small section of Literature Review for your review and final inclusion in your grant or paper. MORE: You can chat with the RA about the papers, your ideas, how are...
How do I use ChatGPT mobile app for brainstorming: Subjective Self-Alignment in LLMs
Переглядів 543 місяці тому
I use ChatGPT app everywhere for brainstorming, from during driving (in this video) to during walking or during shopping. I usually start with a topic, we brainstorm together, I ask it to do the Internet search (which is not too bad, but Copilot is much better at that), and write a report, and a code if needed. Just an example. You can do much more, probably... Summary: In this conversation, I ...
Ahura: Racestar eroad
Переглядів 66010 років тому
This is a driver based on Evolutionary strategy. See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Ahura: Racestar Brondehach
Переглядів 17210 років тому
See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Ahura on Alpine 1
Переглядів 18510 років тому
See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Overtake 4 by Ahura
Переглядів 20410 років тому
How Ahura overtakes sample. See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Overtake 1 by Ahura
Переглядів 12910 років тому
This is how Ahura overtakes. The other vehicles move with constant speed with steady driving style. See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Overtake 2 by Ahura
Переглядів 9910 років тому
This is how Ahura overtakes. The other vehicles move with constant speed with steady driving style. See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Overtake 3 by Ahura
Переглядів 10310 років тому
This is how Ahura overtakes. The other vehicles move with constant speed with steady driving style. See my website: sites.google.com/site/mohammadrezabonyadi/standarddatabases/simulated-car-racing
Ahura: A human-like racer for TORCS using Evolutionary strategy
Переглядів 52710 років тому
This is a driver that I designed for a car racing simulator called TORCS (see en.wikipedia.org/wiki/TORCS and torcs.sourceforge.net/). It uses an Evolutionary Strategy to set some parameters that are used for an in-line decision making on the action necessary to control a car, such as clutch, steering, brake, gear, and acceleration pedal. It has some other mechanisms to adapt parameters during ...
My picks: 1) Your brand through ChatGPT eyes, 2) Copilot Vision, 3) Agents are not enough, and 4) Scaling LLM Test-Time Compute.
Issue #3 is out: ua-cam.com/video/yaa05k0MCUQ/v-deo.html
Issue 2 is out: www.linkedin.com/pulse/applied-ai-news-2-reza-bonyadi-umptf Prefer listening instead of reading? You’re covered! 🎧 Listen on Spotify: open.spotify.com/episode/22MpSSAFVAwuTx4U9nK4db 📺 Watch on UA-cam: ua-cam.com/video/8j0FyXe0Hyo/v-deo.html
AI gents are growing, but designing agents is a mishmash of gut feels and guesses. Anthropic provides a set of best practices for building effective LLM agents here: www.anthropic.com/research/building-effective-agents This is a fantastic, practical and straightforward guideline to creating and deploying LLM-based agents. The article discusses the problem of building effective LLM agents, specifically highlighting the challenges of relying on overly complex frameworks. It also tackles the lack of transparency and control (aka, evaluation) in existing solutions, where direct API interactions are limited. See more here: www.linkedin.com/pulse/applied-ai-news-1-reza-bonyadi-abzxf If you prefer audio then listen to an extended commentary in - Spotify (open.spotify.com/episode/5jY3AujFf1TBJDe9EAJM0Z)
So great, wonderful!👏
This is called Galaxy unpacked 😆
Wonderful car, totally nailed it :)
good stuff :)