euphetar

6 ms·

1.
by euphetar 2mo ago
Honestly I think we will get something in the middle. A lot of stuff will become API calls, but we will also have a lot of people-facing interfaces, and the need to use them automatically (for QA at the very least least). And then there are
2.
by euphetar 2mo ago
I get the point, but you can't always use the API. 1. Presumably you want to trust your agent to do no shady sheningans behind your back when you give it a simple task 2. Sometimes it just doesn't work. https://osworld-
3.
by euphetar 2mo ago
Sorry for the confusion, Steelman CTO here. OSWorld (even the old ones, v1 and verified) runs the model in a full linux VM. And using the terminal is allowed. The models are given tools that provide a subset of possible UI actions. They the
4.
by euphetar 2mo ago
Agreed and a lot of companies seem to have their head in the sand, just hoping AI agents will magic their way through the legacy But I don't think RPA is the way. Or at least, not in the current state. I think AI agents with proper com
5.
by euphetar 4mo ago
Honestly, a lame release of mediocre models. I was most excited about the "frontier tuning." Like, it will actually watch you do stuff and learn to do it for you? That would be actually interesting. But no, it's just a data l
6.
by euphetar 5mo ago
I wouldn't call it a benchmark since it's just one sample. They do highlight a real problem, though. Computer use is immature right now and far behind language agents Try playing fruit ninja via text and llm toolcalls though
7.
by euphetar 5mo ago
No link to edits or specific article. Disappointed. This shouldn't be front page as it contains no substance besides speculation
8.
by euphetar 6mo ago
People don't like having their time wasted
9.
by euphetar 6mo ago
Appreciate the effort, but this is very basic and nothing you need the source code to understand. I was expecting a deep dive into what specific decisions they made, but not how an loop of tool calls works
10.
by euphetar 1y ago
Nice
11.
by euphetar 2y ago
While I do not have much sympathy for Altman, the article is very low quality and contains zero analysis Yeah, maybe on the surface chatbots turned out to be chatbots. But you have to be a poor journalist to stop your investigation of the i
12.
by euphetar 2y ago
the tidbit didn't work though, I just tried with GPT-4o
13.
by euphetar 2y ago
It's 100% clear why his I initiative was buried. The guy is a reckless egomaniac jerk. He is very transparent that he would do illegal stuff on a whim, target civillians, escalate conflict with other nations without even consulting a s
14.
by euphetar 2y ago
Please provide a feature to load data from notion export
15.
Startup Stock Options: The Shortest Complete Guide for Employees(borisagain.substack.com)
10 points by euphetar 2y ago|0 comments
16.
by euphetar 2y ago
The scary part is that perhaps there are in fact LLM generated bot threads, but they are not so easily busted
17.
by euphetar 3y ago
Auqui-fire?
18.
by euphetar 3y ago
Planet Farms | Onsite, Remote Hybrid | Portugal (Coimbra) | Full-time | ML Engineer, Data Engineer We make vertical farms. We are looking for two people to join our data team. For the Data Engineer we expect enough proficiency to help us bu
19.
by euphetar 3y ago
Planet Farms | Data Engineer/Analyst | Office in Coimbra, Portugal We make vertical farms. Hybrid office job, Coimbra, Portugal. Relocation, assistance of lawyers at the expense of the company. The task is to found the data team and gr
20.
Trying All the Alternatives to ChatGPT(borisagain.substack.com)
1 points by euphetar 3y ago|0 comments
21.
Intro to Bayesian A/B Testing(borisagain.substack.com)
1 points by euphetar 3y ago|0 comments
22.
by euphetar 3y ago
Planet Farms | Data Engineer/Analyst | Office in Coimbra, Portugal We make vertical farms. Hybrid office job, Coimbra, Portugal. Relocation, assistance of lawyers at the expense of the company. The task is to found the data team and gr
23.
Understanding the Normal Distribution for Real(borisagain.substack.com)
1 points by euphetar 3y ago|1 comments
24.
by euphetar 4y ago
Assuming web app. 1. Python. 2. Flask. Pure server-rendered HTML, no fancy frontend. Avoid JS as long as I can, just use jinja templates. If my app is not going to work with good old HTML and CSS chances are it's not going to work at a
25.
by euphetar 4y ago
No way to assign probabilities? I prefer metaculus
26.
by euphetar 4y ago
Very shallow article. Would like to see a list of mentioned "recent breakthroughs" about using hashes in ML besides the retrieval applications, because this is genuinely interesting
27.
by euphetar 4y ago
So many people comment about how Snowden can't say things like this because he is in Russia. I wonder: how does it invalidate the facts? He links to Wikipedia. Maybe you don't want to believe Snowden, him being a traitor. Don&#x27
28.
by euphetar 4y ago
I am currently also playing around with this. The best part is that for storage you don't need to store the reconstructed image, just the latent representation and the VAE decoder (which can do the reconstructing later). So you can sto
29.
by euphetar 4y ago
My uninformed take for what it's worth. Let's assume your projections are right and cannibalism is unavoidable like you say. You will never sell any of this technology. There is no need for it until there is a nuclear winter, so n
30.
Privacy and speed: The story behind our face blurring solution(medium.com)
2 points by euphetar 4y ago|0 comments

More ›