HackerTrans
TopNewTrendsCommentsPastAskShowJobs

vdelpuerto

no profile record

Submissions

Opus 4.7 vs. Sonnet 4.6

3 points·by vdelpuerto·2 bulan yang lalu·5 comments

More Than Skills, Frameworks

github.com
1 points·by vdelpuerto·2 bulan yang lalu·0 comments

[untitled]

1 points·by vdelpuerto·3 bulan yang lalu·0 comments

comments

vdelpuerto
·2 bulan yang lalu·discuss
[dead]
vdelpuerto
·2 bulan yang lalu·discuss
[dead]
vdelpuerto
·2 bulan yang lalu·discuss
Not really, although I appreciate the sarcasm. What I'm saying is that Anthropic and other providers tells you that you can rely on lower models for certain tasks that are usually daily tasks that don't require much more than basic understanding. To prepare a quotation doesn't seem like something that requires Opus 4.7 at max effort, don't you think? So the question is: what counts as a task you can delegate to a lower model and still trust the output?
vdelpuerto
·2 bulan yang lalu·discuss
Thanks. I'll take it a look
vdelpuerto
·2 bulan yang lalu·discuss
I was trying to look options outside the box (everything is more context or RAG) and been using this approach for about a month with good results. https://github.com/VDP89/fscars
vdelpuerto
·2 bulan yang lalu·discuss
honest question here, why there are lot of comments about SEO is dead?
vdelpuerto
·2 bulan yang lalu·discuss
That is the same fight the 2D animators were having with 3D aninmation 30 years ago. The resolution is likely to be the same: the tool wins but the fundamentals stay, and the line between competent and incompetent practitioners moves but does not disappear.
vdelpuerto
·2 bulan yang lalu·discuss
I'm still waiting something more specific or groundbreaking too. Feels like a lot of noise with just the goal to get people to talk about it. And now I realize I am talking about it and about nothing at the same time. Just fugazzi.
vdelpuerto
·2 bulan yang lalu·discuss
through agent Harness the user can get a kind of model personalization as it iterates, you adjust patterns, mistakes, KB and preferences. Its the closest we have to add some kind of personality to our LLM agents. mistakes are the hardest one because you put the rule in AGENTS.md or system prompt and the model still goes back to the wrong pattern. context alone don't fix it
vdelpuerto
·2 bulan yang lalu·discuss
Under the premise of low taxes and Itaipú's great energy generation, Paraguay has an open and political interest in bringing data centers and AI centers to the country looking for capital injection to the economy, but with very poor infrastructure and grid base development to face this. Just looking at recent studies and articles about what consuming energy is doing to the grid in the US, it raises the question of how we can prevent this from happening here.
vdelpuerto
·2 bulan yang lalu·discuss
I wrote something about it trying to look other way around the context or memory data in models. The gravitational pull of information stills very hard to manage. Ive been using "functional scars" about 30 days now and getting good results in repetitive mistakes across sesions. https://github.com/VDP89/fscars
vdelpuerto
·2 bulan yang lalu·discuss
and not only feelings, but failures with real consequences is what makes us more creative and aware of our actions.
vdelpuerto
·2 bulan yang lalu·discuss
funny thing, local government in Paraguay is importing a model that USA itself recognizes to not fully understand. Paraguay is proposing to accommodate projects whose demand exceeds the available supply
vdelpuerto
·2 bulan yang lalu·discuss
Pretty nice. Workflow?
vdelpuerto
·2 bulan yang lalu·discuss
Although I don't come from programming per se, but from the civil engineering area, AI has been a real boost and an addictive tool to dismember and try to understand at its core. It tends to make you believe you are above your own capacities and that everything now will be easier.

But the truth is, at least in my case, that by trying to keep myself humble and really testing the barriers and limitations of these tools, it makes me realize that it is not only that the AI is limited in its capacities, but myself. And trying to run through the rabbit hole at the fastest pace possible, I realized I was getting brain burn (difficult to sleep, to talk, to join ideas; I was getting more irritable).

So the first thing I did was delete the coding app from my phone. I try to keep the coding and investigating stuff to office hours. I'm going outside, getting back to the gym, to sports, and not reading or watching any YouTube or podcasts on the subject on Sundays, etc. It's still very hard, to be honest. But clearing my mind makes me see new approaches in a clearer way.
vdelpuerto
·2 bulan yang lalu·discuss
Mmm....His position of "stopped using markdown altogether for almost everything" strikes me as overstated and selfpromotional. The 2-4x higher generation cost is non-trivial when you're operating under rate limits, and Git versioning friction is a real problem for your work in big and complex repos. I use the rule "HTML for deliverables, Markdown for infrastructure" and it seems more defensible than HTML maximalism.
vdelpuerto
·2 bulan yang lalu·discuss
it already feels a huge difference in plan usage since yesterday. Last month I spent x3 times my usuall consume.
vdelpuerto
·2 bulan yang lalu·discuss
Not a computer engineer here, in civil engineering and running our own software in production. In our process we document AI mistakes, decide which ones become scars of the system, put them outside the model and trigger them as hooks at every new session. At session close the system checks for new hooks and whether the scars were activated.
vdelpuerto
·2 bulan yang lalu·discuss
Agree. Human review and iteration. Hooks to gate certain decisions you dont want to happen again, like functional scars.
vdelpuerto
·2 bulan yang lalu·discuss
Someone above already pointed at the move from coding toward architecture. That matches my experience. I had a short stint in CS that I didn't finish, and most of my work is in another engineering domain. With AI tools I spend more time on architecture and judgment and less on syntax, and that's where I actually add value. No product I build gets past prototype without solid architecture and human intervention where real judgment is needed. But there are people who aren't programmers and still understand systems and criteria well. Those people can now ship products that actually work. That doesn't mean every non-programmer should ship; it means the floor for who can build moved. Human capabilities will keep evolving around the tool. It doesn't have to be either pessimistic and restrictive or optimistic and permissive. It's like that Rattaoutille phrase from Anton Ego "not everyone can become a great artist, but a great artist can come from anywhere".