mdp2021

37 ms·

961.
by mdp2021 1y ago
Agency within the world model is sufficient.
962.
by mdp2021 1y ago
> how they think they can force an LLM to pull from the memory bank instead of the training data You have to implement procedurality first (e.g. counting, after proper instancing of ideas).
963.
by mdp2021 1y ago
We do not «replicate». We implemented computing without any need of a brain-neural theory of arithmetic.
964.
by mdp2021 1y ago
Intelligence (solving problems) does not require consciousness.
965.
by mdp2021 1y ago
> this was written using How do you know?
966.
by mdp2021 1y ago
If "context is king" in the LLMs age... Well give us at least some context.
967.
by mdp2021 1y ago
It's relevant given the apparent conservativeness about LLMs, the non-revolutionary architectural improvements that seemed to place too much of the bets on scaling.
968.
Game-Changer for Local LLMs: AMD Medusa Halo Points to 384-Bit LPDDR6 Bandwidth(hardware-corner.net)
3 points by mdp2021 1y ago|1 comments
969.
Cochrane library, global source of indep. health evidence: NZ restricting access(theconversation.com)
4 points by mdp2021 1y ago|1 comments
970.
by mdp2021 1y ago
> According to the WHO, between 2010 and 2015 there were 3,248 cases out of the worlds population of 8.2 billion, of which 584 people died. so we can safely say we are safe If you imagined a counterfactual reality in which the plague h
971.
by mdp2021 1y ago
> same intelligence as gpt-oss-120B Let's hope not, because gpt-oss-120B can be dramatically moronical. I am guessing the MoE contains some very dumb subnets. Benchmarks can be a starting point, but you really have to see how the
972.
by mdp2021 1y ago
It is important to know that some places still have endemic infection in rodents. For example, in a world were some people (even those of a medical background, looking at data) die of rabies because "oh look a wild animal approaching w
973.
by mdp2021 1y ago
(And I really would like to understand the reason under the following expression: > though 50% of cases occur in people ages 12–45 ...that's already near half of the USA population.)
974.
by mdp2021 1y ago
> pretty common The interpretation of "common" will vary: > [In recent decades, a]n average of seven human plague cases are reported each year in the United States > The last urban plague epidemic in the United State
975.
California Resident Tests Positive for Plague. What to Know About the Disease(time.com)
7 points by mdp2021 1y ago|7 comments
976.
"Deeply concerning": reading for fun in the US has fallen by 40%(theguardian.com)
4 points by mdp2021 1y ago|0 comments
977.
by mdp2021 1y ago
Except the chief argument remains the distinction between goods of difficult replication vs goods of cheap replication. And except all the rest in that illogic.
978.
by mdp2021 1y ago
The mission is to give people the tools that enable the mental proficiency that allows to make clearminded statements and avoid skullconcussed ones.
979.
We should thank pigeons for our AI breakthroughs(technologyreview.com)
4 points by mdp2021 1y ago|0 comments
980.
by mdp2021 1y ago
But they will probably thought models, not just language models. The engineering will be different.
981.
by mdp2021 1y ago
> as you seem to be insinuating with The implementation details are not clear, not the goals. I never said that the feature has to be coded explicitly. I said it has to be there.
982.
by mdp2021 1y ago
Personal development. It's a "subtle" skill. You train it (though maybe less directly than other skills).
983.
by mdp2021 1y ago
> An AI might say Those which can say will say. It won't make it much different from what we have to process daily from other utterers. That's why we downplay statements and value analysis.
984.
by mdp2021 1y ago
Sure but, you wrote: > If anything, this shows that some LLMs might win against humans because they can spend more time thinking per wall clock time interval thanks to the underlying hardware. Not because they are fundamentally smarter.
985.
by mdp2021 1y ago
But when an LLM can fail though having all the time in the world, you are pretty certain you hit a wall. So, in a way you have defined a good indicator for a limit for a certain area.
986.
by mdp2021 1y ago
Consider the source: https://www.eff.org/deeplinks/2025/08/no-uks-online-safety-a... I understand from that source that the legislation "mandates that any site accessible in the UK - including social med
987.
by mdp2021 1y ago
> Let me know when you find one of these "rational agents". Hi there. > Again, livability "for whom?" For the rational. The people who tend to dis-want electroshocking cutlery and miasmatic nail polish. > D
988.
by mdp2021 1y ago
With the current market, for a price you can turn things off.
989.
by mdp2021 1y ago
In fact I miswrote: I understand it is about presenting a document to access webpages. In the equivalent proposal in Australia, they spoke about presenting a document to access search engines. Surely my mention of "forums" was rus
990.
by mdp2021 1y ago
> Car hacking will be normalized even more We should be concerned of the increased possibility of malicious third-party hacking of other people's property. And, I should not hack into my desk to open my drawers.

More ›