alach11

10 ms·

91.
by alach11 2y ago
> I mean -- if nobody's surprised, then nobody has learned anything, and then what was the point of doing all this work? I strongly disagree with this view on science. It's extremely valuable to scientifically validate prior as
92.
by alach11 2y ago
This case is interesting because Luigi doesn't fit the normal profile of a shooter - he had almost everything going for him. He graduated from a private high school as valedictorian. He went to an Ivy League college to major in Compute
93.
Suspect in CEO Killing Withdrew from a Life of Privilege and Promise(nytimes.com)
10 points by alach11 2y ago|4 comments
94.
by alach11 2y ago
There's no reliable source that this is his manifesto, and it contains a number of details that conflict with what we know about him. I'd recommend waiting to post and discuss something like this until it's confirmed.
95.
by alach11 2y ago
> How many parameters? What kind of corpus was it trained on? It's rare for the leading model providers to answer these questions. As someone who applies these models daily, I agree with the dead comment from meta_x_ai. Your questio
96.
by alach11 2y ago
There are some recent foundation models pre-trained on time-series data. For example TimesFM from Google. Of course, it's not directly built for classification, and it's meant for univariate datasets, so it would take some work to
97.
by alach11 2y ago
When I teach people how to play a game, I usually omit several of the rules upfront to help people grok the goal faster. For example when teaching Texas Hold'em, we might play all the way to the river with all cards face up and no bett
98.
by alach11 2y ago
It kind of feels criminal to do time-series analysis with multimodel models and not use any traditional numerical models to provide a baseline result. It's an interesting result though.
99.
Unlocking the power of time-series data with multimodal models(research.google)
131 points by alach11 2y ago|26 comments
100.
by alach11 2y ago
That's awesome. We do the exact same thing for prize eligibility on my work leaderboard. The whole point is for it to be fun and challenging. No need to grind to the end unless you want to.
101.
by alach11 2y ago
I just do the puzzles until it stops being fun. For me that's usually around day 14.
102.
by alach11 2y ago
Yeah Elon described it as a "harder" and "faster" landing. So maybe things were a little outside of the safe parameters, necessitating the change in plans. I wonder if we'll get to find out what that was.
103.
SpaceX Super Heavy splashes down in the gulf, canceling chopsticks landing(twitter.com)
272 points by alach11 2y ago|425 comments
104.
by alach11 2y ago
I was being genuine.
105.
by alach11 2y ago
Wow what a fantastic little article. Thanks for writing and sharing that.
106.
Tyson vs. Paul: Netflix suffers significant issues on fight night(usatoday.com)
7 points by alach11 2y ago|4 comments
107.
by alach11 2y ago
Am I the only one who can't stand the "LLM writing smell" on articles like these? There are obvious formatting tells, and then less obvious word choice tells (e.g., delve), and lastly an excess verbosity and genericness that
108.
by alach11 2y ago
My team and I also develop with these models every day, and I completely agree. If models stall at current levels, it will take 10 (or more) years for us to capture most of the value they offer. There's so much work out there to automa
109.
by alach11 2y ago
Exactly. And getting back to the original poster's comment "the CO2 starts underground and ends up underground"... that assumes there are no leaks anywhere in the process.
110.
by alach11 2y ago
That's not correct. If pressure was all that mattered, we'd just run compressors on nitrogen (or formation gas). CO2 has properties that make it especially favorable for flooding. https://en.wikipedia.org/wiki&#x2F
111.
by alach11 2y ago
This seems like something that could be handled easily with a second-pass on the data using an LLM. And the author has made the dataset available... [0] [0] https://www.titledrops.net/
112.
by alach11 2y ago
I think authenticity is being hugely underrated as a factor for why Trump won. People inherently trust someone who is visibly flawed and speaks off-the-cuff. This preference for authenticity has always existed, but is extra strong as a reac
113.
by alach11 2y ago
In an age of inauthenticity on social media, people are inherently drawn to someone who appears authentic. Trump comes across as a straight-shooter. People may not love everything he says, but they feel like they can trust him because he is
114.
by alach11 2y ago
> What really saves London are the vast number of remarkably large parks and tree cover. I don't disagree that shade is a good idea. But what really helps London is it's vastly milder climate [0] caused by the Gulf Stream. [0]
115.
by alach11 2y ago
This is largely in anticipation of the devastating drop in revenue if Google can no longer pay them to be the default search engine. That's where they get 86% of their revenue [0]. [0] https://fortune.com/2024/08&#
116.
by alach11 2y ago
> There will always be a good offering who doesn't water mark There's a possible future where this gets legislated, right? Of course, there are lots of implementation challenges to this and it's probably a bad idea...
117.
by alach11 2y ago
A casual look at the source shows the architecture won't allow the attacks you're talking about. Since each request runs separately, there's no way for prompt injection on one request to influence a future request. Same thing
118.
by alach11 2y ago
This is a common workflow for these sorts of problems. I've done similar a few times. The downside is the additional cost.
119.
by alach11 2y ago
I have several use cases where I realize ~75% cost reduction by using prompt caching. It's not clear to me while reading this - does the Batch API stack with prompt caching?
120.
by alach11 2y ago
It's great to see the arms race between OpenAI and Anthropic benefitting end users. I have several use cases where I realize ~75% cost reduction by using prompt caching. It's not clear to me while reading this - does the Batch API

More ›