padolsey

15 ms·

211.
by padolsey 2y ago
Yikes you're right. I need to come up with a way to solve that. Maybe a param or path to lock in the username like abc.at.hn/Abc ... I'll have a go. EDIT: for now I'm tired and off to bed but if someone has a graceful&#
212.
Show HN: Markdown HN profiles at {user}.at.hn(at.hn)
459 points by padolsey 2y ago|154 comments
213.
Tipping AI for Better Responses?(blog.j11y.io)
2 points by padolsey 2y ago|0 comments
214.
by padolsey 2y ago
Yes, server-rendering definitely helps, though I have suspicions about its compiled outputs still being very heavy. There's also a lot of CSS frameworks that have an inline-first paradigm meaning there's no saving for the browser
215.
by padolsey 2y ago
What an incredibly naive and dismissive thing to say.
216.
by padolsey 2y ago
Whether you like it or not, over 15% of the world's population lives in China.
217.
by padolsey 2y ago
A lot of this resonates. I'm not in Antartica, I'm in Beijing, but still struggle with the internet. Being behind the great firewall means using creative approaches. VPNs only sometimes work, and each leaves a signature that the f
218.
by padolsey 2y ago
I find it useful. I'm on a pretty limited mobile data plan. Imagine others might be too. Though I wish there was a way my browser could warn me.
219.
by padolsey 2y ago
I get the hate, but I find this semi-interesting. I flew business recently and the airline had a '3d' walkthrough of their cabin (i.e. 360deg photos at multiple points), a bit like property market listings where you have to press
220.
by padolsey 2y ago
Cool! Feature req: a way to 3d-skew the screenshot.
221.
by padolsey 2y ago
https://lobste.rs
222.
by padolsey 2y ago
The thing is, such prophecies are all very wrong until they're very right. The idea of an LLM (with capabilities of e.g. <1 yr away) being given access to a VM and spinning up others without oversight, IMHO, is real enough. Biases l
223.
by padolsey 2y ago
They know, as well, that opt-in simply wouldn’t give them the scale they’d need for meaningful training data. They’re being very intentionally self interested and unconcerned with their customers best interests.
224.
by padolsey 2y ago
It seems really clever, especially the encoding of a signature into LLM token probability selections. I wonder if synthid will trigger some standarization in the industry. I don't think there's much incentive to tho. Open-source g
225.
by padolsey 3y ago
Do you have pay-as-you-go pricing? I feel that’s always missing from things like this. Cool otherwise.
226.
by padolsey 3y ago
Yeh agreed. Function calling FTW— just need a bit more reliability/(semi-?)-idempotence.
227.
by padolsey 3y ago
It’s hard to communicate about this stuff. I think people hear ‘tools’ and ‘function calling’ and assume it provides an actual suite of tools or pre-made routines that it calls upon on the Anthropic backend. But nope. It’s just a way of gen
228.
by padolsey 3y ago
WHAT LOL. You're on hacker news and you're complaining that people are hacking?! Literally the entire tenet is playing around and experimenting for our own edification and fun. On August 25, 1991, Linus Torvalds posted a message o
229.
by padolsey 3y ago
Biggest bugbear is that 90% of the time, non-technical people are just looking for a CRUD app. Basic user auth, chat, forum, mobile app. Etc. The age-old scaffold. It's just dull. I'm a technical person looking for a non-technical
230.
by padolsey 3y ago
Location: UK, able to work Asian timezones only Remote: Yes Willing to relocate: No but open to travel Technologies: JavaScript, React, Neo4j, Node.js, Next.js, LLMs, Generative AI Résumé/CV: See https:/&#x
231.
by padolsey 3y ago
Why does this feel like an AI wrote it? And the whole website feels the same. The OP is here commenting as well and, though polite, just gives off an AI vibe. Really off.
232.
by padolsey 3y ago
2c: I guess it's a "wait and see" approach. The AI Bill is making good progress through EU parliament and will soon (?) be enforceable. It includes pretty hefty fines (up to 7% annual turnover for noncompliance) and lots of c
233.
Tipping AI for Better Responses?(blog.j11y.io)
2 points by padolsey 3y ago|0 comments
234.
by padolsey 3y ago
Best approach is just to do an initial call to an LLM to classify and filter user inputs, and then after that you can safely send it along to your main agent.
235.
by padolsey 3y ago
This unthinking encoding of thresholds feels like a 'Systems' problem that percolates into almost every part of life, whether its income brackets or hard limits on how many books you can borrow from the library. Some make more sen
236.
by padolsey 3y ago
Yeh it’s weird they don’t mention parameter size or other reasoning metrics. It’s a very cool approach to getting structured output from an LLM, but the benchmarks don’t show us the whole picture. I’m wondering if their approach can be used
237.
by padolsey 3y ago
The follow-up's better imho: https://digitopoly.org/2024/03/07/after-the-gold-rush/ E.g. On biases in LLM responses and lack of explainability: > Believe it or not, there is a historical preceden
238.
by padolsey 3y ago
I don’t think you’d be fine-tuning a whole model in such cases. That seems over the top, no? I assume you’d get sufficiently far with big context windows, vector search, RAG. Etc.
239.
Learning to Decode Collaboratively with Multiple Language Models(arxiv.org)
2 points by padolsey 3y ago|0 comments
240.
by padolsey 3y ago
Ooh intriguing. I’m learning mandarin at the moment, and also happen to have some sodium valproate lying around for my epilepsy. I’m not currently taking it but I could give it a go and report back.. :D

More ›