afro88

14 ms·

211.
by afro88 1y ago
The author is missing the bit that the LLM provides: automatically mapping input parameters to things the user wants to do, and responses to the way the UI displays them. Take out the LLM and you're not that far away from existing prot
212.
by afro88 1y ago
Strangely the image change examples (edits, style transfer etc) have that slight yellow tint that GPT Image 1 (ChatGPT 4o's latest image model) has. Why is that? Flux Kontext doesn't seem to do that
213.
by afro88 1y ago
Same. Agents can be very expensive. Plus I have no idea how reliable or effective this actually is in practice. Would love to try it first.
214.
by afro88 1y ago
My social anxiety sky rocketed at 19 and didn't start to settle down until 23/24. I never knew exactly why, and that's always troubled me a bit. A microbe population getting out of hand sounds possible, as my diet and surroun
215.
by afro88 1y ago
ADHD-I has a range of symptoms where the person needs 5 or more that are significantly disruptive to their life for at least 6 months. So when someone reads the first paragraph and immediately thinks the author has to be ADHD because they t
216.
by afro88 1y ago
Genuine question: isn't this everyone? Don't we all find large tasks hard to start and so we procrastinate? Isn't it common general advice for all people to break things down into smaller steps so you can get going
217.
by afro88 1y ago
Great post, and sums up my recent experience with Cursor. There has been a jump in effectiveness that only happened recently, that is articulated well very late in the post: > The answer is a critical chunk of the work for making agents
218.
by afro88 1y ago
> What this interaction shows is how much knowledge you need to bring when you interact with an LLM. The “one big flaw” Claude produced in the middle would probably not have been spotted by someone less experienced with crypto code than
219.
by afro88 1y ago
The same way you context switch between whatever your day to day is and reviewing peers code
220.
by afro88 1y ago
He means 30% of the time is waiting around for agents to reach a conclusion and need human review and a next task to tackle.
221.
by afro88 1y ago
Our minds are used to that: dreams
222.
by afro88 1y ago
If you look at the sources behind the AI responses for this search, they clearly don't mention the 280. Google are probably using a dirt cheap model for these responses and it's harming user trust
223.
by afro88 1y ago
What's the app you created and published with Rork?
224.
by afro88 1y ago
You're showing me pricing in euros, which is a bit off putting (I'm in Australia). It wouldn't stop me if I knew your product already and knew I needed it, but as a "cold" visitor it's a small thing that stops
225.
by afro88 1y ago
A lot of it is due to the tokenisation system though. It just doesn't represent individual letters in words. And at the moment reasoning models aren't trained to know how to spell all words, and to know when to spell them out to u
226.
by afro88 1y ago
Though I agree with the sentiment that Miyazaki must hate OpenAI commodifying his style, I find it so ironic that people misuse a quote of his from 2016 to be about AI generally. Another kick in the nuts for the poor guy. He was talking abo
227.
by afro88 1y ago
Cryptic crossword clues that involves letter shuffling (anagrams, container etc). Or, ask it to explain how to solve cryptic crosswords with examples
228.
by afro88 1y ago
> most importantly, it's just 1,823 well organized commits (1,329 of which belong to the creator). That gives me confidence to keep it around long term. Genuinely curious: why does a low number of commits, and mostly commits from on
229.
by afro88 1y ago
For what it's worth, I have tried it on a couple of private repos today and the quality is incredible. Give it a shot. I think Gemini 2.5 Pro is doing a lot of the heavy lifting here. I have tried this sort of thing before (documentati
230.
by afro88 1y ago
This is part of what the article is about
231.
by afro88 1y ago
Fair, there are tons of docstrings. I have had the opposite experience with LLMs explaining code, so I am biased towards assuming this works. I'm keen to try it and see.
232.
by afro88 1y ago
You don't point this tool at the documentation though. You point it at a repo. Granted, this example (and others) have plenty of inline documentation. And, public documentation is likely in the training data for LLMs. But, this is more
233.
by afro88 1y ago
> You didn't "build an AI". True > It's more like you wrote a prompt. False > I wonder why all examples are from projects with great docs already so it doesn't even need to read the actual code. False. This
234.
by afro88 1y ago
> First, there’s a signal being given about the complexity and fragility of the software. A 16GB RAM requirement just screams “this is a big, complex piece of software that will break in mysterious ways”. Not very scientific, I know, but
235.
by afro88 1y ago
Or errors an LLM agent would make...
236.
by afro88 2y ago
Great post, and nice recreation! I went through a period of recreating songs during Covid. Here's my attempt at Short Circuit from the same album: https://on.soundcloud.com/F5dcikiLb9RNQ4jC9 The chords at the end were
237.
by afro88 2y ago
Can a model even know that it used a hint? Or would it only say so if it was trained to say what parts of the context it used when asked? Because then it's statistically probable to say so?
238.
by afro88 2y ago
For 99.9% of people, the funny of the prank would only hit later. Ie, upon finding out it was a prank, and hearing about the "insert 5 cents" part that they probably didn't see with their own eyes. Plus the retraction, and 2n
239.
by afro88 2y ago
Thank you. This is the insidious thing about black box LLM coding.
240.
by afro88 2y ago
The linked promotional materials [0] say that they remote into a mac mini running Avid. > he works on iMac, which remotes into a separate Mac mini that runs Avid So the conjecture from the article that the mac mini isn't powerful

More ›