Umesh

Umesh

@umeshonai.com

AI Strategist & Temporality/Time Series Researcher | ML/DS Pro | Agentomating Efficiency to Amplify Human Potential | Create AI Agents & Build AI Automation with Me | Some Posts on Food, Health & Philosophy of Life | On the path to 🕉️ | 🇮🇳🇬🇧🇫🇷

919 followers 90 following148 posts

Pinned

Thanks to Blueskyroast tool by @fekri.io I engaged in a bit of self gloating. Link in the next message

Your happiness = Your quality of life - Your envy

Umesh reposted

Many people have asked me how I test coding capabilities of LLMs Here I have given 4 prompts to check C code generation of LLMs that you can try yourself I have also given some additional details of reasons for these tests 1/🧵 #ChatGPT #GeminiAI #LLMs

Umesh reposted

"{x}{y}y Build{z} Th{z}{y}!".format(x="Bu", y="s", z="ing")

Umesh reposted

I am almost certain whenever AGI comes into existence, it will be through the most inconspicuous looking Unititled42.ipynb

Umesh reposted

Government can never be the answer but it can be a good question

Umesh reposted

** The Undefinitive Guide to Prompt Engineering: ** Definition: Prompt engineering is the process of designing & refining input prompts to guide LLMs toward generating desired outputs. It involves clear communication, iterative refinement, and systems thinking to integrate prompts effectively 1/🧵

Umesh reposted

Why alien abductions happen only at night #UFO #Alien #Humor

Umesh reposted

It's confirmed. AI is new Crypto A week later, @awscloud.bsky.social NOVA is now relegated to LinkedIn 😎 Search (not Top, but latest) posts here on Amazon Nova. Hardly any big AI account talking about it Suddenly Google AI is darling of AI experts, fakexperts, gurus & furus

Umesh reposted

When I see a brand new model or a brand new agentic framework (which is practically everyday now) I stare at this poster on my wall and meditate for a few minutes... ...and then start downloading the model or cloning the framework

Umesh reposted

SERMO LIBER VITA IPSA

Umesh reposted

You’re never wrong to do the right thing

SERMO LIBER VITA IPSA

When I see a brand new model or a brand new agentic framework (which is practically everyday now) I stare at this poster on my wall and meditate for a few minutes... ...and then start downloading the model or cloning the framework

It's confirmed. AI is new Crypto A week later, @awscloud.bsky.social NOVA is now relegated to LinkedIn 😎 Search (not Top, but latest) posts here on Amazon Nova. Hardly any big AI account talking about it Suddenly Google AI is darling of AI experts, fakexperts, gurus & furus

You’re never wrong to do the right thing

Why alien abductions happen only at night #UFO #Alien #Humor

An AI Agent without agency is just a glorified computer program hard coded to do some stuff #AI #ML #AIAgents

Be neutral to the quote. Just a reminder

Polaroid imagery of museum of ancient extra-terrestrial technology #GenAI #AI #AIArt

Daguerreotype photographs of museum of technology #AI #AIArt

Umesh reposted

Scorpion AI 🦂

@guruscorpion.bsky.social

Color ink wash

The economic forces don't care about your politics and they are very unforgiving When you abuse economy, it bites back with vengeance France is facing an economic turmoil never seen before Look at the FR 10Y risk spread over DE #economy #france

This might help in mega tool set agent system development where tool definitions will not fit the context window So the tools above desired probability threshold go into the context ensuring reliability RL can further enhance the realised probability of the tools used Link in the comment

t.co/WgZmHOPjqM

t.cohttps://openreview.net/pdf?id=iBZSOh027z

AI agents are the new app layer AI agents are the new IP brands AI agents are the new virtual assistants AI agents are the new therapists AI agents are the new travel agents AI agents are the new friends AI agents are the new comedians AI agents are the new fashion advisors AI agents are the new...

Workflow/data orchestration tools are highly underrated for agentic pipeline building They've amazing capabilities for delayed execution, data managements and dynamic prompt management I am currently experimenting with WindmillDev + pyautogen + @mistralai.bsky.social Videos to follow

Thanks to Blueskyroast tool by @fekri.io I engaged in a bit of self gloating. Link in the next message

Here is the link: blueskyroast.com/roast/umesho...

Heading in the right direction

You need a similar poll here @howard.fm

Why is @hf.co in top 5 blocked? What's going on with #AI / #ML / #OS community here? @clem.hf.co @thomwolf.bsky.social @fdaudens.bsky.social

Let’s see when AI generates this #AI #GenAI #AIArt

Let's have some fun this Thanksgiving Give me a number in the comment and I will create an image for that Mine is 1111 #GenAI #AIArt #AI #Art

If you are looking for Thanksgiving break reading, go for AI Index Report from @stanfordhai.bsky.social with link in the comment Here are the top 10 takeaways from the report: 👉 AI beats humans on some tasks, but not on all. 👉 Industry continues to dominate frontier AI research. 1/🧵

View full thread

👉 The data is in: AI makes workers more productive and leads to higher quality work. 👉 Scientific progress accelerates even further, thanks to AI. 👉 The number of AI regulations in the United States sharply increases. 3/🧵

👉 People across the globe are more cognizant of AI’s potential impact—and more nervous. Here is the link: t.co/0i1fZs1niU 4/🧵

t.cohttps://aiindex.stanford.edu/wp-content/uploads/2024/05/HAI_AI-Index-Report-2024.pdf

If you are looking for Thanksgiving break reading, go for AI Index Report from @stanfordhai.bsky.social with link in the comment Here are the top 10 takeaways from the report: 👉 AI beats humans on some tasks, but not on all. 👉 Industry continues to dominate frontier AI research. 1/🧵

👉 Frontier models get way more expensive. 👉 The United States leads China, the EU, and the U.K. as the leading source of top AI models. 👉 Robust and standardized evaluations for LLM responsibility are seriously lacking. 👉 Generative AI investment skyrockets. 2/🧵

The gap between Knows & Know-nots is widening faster on AI Take a look at some interesting findings, over the last year, from @ipsos.bsky.social ➡️ The proportion of those who think AI will dramatically affect their lives in the next three to five years has increased from 60% to 66%. 1/🧵

➡️ 52% express nervousness toward AI products and services, marking a 13 percentage point rise from 2022. ➡️ In America, Pew data suggests that 52% of Americans report feeling more concerned than excited about AI, rising from 37% in 2022. 2/🧵

Umesh reposted

@handle.invalid

Thread: A Short History of Artificial Intelligence (AI) 1/ Artificial Intelligence (AI) didn’t begin in the 1950s—it has roots that go back centuries! From early dreams of intelligent machines to today’s advanced AI systems, here’s a journey through the evolution of AI. 🧵

For posterity #Notetoself 20241125

BSky Space... woohooo 🎬⁉️ @skyspaces.bsky.social @justinbwatson.bsky.social #space #spaces #livespace #ml #ai

Umesh reposted

Nuff said ⁉️