Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
ilaksh
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
26 ms
·
451.
▲
by
ilaksh
1y ago
The solution is advanced cryptocurrency. Obviously. Almost no point in writing any comments on this site that use that word unfortunately.
452.
▲
by
ilaksh
1y ago
1. What hardware for that. 2. Can you do a benchmark?
453.
▲
by
ilaksh
1y ago
Where do you put the 480 GB to run it at any kind of speed? You have that much RAM?
454.
▲
by
ilaksh
1y ago
To run the real version with the bench arks they give, it would be a nonquantized non distilled version. So I am guessing that is a cluster of 8 H200s if you want to be more or less up to date. They have B200s now which are much faster but
455.
▲
by
ilaksh
1y ago
Its hard for me to get past the name. It's like naming a ceramics manufacturer "punch brittle".
456.
▲
by
ilaksh
1y ago
Can you explain? I'm going to ignore your arrogant decision because I am genuinely curious. What is it about Israeli "settlements" that makes it different from stealing land?
457.
▲
by
ilaksh
1y ago
I remember my first year in college as being the time when I solidified my own first worldview. Prior to that, I had some ideas like the existence of God (in some form) that I was ambivalent about or maybe deferring final judgement. That&#x
458.
▲
by
ilaksh
1y ago
I didn't say I thought it was objectively righteous. I didn't say anything about my worldview. It seems objective that many civilians have been killed and a lot of land stolen. But my point is that things that seem objective to on
459.
▲
by
ilaksh
1y ago
They have a worldview that is so different it is effectively an alternate reality. This mostly comes from seeing different information streams or being in different social circles.
460.
▲
by
ilaksh
1y ago
Seems like renewable fuels of some type could make it much more realistic to move away from fossil fuels. Even hydrogen could work if people really wanted to make it work. But surely there is a way to make ammonia, ethanol, biodiesel, or so
461.
▲
by
ilaksh
1y ago
Yes, prompt caching helps a lot with the cost. It still adds up if you have some tool outputs with long text. I have found that breaking those out into subtasks makes the overall cost much more reasonable.
462.
▲
by
ilaksh
1y ago
That's obviously broken but part of this is an inherent difficulty with names. One thing they could do would be to have a default question that is always present like "what other people named [_____] are there?" That wouldn&#
463.
▲
by
ilaksh
1y ago
I think it's a fair article. However I will just mention a few things. When you make an article like this please take note of the particular language model used and acknowledge that they aren't all the same. Also realize that the
464.
▲
by
ilaksh
1y ago
https://github.com/runvnc/tersenet
465.
▲
by
ilaksh
1y ago
Watch the video?
466.
▲
by
ilaksh
1y ago
I think Gemini is a step in the right direction for some things. I usually mention my "tersenet" ideas when I see Gemini. Now that we have the Gemini LLM and Claude etc. there is less excuse for me to not finish any real software
467.
▲
by
ilaksh
1y ago
Some is marketing, but it's not _just_ marketing. Many people have a worldview now where AI progress is inevitable. So we really believe it I would be interested to hear other ideas or plans that don't involve AI progress. My prem
468.
▲
by
ilaksh
1y ago
There are Ethereum, Algorand, many alternatives with low fees.
469.
▲
by
ilaksh
1y ago
You were using a weak LLM then. The difference between one of the leading edge LLMs like Gemini 2.5 Pro, o3, or Claude 4 and an average LLM or one you can run on your typical PC/laptop is night and day.
470.
▲
by
ilaksh
1y ago
Very cool way to play classic games. I am more into form over function though. So the projects that are more novel are more interesting to me in a way. Like Commander X16.
471.
▲
by
ilaksh
1y ago
The game generator is not an LLM.
472.
▲
by
ilaksh
1y ago
For the skeptics, think about how bad the video generation models were last year versus today. Besides optimizations, the biggest difference between those models and todays are more training and clever tweaks. A lot of core concepts are the
473.
▲
by
ilaksh
1y ago
Perspectives on this may shift as AI video generation continues to improve. Also I think this intersects with the efficacy of the country's social safety net. What is that like in Sweden?
474.
▲
by
ilaksh
1y ago
What I suggested was to use an actual agent. I also did not say there was no human in the loop. The process is encoded in natural language and tool options. I'm not YOLOing anything.
475.
▲
by
ilaksh
1y ago
I didn't say to skip any kind of problem modeling. I just didn't emphasize it. The goal _should_ be to avoid doing traditional software engineering or create a system that requires typical engineering to maintain. Agents with lead
476.
▲
by
ilaksh
1y ago
I think this was true late 2023 or early 2024, but not necessarily in mid 2025 for most tasks (as long as they require some AI and aren't purely automation) and you use SOTA LLMs. I used to build the way most of his examples are just f
477.
▲
by
ilaksh
1y ago
This isn't really a problem with video generation or AI in general. Sure, there is an aspect of ragebait to it, but the reality is that racism is extremely widespread. If it were not, this kind of content would not be so popular. The p
478.
▲
by
ilaksh
1y ago
AI isn't destroying anything. Don't blame the technology for what humans do with it. AI should allow every student to have personalized instruction and tutoring. It should be a massive win. If everyone instead of taking advantage
479.
▲
by
ilaksh
1y ago
It would be cool to get a follow up on how long it's been since this write up and how well it's been doing since they revised the prompts and tools. Anyone know someone from Andover Labs?
480.
▲
by
ilaksh
1y ago
It said they had a few tool commands for note taking.
More ›