sdenton4

32 ms·

541.
by sdenton4 2y ago
This is still too reductive, IMO. The kinda dumb colleague who has memorized all the literature is not just auto-complete. When I ask about some statistical technique that I didn't know the name of, being able to connect my description
542.
by sdenton4 2y ago
Just like the SATs!
543.
by sdenton4 2y ago
The distinction here is between 'the /kind/ of data you're throwing at it' and 'data you have already thrown at it.' The result shows that the LLM has a hard time generalizing to new (basic!) problems that
544.
by sdenton4 2y ago
That's the same limitation as sqlite, actually.
545.
by sdenton4 2y ago
And thinking about my specific project, it's clear that a great solution would allow me to pick row-vs-column storage on a per-table basis. I basically have a graph over vectors, which is a very 'row-storage' kind of things:
546.
by sdenton4 2y ago
Ha, looks like the DuckDB devs have a great and balanced answer for this: https://marclamberti.com/blog/duckdb-getting-started-for-beg... If you want row storage, use sqlite. If you want columnar storage, use DuckDB.
547.
by sdenton4 2y ago
So, hype aside, what's the over/under on DuckDB vs Sqlite these days? I'm working on a Thing right now, has started with sqlite due to being a) good enough, b) stupendously optimized and nails hardened, and c) runs on your ph
548.
by sdenton4 2y ago
This seems like a place where generative augmentation could help - generate an image of a person with demographic values x, y, z and an early stage lung cancer. Then you can generate images for more corners of the days, more evenly, and tra
549.
by sdenton4 2y ago
Exactly this! Fusion power is eight minutes away. Always has been.
550.
by sdenton4 2y ago
Looks like the hallucination rate doesn't improve significantly, but I suppose it's still a win if it helps humans review things faster? Though I could imagine reliance on the tool leading to missing less obvious problems.
551.
by sdenton4 2y ago
Morse code would do the trick! One check for short, two checks for long, two blanks for end-of-character.
552.
by sdenton4 2y ago
The definition of blackout drunk is not remembering what happened. This /could/ simply be a failure to form lasting memories, rather than a loss of 'self-ness.' If we imagine a switch which put a person's long-term
553.
by sdenton4 2y ago
World population grew from 4.6 billion to over 8 billion over the last 40 years, rather more than 50% growth...
554.
by sdenton4 2y ago
Traditional* p-zombies presumably form and act on memories, as well - another strike against the 'drunk people are p-zombies' argument. (* - Not to be confused with the 'fast p-zombies' which were briefly popular in 90&#
555.
by sdenton4 2y ago
In short: "Commoditize the complement."
556.
by sdenton4 2y ago
I think I'll title my memoir: "Drunk people are p-zombies, and other facts I learned on the Internet."
557.
by sdenton4 2y ago
Limited page count?
558.
by sdenton4 2y ago
Just replace the visible buttons with low-power bluetooth beacons...
559.
by sdenton4 2y ago
You can pretty reasonably prune the tree by a factor of 1000... I think the problem that others have brought up - difficulty of the value function - is the more salient problem.
560.
by sdenton4 2y ago
In my experience, LLMs do a very poor job of generalizing. I have also seen self supervised transformer methods usually fail to generalize in my domain (which includes a lot of diversity and domain shifts). For human language, you can paper
561.
SurfPerch model for classification of sounds in coral reefs(techcrunch.com)
2 points by sdenton4 2y ago|1 comments
562.
by sdenton4 2y ago
I worked with Ben Williams to create this model! It's trained on a combination of reef sounds... and bird vocalizations. The model is available to play with here: https://www.kaggle.com/models/google/surfperch
563.
by sdenton4 2y ago
If you make incredibly loud noises above water, humans will complain, and you'll end up reducing the noise, due to some combination of workplace safety regulations, litigious neighbors, and self-regulation. People don't experience
564.
by sdenton4 2y ago
Of course, if you end up with an incorrect result due to assuming the Riemann hypothesis, you'll get a fields medal anyway...
565.
by sdenton4 2y ago
Was the song actually that good, or did your brain simply tickle the 'appreciation for incredible beauty' neurons while playing back some Nickelback memories?
566.
by sdenton4 2y ago
... The replication crisis does not - I repeat, does not - lower the standard for acceptable evidence in the sciences.
567.
How to produce the anarchy symbol in TeX(tex.stackexchange.com)
4 points by sdenton4 2y ago|0 comments
568.
by sdenton4 2y ago
Well, the jokes on you: Now we all know that you're in northern Iowa.
569.
by sdenton4 2y ago
It's often people who have a hard time saying no who end up spread thin over a number of gigs - this leads to burn out. Or not showing up to meetings... The 3/5 of the board who can't be rounded up are also signed up for too
570.
by sdenton4 2y ago
Parquet is columnar storage, which is much faster for querying. And typically for protobuf you deserialize each row, which has a performance cost - you need to deserialize the whole message, and can't get just the field you want. So, o

More ›