fsndz

25 ms·

241.
by fsndz 2y ago
I get it. But how many people travel long-term every year? I usually get 4 weeks and split that between 2-3 locations, both in and out of the country. And I don't necessarily want to cook a lot on vacation (I guess I am not alone?), so
242.
by fsndz 2y ago
Can't the government already do the same using images from identity cards, passports, etc.? Is the problem mainly because it's a private company doing it? I genuinely want to better understand this issue.
243.
by fsndz 2y ago
useful for one shot cases, but not more for the moment imo.
244.
by fsndz 2y ago
Finishing is super important. Just focus on completing a version 0. Then, you can improve it if you feel like it. It doesn't have to be perfect, just finished under a reasonable timeframe. Not finishing and endlessly moving from one pr
245.
by fsndz 2y ago
anyone ever found a job with this ?
246.
by fsndz 2y ago
I couldn't understand what this is for and why it is useful. Gotta work on that documentation to make it more clear.
247.
by fsndz 2y ago
Regarding Boeing, it's true that the principal-agent problem and misaligned incentives are issues. However, you must also consider the impact of promoting business school graduates and McKinsey-type consultants—who often believe they a
248.
by fsndz 2y ago
I frankly don't know why people keep using Airbnb now. It's costly, and the service is worse than a hotel. My worst vacations have happened in Airbnbs. Even for group vacations with friends, we now choose hotels. The good Airbnbs
249.
by fsndz 2y ago
you have a lot of resources about similar problems on the internet. so LLMs will have some patterns to leverage even if it is not even sure if you will get correct answers.
250.
by fsndz 2y ago
The reversal curse is mostly about understanding specific logical relationships rather than identity statements. It's not about statements like 'Joe is human' implying 'human is Joe,' but rather about LLMs strugglin
251.
by fsndz 2y ago
Each argument could easily fill a book. Some valuable insights are discussed in the linked article, but most of the evidence comes from a series of research papers demonstrating that LLMs perform poorly when confronted with out-of-training
252.
by fsndz 2y ago
The chain of thought that is apparent in the conversation is what's really interesting and what OpenAI exploits. OpenAI does not have to evaluate the quality because the conversation itself is already valuable. Whether the conversation
253.
by fsndz 2y ago
Typically, when people aren't satisfied with an answer, they continue asking questions or prompting further, which makes the whole exchange valuable even without the use of a thumbs-up or thumbs-down button. That's the secret weap
254.
by fsndz 2y ago
and sometimes they generate two answers and the same time and users have to choose one
255.
by fsndz 2y ago
I don't know the history of the original tribes in Brazil. Why? Because I haven't studied it, read about it, or been taught anything on the subject. I know that such tribes exist, but I don't know their history.
256.
by fsndz 2y ago
you get an AI, you get an AI, you get an AI, but all of you won't get any AGI: https://www.lycee.ai/blog/why-no-agi-openai
257.
by fsndz 2y ago
> And yet I can provide modern LLMs unique never before asked scenarios that require the ability to reason about real world phenomenon and they can respond in ways more thoughtful than the average person. Give me an example of that
258.
by fsndz 2y ago
There are no emergent behaviors; LLMs are essentially memorizing statistical patterns in the data and using lexical cues to generate responses. They cannot explain themselves reliably because they don't know what they know, nor do they
259.
by fsndz 2y ago
We all know the saying: "If it's free, you're the product." I believe the free version of ChatGPT is primarily a data acquisition tool that OpenAI uses to fine-tune its models and conduct reinforcement learning from huma
260.
by fsndz 2y ago
The Machine is free. Karpathy is Harold Finch.
261.
by fsndz 2y ago
The thing is, the killer app of the generative AI space—at least for large language models (LLMs)—might already be ChatGPT. While people are still searching for the next big application in this field ( https://www.lycee.ai/bl
262.
by fsndz 2y ago
Elastic trying to reclaim the lost ground in their original territory by re-embracing open source is a bold move. They might be hoping to attract developers and businesses back by offering more transparency and community involvement. Could
263.
by fsndz 2y ago
Context windows are becoming larger and larger, and I anticipate more research focusing on this trend. Could this signal the eventual demise of RAG? Only time will tell. I recently experimented with RAG and the limitations are often surpris
264.
by fsndz 2y ago
some people are just addicted to unnecessary complexity
265.
by fsndz 2y ago
can't you just design your database so as to be able to do that independently ?
266.
by fsndz 2y ago
I have to agree here. Feels like pure marketing/selling blogpost. I was not convinced.
267.
by fsndz 2y ago
not being a criminal is really good, I don't have to worry about any of these stuff
268.
by fsndz 2y ago
Data is great, only if you always think critically about it. The problem is, for some, data has been elevated to the status of a new religion. "If the data says it, it must be true." They refuse to see how biased any data could be
269.
by fsndz 2y ago
The premise of using an LLM to interpret data assumes that LLMs can reliably reason over data, which is not always the case. Essentially, you're trusting a black box to analyze health data and blindly trusting the output. This doesn&#x
270.
by fsndz 2y ago
A bit clickbait, but I really learned a lot from reading this. Specifically, when dealing with large transaction amounts, it's probably better to opt for wire transfers instead of using Stripe or other payment processors.

More ›