Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
luckydata
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
8 ms
·
121.
▲
by
luckydata
1y ago
Gemini is notoriously bad at tool calling and it's also widely speculated that 3.0 will put an emphasis on fixing that.
122.
▲
by
luckydata
1y ago
This is dumb. Llms are a tool, a very useful one. Bad PRs should be rejected always no matter the source, but banning a tool because some people can't use it is not what engineering is about.
123.
▲
by
luckydata
1y ago
what are the advantages vs something like https://caddyserver.com ?
124.
▲
by
luckydata
1y ago
no, that factory makes a ton of cars. had teething problems like every other car factory in the world. source: my significant other worked at tesla for a decade, now at Rivian.
125.
▲
by
luckydata
1y ago
There's a great Kurz gesagt video about it. Whales get the least cancer! https://youtu.be/1AElONvi9WQ?si=rhZjgpkORBTC0D8r
126.
▲
by
luckydata
1y ago
you work for the wrong people
127.
▲
by
luckydata
1y ago
speaking to a child at their level is the best way to keep them from speaking well. I never did it with my son and it didn't hold him back one bit. Everyone remarks his incredible vocabulary and language skills for his age. IMHO holdin
128.
▲
by
luckydata
1y ago
In an immersion context I can get to conversational fluency in about 3 months and to complete mastery in a year. I've done it twice, once for Spanish and once for English. A few things in my approach helped me move quite a lot faster t
129.
▲
by
luckydata
1y ago
new models are always magical, let's see how it feels after the cost cutting measures get implemented in 2-3 months.
130.
▲
by
luckydata
1y ago
your pricing is nuts. let me know if revenue are good, I'll copy you.
131.
▲
by
luckydata
1y ago
does this approach actually work? is it more reliable than plain mcp calls?
132.
▲
by
luckydata
1y ago
I downloaded the app and it failed at the first screen when I set up the models. I agree with the spirit of the blog post but the execution seems lacking.
133.
▲
by
luckydata
1y ago
yeah but it's hard for others to know what you're up to, you force everyone else to do investigations and waste time. Just open a ticket and be a good team mate.
134.
▲
by
luckydata
1y ago
have you ever witnessed how sometimes Gemini makes multiple attempts at writing a file only to give up and start chanting "I'm worthless...". That's tool use failure :)
135.
▲
by
luckydata
1y ago
Gemini is great, it's just incredibly clumsy at tool use and that's why it fails so often in practice. I'm looking forward to the next version, it will for sure address it, it's a big issue internally too (I'm a rec
136.
▲
I made a simple harness for AI-assisted coding
(github.com)
2 points
by
luckydata
1y ago
|
1 comments
137.
▲
by
luckydata
1y ago
I put together a small harness to keep AI-assisted coding under control. It’s basically a set of git hooks and scripts that enforce simple things AIs tend to skip—running tests, checking basic metrics, that sort of guardrail. I think it’s u
138.
▲
by
luckydata
1y ago
This is a fundamental misunderstanding of how LLMs work.
139.
▲
by
luckydata
1y ago
Local models can do embedding very well, which is useful for things like building a screenshot manager for example.
140.
▲
by
luckydata
1y ago
Do you have some kind of tooling to automate the process? Would like to try it.
141.
▲
by
luckydata
1y ago
I'm over simplifying here but graph database and knowledge graphs exist. An LLM doesn't need to preserve everything in context, just what it needs for that conversation.
142.
▲
by
luckydata
1y ago
Because that's not the job it was designed to do, and you would know by reading the article.
143.
▲
by
luckydata
1y ago
The article says it's not a good conversational model but can be used for data extraction and classification as two examples.
144.
▲
by
luckydata
1y ago
How many times do we need to hear "exercise and healthy eating" prolong life and improve cognitive function?
145.
▲
by
luckydata
1y ago
is there some sort of utility to automatically backup and retrieve dotfiles from github to keep different computers synchronized?
146.
▲
by
luckydata
1y ago
if a kid is being lazy there's simply no way around "cohercion" as you put it. You know how I know he's being lazy? Because I used to do the same stuff for the same reasons, and my parents and teachers saw through it and
147.
▲
by
luckydata
1y ago
The crowd you see is because options are dwindling faster than people willing to use them.
148.
▲
by
luckydata
1y ago
As an immigrant to the USA teaching in this country is a mess. Teachers apply a lot of semi scientific mumbo jumbo to justify a completely inadequate amount of work required from students to learn. I know it's not popular to say it but
149.
▲
by
luckydata
1y ago
but why?
150.
▲
by
luckydata
1y ago
and who wants that exactly?
More ›