Dhaya

[Dhaya @dhaya]

@jsuarez @kalomaze @stalkermustang > fwiw I don't think you can do Go wellin 50k

what hard/non-puzzle like problems can be solved in 50k params. genuinely curious on what kind of problems you have been working on with this.

[Dhaya @dhaya]

@vikhyatk He is not talking about the older RTX 6000. RTX Pro 6000 Blackwell has tcgen. No?

[The Figen @TheFigen_]

The cat doesn't want to leave even while working.
Like a kid. ❤️

[Dhaya @dhaya]

@svembu @fizzbhau @Bluearcsys So you are urging indian parents to stop vaccinating their children based on an unverified hypothesis? Do you see the problem here?

[Dhaya @dhaya]

@jsensarma When I read the OP, I interpreted it something like the Talmudic parable of the carob tree. I doubt Sridhar Vembu needs a lecture in network effects. I think he is betting on the long arc of creating them. After all, Bangalore was once just a sleepy town too.

[Dhaya @dhaya]

@hkproj To "live for ideals" means demanding accountability—not pretending murderers and victims are morally equal. Condemn the machine, not just its operators. Until then, your idealism fuels the violence it claims to oppose.

[Dhaya @dhaya]

@hkproj To dismiss politics as a choice is a luxury only the unaffected can afford. The Jews of Europe did not "die for politicians"; they were exterminated by a regime embedded in societal complicity.

[Josh Bongard @DoctorJosh]

Lecture 2 of evolutionary robotics course:

A short history of AI.

https://t.co/sRZgE3OlLq

(Fun stuff starts at 8m30.)

[Dhaya @dhaya]

@semiDL @bindureddy @ClementDelangue Also, it looks they did better than GPT-4 in 2 of the 4 categories and had comparable performance to GPT-4 in the rest two categories.
I was looking at this more recent version of the paper here: https://t.co/1oo5X9xOOL

[Dhaya @dhaya]

@semiDL @bindureddy @ClementDelangue I enjoyed reading the paper. Thanks for the reference.

May be I am missing your point, but it looks like the
"...results show that domain-adaptive pretraining was the primary technique driving enhanced performance in domain-specific tasks."

[Dhaya @dhaya]

@vgr Hmm..These attacks are mentioned in this (rather devastating) critique of NFT/web3. Any thoughts on this video essay?
https://t.co/QTZQyCp45o

[Dhaya @dhaya]

@vgr Time to start cranking out the drills.

https://t.co/Lj57rapUJV

[Alex Kontorovich @AlexKontorovich]

well this is one you don't get every day. Very nice to hear. Way to go @3blue1brown!

[Dhaya @dhaya]

@lemire No masks outside even when two people are talking standing near each other? I know atleast some people who wear masks outside when conversing with others so that the *others* feel safer. I'd invoke the principle of charity and assume that is what is happening in this picture.

[Dhaya @dhaya]

@JohnDCook @lemire Fact check: https://t.co/6ODY6UYCt4
The 94% 'includes both serious complications from COVID-19 infection -- such
as respiratory failure, pneumonia, and adult respiratory distress
syndrome – as well as pre-existing conditions..'

[Dhaya @dhaya]

@mla_sudhakar Dr Sudhakar, Can we please stop counting the total tests done so far and focus on test positivity rate? It is crucial to keep ourselves honest and that metric right in front of our eyes day over day is how we will beat the virus.🙏

[Dhaya @dhaya]

@tubelite @asuph I have a feeling it is going to be epic.

[Dhaya @dhaya]

@kiranshaw There was a directive to this effect. Are hospitals not following this?

[Dhaya @dhaya]

@mpi_jcmace Any favorite thesis that you would recommend?

[Dhaya @dhaya]

@AnooBhu Didn't see this interview. Did he try to lift the distinction between IFR and CFR and assign it to infections and cases?

[Dhaya @dhaya]

@hchawlah Probably no retesting to see if the inmates subsequently developed symptoms. Buried in the middle of the original Reuter's report: "Some people diagnosed as asymptomatic when tested for the coronavirus, however, may go on to develop symptoms later, according to researchers."