Rajesh (@_rajesh_ganesan) 's Twitter Profile
Rajesh

@_rajesh_ganesan

Human (not a bot)

ID: 218929692

calendar_today23-11-2010 14:30:47

786 Tweet

171 Takipçi

3,3K Takip Edilen

Albert Edwards (@albertedwards99) 's Twitter Profile Photo

Personally having been at my desk through the 1987 and 2000 crashes I'm surprised that anyone thinks this AI induced bubble is NOT going to end with M7/tech crashing back to earth. And when I say crash, I mean >50%. marketwatch.com/livecoverage/s…

opdroid1234 (@opdroid1234) 's Twitter Profile Photo

Arpit Bhayani I dont think you are getting the "platform play" aspect of it. If one user manages to brick a PC, it is the users fault. If millions of PCs are bricked worldwide it is the fault of the company that shepherds the platform. This should not have been possible by design.

Jonathan Ross (@jonathanross321) 's Twitter Profile Photo

What can you do with Llama quality and Groq speed? You can do Instant. That's what. Try Llama 3.1 8B for instant intelligence on groq.com.

Yangqing Jia (@jiayq) 's Twitter Profile Photo

People often ask why prices like $2.8/m token for Llama 405B, while being super fast, are still profitable at Lepton AI. We've even been asked by a leading GPU provider! So, I figured we should share some technical analysis. This information could benefit the community. We've

Rohan Paul (@rohanpaul_ai) 's Twitter Profile Photo

WOW, SambaNova is giving over 123 tokens/second on llama 3.1 405B with their ASIC hardware 🤯 Llama 3.1 8B at over 1000 tokens / second 🤯 Also, you use it without any signup

WOW, SambaNova is giving over 123 tokens/second on llama 3.1 405B with their ASIC hardware 🤯

Llama 3.1 8B at over 1000 tokens / second 🤯

Also, you use it without any signup
Garry Tan (@garrytan) 's Twitter Profile Photo

“Working from home was more important than winning” I guess I get why they are trying to delete this interview off the Internet. Too much truth shared. In a time when AI is speeding up the rate of change, you can’t compete if you don’t have a work ethic.

Arnaud Bertrand (@rnaudbertrand) 's Twitter Profile Photo

This is key to understanding China. It has only 11.6% arable land, concentrated to the East of the Hu line. Which means that it has an enormous density of population for its homegrown food resources: about 12 people per hectare of arable land. For comparison, India - a country

This is key to understanding China. It has only 11.6% arable land, concentrated to the East of the Hu line.

Which means that it has an enormous density of population for its homegrown food resources: about 12 people per hectare of arable land. 

For comparison, India - a country
Rajesh (@_rajesh_ganesan) 's Twitter Profile Photo

After listening to Prof. Subbarao Kambhampati (కంభంపాటి సుబ్బారావు) talks, I tried a simple experiment on #Claude. The results show the limitations of an #LLM in completing a word problem of a multiplication which we think should be something simple given it could do a direct multiplication of the same numbers

After listening to Prof. <a href="/rao2z/">Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)</a> talks, I tried a simple experiment on #Claude. The results show the limitations of an #LLM in completing a word problem of a multiplication which we think should be something simple given it could do a direct multiplication of the same numbers
Anton McGonnell (@aton2006) 's Twitter Profile Photo

There are two important insights that should come from this launch: 1. Three AI chip startups have proven that GPUs are the wrong architecture for inference. This is a huge signal to the market. Developers want fast inference and GPUs can't give it to them. 2. It is not just

Mohamed A. El-Erian (@elerianm) 's Twitter Profile Photo

Up 11 basis points to 4.51%, and down 3%: The reactions of markets — specifically the yield on the US 10-year government bond and the S&P stock index — to this afternoon’s Federal Reserve communication, the latest example of what has tended to be confused and confusing flip flops

Up 11 basis points to 4.51%, and down 3%:
The reactions of markets — specifically the yield on the US 10-year government bond and the S&amp;P stock index — to this afternoon’s Federal Reserve communication, the latest example of what has tended to be confused and confusing flip flops