cdrini

18 ms·

241.
by cdrini 2y ago
I've been pleasantly surprised by GitHub's "copilot workspace" feature for creating near production code. It takes a GitHub issue, converts it to a specification, then to a list of proposed edits to a set of files, then
242.
by cdrini 2y ago
Not a huge fan of the way it's framed here. This is a very personal experience framed as very general guidelines. And some of the assumptions I find a little much: - if you drink alcohol, you probably need therapy - you only think you
243.
by cdrini 2y ago
Yes exactly this. I found the article kind of confusing cause it didn't describe anything I'd ever experienced, and then she mentioned London and it all made sense. Drinking is a muuuuuch bigger part of the culture there from what
244.
by cdrini 2y ago
Oh I've also seen some things recently about encoding your data in a knowledge graph to be used in conjunction with RAG? But that seems more cutting edge; not sure of its efficacy.
245.
by cdrini 2y ago
Your two options are RAG or fine-tuning. RAG (with vector db) I've found to be a little finicky, and requires a decent amount of preprocessing of your data in order to split it up into reasonable chunks that vector encode well. I also
246.
by cdrini 2y ago
> I would do and say things I wouldn't have were I not inebriated Although I mean, isn't that the point? If your social anxiety was inhibiting you from eg going to a social event, asking someone out, or speaking in front of a l
247.
by cdrini 2y ago
It's not quite so black and white, since poisons are generally defined in conjunction with their dose. Eg "The dose makes the poison". With alcohol, we don't really have a dose, like we do with eg vitamin A, which is tox
248.
by cdrini 2y ago
> we say "the dose makes the poison" but it doesn't change that regularly drinking a low dose of poison is not good. I'm not sure that's the right conclusion there. Vitamin A is a poison, because it's toxic
249.
by cdrini 2y ago
Downvoting isn't really a mechanism to censor, flagging is. The intent of downvoting is usually just disagreement. My guess would be you're being downvoted for a few reasons: - A few of the people you mentioned are somewhat polari
250.
by cdrini 2y ago
I'm not sure I follow that logic. Nothing about justinhj's statement implied "we should do this regardless of what anyone else thinks". They just stated something which is a priority to them. They're taking part in
251.
by cdrini 2y ago
I wish folks wouldn't speak in absolutes this much, it's incorrect and tends to raise the temperature of a generally already heated discussion. Biden's approval rating was like 40% for most of his term. So many many many peop
252.
by cdrini 2y ago
Hmm why should they consider those things? They stated their opinion and priorities, which will likely influence their voting decision. The views of HN or the American electorate play no role in the statement of that individual's prior
253.
by cdrini 2y ago
Hmm interesting take. If I understand you correctly, your theory of change is that by not voting in this election (since you see either candidate as similarly good/bad), you might have a long term impact where, in the next election, pa
254.
by cdrini 2y ago
The previous post said "indirectly" controls; maybe a misread?
255.
by cdrini 2y ago
Yeah; it's a shame, cause at its core a lot of these platforms are meant to connect people who create things with people who might be interested in those creations. But it all hinges on the algorithm to make that connection, and that a
256.
by cdrini 2y ago
This is a topic which I've been thinking about a lot, and I think one problem is with the recommendation algorithm using likes/views/comments as a signal for what to propogate in people's feeds. This creates an incentive
257.
by cdrini 2y ago
Don't know what your problem is, I'm literally agreeing with you and have said three times now that LLMs can memorize. Good bye.
258.
by cdrini 2y ago
You did it again, memory != memorization! People don't claim that "there is no memory"; the argument you're trying to make is "people claim LLMs do not memorize". Memory means something completely different. Me
259.
by cdrini 2y ago
At no point did I claim whether they could or could not memorize. I was disagreeing with your statement "How can the AI spit out facts about a metal band if it has no memory". No one has ever said "AI's have no memory&qu
260.
by cdrini 2y ago
I think there are a few different ways you can define "memory" and "memorization" here. When folks say "memorising" in the context of AI they mean "Does they AI have chunks of its training data fully/
261.
by cdrini 2y ago
You're free to speculate, but they have on multiple occasions said they don't train on private repos. Furthermore, there's no real incentive for them to do so, since (1) there are a lot of public repos, and (2) training on pr
262.
by cdrini 2y ago
I disagree with the statement "they are trained to duplicate" because "to" implies a purpose/intent which is incorrect. I.e. "they are trained with the purpose of duplication". This is I believe pretty unc
263.
by cdrini 2y ago
They're not "inherently trained to duplicate"; I think that's a bit of a disingenuous oversimplification. They're trained to learn abstract patterns in large datasets, and remix those patterns in response to a promp
264.
by cdrini 2y ago
I mean, I imagine it used a lot of their public code, like VS code, typescript, the new windows terminal, or anything on https://github.com/microsoft . They didn't use their private code, but they didn't use anyon
265.
by cdrini 2y ago
I think there's a difference between a mathematical proof and legal proof. The mathematical proof would be "show that it never copies a code snippet verbatim", and you of course cannot prove that by example. Legal proof is I
266.
by cdrini 2y ago
Much ado about a single line of CSS :P I thought this would lead to something a little more interesting, like dynamically modifying/drawing the elements of the map or something. CSS filter is the way to go; and I'm glad to see tha
267.
by cdrini 2y ago
Good catch! Today I learned.
268.
by cdrini 2y ago
Same actually! Here's an old project: https://rawgit.com/cdrini/wikidata-spiral/master/index.html?... Do you have any of your old projects public to share? I love creative UI stuff!
269.
by cdrini 2y ago
Yep that can be confusing at times! Although I've found SVG to be pretty predictable with regards to this.
270.
by cdrini 2y ago
Not a huge fan of this hot-take that pops up every once in a while. I love SVG and use it all the time, and think it's a fantastically designed language and system. The places where it's lacking have nothing to do with bloat or ve

More ›