Daniel Erenrich

@PCBWayOfficial W134303A

@derenrich Message me your PCBWay account number. You can register as a PCBWay member here: pcbway.com/Member/Login/

👇Hurry over to claim your #Halloween "Treat"! Rules: ①Follow @PCBWayOfficial ②Post A Picture of your pumpkin in the comments👇 Treats: 🎁You'll get a $2 coupon. 🎃For every 50 pumpkins posted, one lucky winner will get a FREE ORDER! 50 - 100 pumpkins, 2 FREE ORDER! 100-300pumpkins, 6 FREE ORDER! #draw #HalloweenGiveaway #Giveaway #Contest #Win #Prize #LuckyDraw #Gift #ShareToWin #EnterToWin #WinFree

@antirez @sergeax But you are notable? You exist in both wikidata.org/wiki/Q106368090 and google.com/search?kgmid=/…

@sergeax This is exactly the reverse: I used my name to be sure the LLM didn't know my dates because I'm not notable.

Well, this is extremely convincing...

@SenJoshBecker QR code seems to be dead? And the link isn't in your press-releases or other obvious places

@molly0xFFF Technically the blockchain portion of golden was split off into its own project that is limping along. But it's not looking good

the page was last edited on May 24 by "Golden AI" the two edits before that (May 2) were by a person who works for Golden as a "QA editor" they don't appear to have a publicly visible feed of all changes, but this sure feels like a ghost town

remember golden, which was supposed to be web3's project to "fix" wikipedia by adding crypto incentives for editing? you don't get points for editing it anymore and they've done a full pivot to AI twitter.com/molly0xFFF/sta…

@Ben_Reinhardt Does the Allen institute comin count? They do ai research (amongst other things)

Some examples (I think?) are: - HHMI Janelia - The Salk Institute - Otherlabs (Particularly interested in orgs outside of human health!)

Input needed! What are your favorite examples of successful non-academic* research organizations outside of industrial and government-run labs? *"academic" is hard to define, but I think it's some combination of "PIs are professors" and "a lot of labor done by grad students"

@chrisalbon As others have said, pyenv

What is "the right way" to install Python on a new M2 MacBook? I assume it isn't the system Python3 right? Maybe Homebrew?

@weel @GeorgeHagstrom i'm no expert but if you're planning on filing an insurance claim against the driver the big bill will be paid by them? i was recently in an accident and people advised me to see a doctor just in case (I went to urgent care not a hospital and got some x-rays). good luck.

@GeorgeHagstrom got a bruised shin and val has a bruise on their hip. we managed to get some ice packs from the paramedics. didn’t want them to take us to sf general just to sit around for hours and be hit with a big bill so now we’re at a restaurant on mission calming down.

just got hit by a car at 19th and guerrero in sf. ON THE SIDEWALK. wide open intersection, not even bump outs, nothing to keep a car that’s hit in a crash from sliding right up the sidewalk.

@judegomila my cat "exists" but I don't think there should be a Wikipedia/Golden page for her. What matters is verifiability.

notability without mathematical definition is an irrelevant measure; existence is all normalize not by not just popularity but by meaning aka vector space clustering, graph topology, density and any other measure you wish to see as a structure of meaning

@PeterRoelants @dshieble I'm really surprised this is the suggested way to encode semi structured data? Are there not better ways to encode this? Maybe if you're fine tuning?

@dshieble > replace raw numbers or computed values with descriptive text like small/medium/large or comparative figures like “twice as large as the median value” Interesting take in how to add context to interpret these values.

@steve47285 @Ben_Reinhardt also consider dbpedia.org which is an effort to structure the information in Wikipedia using algorithms. turns out there's like a million parallel open data efforts.

@Ben_Reinhardt that's what wikidata is for though right? en.wikipedia.org/wiki/Wikidata

Has anybody made an actually good publicly editable database? or is it the sort of thing where you need a single curator and an easy way of suggesting changes?

@Ben_Reinhardt Wikipedia, Wikidata, Golden, DMoz, MusicBrainz, Open Food Facts, OpenStreetMap, RECAP, IMDB. A lot of citizen science initiatives e.g. BirdTrack. Tons of genealogical databases. Really depends what you mean by "actually good" and "database".

@Landgren Unfortunate, maybe cache the list of usernames locally then. It feels a little abusive of the Wikidata query service to do this fetch so often. Though it probably doesn't matter

@derenrich The AI suggested a similar query but it is now too slow. I'm reverting back.

I wrote an "Verified by Wikipedia badge" Chrome extension in less than two hours using GPT-4. Only using Swedish. Here is the transcript. shareg.pt/Wm4Ap6u

@Landgren ok the fix should fix it but a consequence of the fix is that you are downloading 357,453 rows of data every time you see a new user. try a query like w.wiki/6cXE

@derenrich Lets ask the AI to fix it

@Landgren right. but it shouldn't be case sensitive. twitter handles are not case sensitive.

@derenrich It is case sensitive. Could you add an issue describing a case where it’s not working as expected?

@Landgren there's a capitalization/case issue here. might need to normalize the case on both sides or do the search using twitter numeric IDs

Here is the working code: github.com/irony/verified…

@andy_matuschak can't powerful chemical weapons already be assembled for prices cheaper than that? how much is chlorine gas?

At least they didn't publish the weights. But they trained it with 16 V100s, so it wouldn't be very expensive to reproduce. Feels a bit like reading a paper with detailed synthesis steps for a powerful chemical weapon which can be assembled for ~$10^4-10^5. Cool, good job!

Microsoft researchers, in their paper on a model which can synthesize high-quality speech in any voice with three seconds of training data, closes with this powerful ethical statement: "we appeal to our practitioners not to abuse this technology." The Microsoft AI Principles!

@JesseThorn Sorry for the unsolicited advice but just in case you haven't, have you looked into nurtec? Works well for me. (Also other cgrp inhibitors). Feel better

@MagnusManske @wikidata i think it would depend on what classification scheme item is being used.

@chrisalbon i would use this if WMF made this (or similar) accessible for all articles on an ongoing basis