machiaweliczny

5 ms·

1.
by machiaweliczny 2mo ago
> It is horrible until AI gets as good as a perfect project from a prompt or we need to abandon it. Getting close to it with my recent test of cursor cloud workflow. I've spent 300M tokens in a day and it achieved what I wanted. Man
2.
by machiaweliczny 4mo ago
Good news is that HTML in canvas might bring back these cool days :)
3.
by machiaweliczny 4mo ago
Yes it's still bad there's no viable headless UI in browser one can really style and it has all the a11y etc. but need extra library for selects that work etc. Invented work for no good reason. The real complexity is in diversity
4.
by machiaweliczny 4mo ago
At current rates you are losing when renting (excluding possible appreciation), that's why it's hard to sell and easy to rent. Due to low rates before homes went big in value cause you were earning even not counting appreciation,
5.
by machiaweliczny 4mo ago
The thing with home is that it's leveraged investment for normal people. You can pay 10% of value and resell later. Buying home for cash is less optimal but for credit it makes sense usually due to leverage and little downside risk. Pl
6.
by machiaweliczny 5mo ago
Tested and it's good. Fast version is bad though. I like planning model in Cursor that it works more like human written design doc instead of too detailed AI plan. Seems like this is more responsible for results that model but still on
7.
by machiaweliczny 5mo ago
The practical case is less time spend on rebasing/formatting code - IMO formatting standard is very helpful as then it's the same as storing AST. There's also better preserving of git blame and that's likely why they hav
8.
by machiaweliczny 5mo ago
They would compete with Meta then which already does something like this but has mature AD tech
9.
by machiaweliczny 7mo ago
Rookie mistake, you just conflating concepts pretending things don't exist. Good API is one that is minimal, explicit and acomplishes the job without conflating with somethings else. Congrats now I don't know if this variable is d
10.
by machiaweliczny 7mo ago
I personally go 3x for gym classes like "healthy spine", "mobility", "core" etc. and then 2-3x hard training. But I would say I am very active recently. Do I feel better? Yes. Was it hard first 2 weeks? Yes, I
11.
by machiaweliczny 7mo ago
5.2 and 5.3 are strong/best for coding, 5.0 and 5.1 were garbage
12.
by machiaweliczny 8mo ago
We need that for this chinese 3B model that think 45s for hello world but also solves math.
13.
by machiaweliczny 8mo ago
A lot of NLP tasks could benefit from this
14.
by machiaweliczny 8mo ago
No it's not because cost is much lower. They do some kind of speculative decoding in monte-carlo way If I had to guess as humans do it this way is my hunch. What I mean it's kinda the way you describe but much more efficient.
15.
by machiaweliczny 8mo ago
Windows Phone was actually good. I would even say that my Lumia something was one of best experiences ever on mobile. G+ was also good. Efficient markets mean that you can "extract" rent, via selling data or attention etc. not rea
16.
by machiaweliczny 8mo ago
--help explains this, -s for sandbox and -a for approvals
17.
by machiaweliczny 8mo ago
For now until they are in the lead
18.
by machiaweliczny 8mo ago
Claude: you get rate-limited with one prompt so hard to validate 4.6 Codex: better with rate-limits, 5.2 strong with logic problems Cursor: cursor auto - a bit dumb still but I use the most for writing not really thinking, it's also go
19.
by machiaweliczny 8mo ago
I think this will work but requires quality data pipelines and scaffolding same as coding
20.
by machiaweliczny 8mo ago
Yeah, I think people have had enough this time its not gonna fly
21.
by machiaweliczny 8mo ago
Man this is art
22.
by machiaweliczny 9mo ago
Yeah, that's my current gripe but I think this just needs some good examples in AGENTS.md (I've done some for hooks and it kinda works but need to remind it). I need good AGENTS.md that explain what good abstraction boundary is an
23.
by machiaweliczny 9mo ago
But detailed specs exists for both HTML and JS and tests also exists and unlimited amount of test data. You can just try running webpage or program and also have reference implementations - it's much easier for agents to understand tha
24.
by machiaweliczny 9mo ago
Yeah, seems like this could be achieved by using https://github.com/streamich/memfs/blob/master/docs/snapshot... Weird they don't use it - might backfire hard
25.
by machiaweliczny 9mo ago
BTW does anyone know of good assistant voice stack that's Open Source? I used https://github.com/ricky0123/vad for voice activation -> works good, then just using Web Speech API as that's the fastest and t
26.
by machiaweliczny 9mo ago
I tried Kokoro-JS that I think runs in browser and it was too way too slow with latency also not supporting language I wanted
27.
by machiaweliczny 10mo ago
> Potential accidental data leakage from the “future” Exactly. Makes no sense with models like grok. DeepSeek also likely has this leak as was trained later.
28.
by machiaweliczny 11mo ago
AFAIK the whole morphology is decided in distributed computation fashion via electrical potential changes, at least according to experiments by Michael Levin
29.
by machiaweliczny 11mo ago
Yeah compared to animals we have a lot of extra bootstrapping outside of physics/chemistry alone via culture and stored information similar to how cell DNA bootsraps via physics, human mind boostraps via stored information in human &qu
30.
by machiaweliczny 11mo ago
I can see a perl comeback

More ›