dmos62

4 ms·

1.
by dmos62 6d ago
I disagree. Parallelization, coordination shouldn't be model-level functionality. Further, how do you know if or when we'll get orders of magnitude increases in context sizes. Or, do you mean that the labs will just nail the perfe
2.
by dmos62 7d ago
Yes. I want reproducability, open-sourcing, accessibility, correctness, and most of all: usefulness. I don't care how it was written or reviewed, as long as some assurances regarding above things can be made, and I don't see why L
3.
by dmos62 8d ago
That's legal?
4.
by dmos62 8d ago
I commend your understanding. I would have been losing my mind with rage.
5.
by dmos62 8d ago
There's definitely something to be said about not trying to spin stories into something more. And, I'm usually on the other side of the barricade in these discussions. But, when I'm reading about apples, it's me being go
6.
by dmos62 8d ago
An excuse to say "apple detectives"? Choice of words drew me in, and I'm happy it did. This is quirky news; it's making a bit of geekery interesting to distracted geeks, like me.
7.
by dmos62 8d ago
SpecKit and Superpowers do have significant overlap, in that they can orchestrate work, diverge into parallel streams and reconverge, etc. But, SpecKit is about making spec changes as safe and explicit as possible. OpenSpec is a very lightw
8.
by dmos62 9d ago
You can't make that generalization. What you describe will be different city to city, not to mention country to country. Further, privacy and safety regulations (which are EU-wide) will not be available in US, independent of income. Th
9.
by dmos62 9d ago
Until LLM context sizes explode by two orders of magnitude, I cannot envision significant agent programming without leaning heavily on spec-driven workflows. It's also worth noting that SDD is such a wide variety of approaches that the
10.
by dmos62 9d ago
Checkout SpecKit and SpecDD too. There is a significant variety of spec-driven development approaches.
11.
by dmos62 10d ago
> as long as we don't need to be involved in their regulatory or legal regimes EU has such a bad name in these circles. As a EU citizen, I cherish these same regimes.
12.
by dmos62 17d ago
I'm describing reinforcement learning.
13.
by dmos62 17d ago
LLMs are trained to obey instructions, and they try their best to game the reinforcement learning by including reports of how they're obeying your instructions. Therefore, not talking about followed instructions is a sort of conflict f
14.
by dmos62 17d ago
I have found that asking the agent why it did X or Y is an essential part of iterating on its harness. It was non-obvious to me, probably because I'm used to having to reverse-engineer why a system does something, but that's not u
15.
by dmos62 21d ago
It would make sense to have this integrate with a self-learning routine for an agent: e.g. at night it looks through what it did and writes NPC-Forge recipes. Tomorrow it can answer queries (which he turned added to NPC-Forge) without an LL
16.
by dmos62 22d ago
What is happening to your batteries? I'm on something like year 6 on my Redmi and my battery is fine. Am I simply lucky?
17.
by dmos62 25d ago
> 'Why not both' is an argument, but clear as day what Apple does is worse. There's no alternative. 'Just start somewhere, anywhere, please' is another argument.
18.
by dmos62 25d ago
AI does make me suck more, which is a good thing. It sped me up and has me iterating on things in 3 days that normally would have taken me more than a month to do. Surprise, surprise: I'm discovering that I have a lot of blind spots an
19.
by dmos62 27d ago
Everything south of Poland is Southern Europe now.
20.
by dmos62 29d ago
I wonder if immersion in fiction (interactive or otherwise) is lower nowadays, given the rise of dopamine-saturating short-form content like scrolling. I definitely enjoyed more long-form content (like the mentioned games) 10-20 years ago.
21.
by dmos62 1mo ago
Sounds like your old, random clients might be a problem?
22.
by dmos62 1mo ago
I disagree. An editor can definitely make a piece of writing clearer without further input from the author. I often use an LLM to do that. Of course, I didn't mean to imply that there's a dichotomy between personality and clarity.
23.
by dmos62 1mo ago
Spent 2 minutes skimming, didn't manage to find a single talking point. I've no idea what the thesis is. Seemed like a list of things this person built. People talk shit about LLM writing. Well, here's a prime example of why
24.
by dmos62 1mo ago
Judging by the comments here, Ox Alpha routes to multiple models from different vendors. A tactic to make identification harder?
25.
by dmos62 1mo ago
> Comes across cheap Given this is open-source, have you considered contributing a more expensive-feeling landing page? Or, donating to finance the project?
26.
by dmos62 2mo ago
At the same time, the experiment doesn't corroborate this take at all. If it were a mix of different "unexpected events", then you might have an argument there. As the experiment is, I'd say, it would have been better to
27.
by dmos62 2mo ago
New understanding, you have added to mine, young padawan. PS: fascinating insight.
28.
by dmos62 2mo ago
Ehem, we already do that. USB-C is an umbrella term and everyone has a homebaked organizing system to tell which USB-C cable can do what, or a homebaked testing routine where we run through 5 cables each time we need to connect something. T
29.
by dmos62 2mo ago
I don't recall running into ads, don't know about trackers. PlantNet hasn't triggered any red flags for myself.
30.
by dmos62 2mo ago
Better than PlantNet?

More ›