realvenki

GPT/LLMs, Low Code, Technology Enthusiast, Enterprise Social Nut

[realvenki @realvenki]

@wabi build me an app that can help me make progress on my yearly priorities

[realvenki @realvenki]

@rao2z @iclr_conf Please do share your schedule once finalized. Will turn up with Banganapallis at one of the locations to thank you for your videos and posts (including the Sunday Harangues) 😀

[realvenki @realvenki]

https://t.co/REhKSAeEMV

A long and well structured article from Dario Amodei the founder of Anthropic (makers of Claude). He explains the economics of innovation/progress in GenAI, dissects what happened with DeepSeek’s new models and argues for Export Control by the US to prevent China from getting access to latest US tech. Fairly articulate and well laid out argument (of course from a US perspective). Helps in understanding the DeepSeek saga from one of the industry insiders. Please note that he is an interested party, and brings his bias/glasses to the topic. Still worth a read for the perspective.

[realvenki @realvenki]

With Think Deeper being the feature name from Microsoft, as and when Copilot reaches a big user base, they can swap out the model in the back-end without much worry about messaging. And the ability to perform reasoning on my knowledge store which are all in M365 would make it easier to adopt if the feature works well.

[realvenki @realvenki]

https://t.co/fvgKhImBHb

My take on DeepSeek-R1's impact on the markets, India and the world

[realvenki @realvenki]

https://t.co/ZKIfPPqTvW ... $500 Billion Investment. In AI Infrastructure. Unbelievable numbers and scale. Dylan Patel called it in the Dwarakesh podcast 4 months back. To see Deepseek R1 match o1 performance in benchmarks and the controversy around OpenAI having access to Epoch benchmark data, and still have them line up such an investment points to something. I am not too sure what it is. Maybe OpenAI has tech that it has not yet released. Maybe the thesis about lining up the investments to continue to reap benefits of scaling laws .. which might leave others without the compute behind in the use cases that rake in the $$. I can say one thing for sure: exciting times ahead.

[realvenki @realvenki]

Elegant way to visualize a few more dimensions beyond the 3rd. This solves a long time ask for me.

[spike @spikedoanz]

they don't want you to know this, but you can visualize arbitrarily hyper dimensional spaces in your head.

just keep stacking!

[realvenki @realvenki]

@rao2z Though Mumbai's density and crowds can be forbidding, I find that I surprisingly adopt very well to Mumbai. I suppose Chennai to Mumbai would be an easier cope than Tempe to Mumbai. Good to see you take in the Marine Drive vibe.

[realvenki @realvenki]

🚀 NVIDIA's Project DIGITS is shaking up desktop AI! • $3k starting price • 128GB unified memory • May 2024 release • Potential game-changer for home LLM inference
Exciting times ahead for AI enthusiasts! Can't wait to see real-world benchmarks. The future of personal AI is looking brighter (and more compact) than ever! 🤖💻 #ProjectDIGITS #NVIDIA #AIComputing https://t.co/aLucRqxOJb

[realvenki @realvenki]

@iruletheworldmo

[realvenki @realvenki]

@Scobleizer @iruletheworldmo @xai To be fair to the doubters, Elon's ambitions have been outsized. Crazy to see that he has delivered on a lot of those.

[realvenki @realvenki]

@iruletheworldmo @xai How is the alignment, safety culture at xAI?

[realvenki @realvenki]

@OfficialLoganK Would the team at SSI be solely guided by benchmarks and internal evaluations? Would take a lot of motivation to stay the course.

[realvenki @realvenki]

o3 sounds exciting. Gemini 2.0 Flash Thinking Experimental is already here and is cool. Exciting times ahead.

How did we get here? From Sam Altman's The Intelligence Age:

How did we get to the doorstep of the next leap in prosperity?

In three words: deep learning worked.

In 15 words: deep learning worked, got predictably better with scale, and we dedicated increasing resources to it.

[realvenki @realvenki]

https://t.co/G5X6w2DhtP Finally after weeks of speculation, an interesting announcement from OpenAI. Waiting to see if it lives up to the hype. Reasoning and Long Range Task Coherence would be game changers for GenAI

[OpenAI @OpenAI]

We're releasing a preview of OpenAI o1—a new series of AI models designed to spend more time thinking before they respond.

These models can reason through complex tasks and solve harder problems than previous models in science, coding, and math. https://t.co/peKzzKX1bu

[killian @hellokillian]

●

In less than 48 hours, over 200+ engineers and designers have reached out to join this project.

Over the next six weeks, we are going to collectively build the world's first open-source Language Model Computer (LMC).

We will also publish a family of standard protocols to advance the LMC ecosystem as a whole.

The 01 design will be an attempt to build the Linux of this space—open, powerful, and free as in freedom.

The core team is in Seattle. We start Monday.

[realvenki @realvenki]

I think I first heard this from Anthony Robbins: "We tend to overestimate what we can do in one year. We tend to underestimate what we can do in ten years".

Should be true because I often think of and hear about "New Year Plans" and rarely/never about "New Decade Plans" 🤔

[realvenki @realvenki]

Frontier models are no doubt important and India needs to find its way into the space for long term purposes. However the possibilities of inclusive growth represented by smaller open source/open weight models that work well with Indian languages is exciting. Kudos @SarvamAI

[realvenki @realvenki]

Excited to see Sarvam AI's first version of OpenHathi model for Hindi being launched. Check out the video here: https://t.co/psQWWFZ2jL. You can find their blogpost here: https://t.co/GNhytsJFPJ

This could be a leapfrog moment for India.