Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

deepmind.google

Post reply on HN
Print view View original

131–140 of 143 posts

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#131

Post by yieldcrv »

tcumulus wrote: post #32

Everything in AI seems to be focused on LLMs lately. But in my opinion, powerful problem-specific models like this are even more interesting. The SOTA AI models used in weather forecasting are already outperforming the classic NWP models while being orders of magnitude more efficient (inference). Most are based on multi scale (hierarchical) Graph Neural Networks, an architecture which is not often talked about. The o…

LLM’s are an evolutionary deadend with the power and resource demands to make and run them for diminishing returns, but that may be okay as their own reasoning capabilities are big enough, and spur investment into the supporting infrastructure for them

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#132

Post by auspiv »

algo_trader wrote: post #37
trescenzi wrote: post #16

Earlier quoted context omitted.

If you’re just getting into this tropical tidbits[0] is my go to for more raw data. Less pretty than zoom earth but also an interesting place to see what the models are predicting on each of their runs which is then interesting to compare to actual forecast guidance. 1: https://www.tropicaltidbits.com/

I am getting into cyclone predictions (for maritime scheduling) Is there a basic/freemium resource for past events? Mostly just very coarse spatial/temporal maps of past events

https://www.nhc.noaa.gov/

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#133

Post by counters »

pseudocoup wrote: post #129

call me a cynic, but in the age of government underinvestment in science, isn’t this how google could actually change the world?

No, because 100% of the data that Google trains their models on comes from federal research investment.

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#134

Post by vickychijwani »

PunchTornado wrote: post #100
vickychijwani wrote: post #89

Earlier quoted context omitted.

Are you being sarcastic? Even if “Google is struggling so badly” (which it really is not - the narrative will flip again at some point), these efforts will have a lasting impact on the world. Not everything good is about bringing in revenue.

shareholders are pretty unhappy about demis. think about alphafold. huge investments from the company, tens of billions. at a critical time. and absolutely 0 revenue. it got demis a nobel though. as a shareholder you'd be unhappy too.

This line of thought strikes me as very short-term. I’m already a shareholder and employee of the company, but that’s beside the point.

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#135

Post by nbardy »

fcanesin wrote: post #10

Maybe was this that was the last drop for Sundar. Demis: "I have a new amazing breakthrough" Sundar: "Great! We really need a answer to Sol and Fable" Demis: "They are completely owned in typhoon forecasting"

he got promoted to chief scientist and them being an expert at all modeling will really help compared to being llm only

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#136

Post by vasco »

purplemoonx wrote: post #130
vasco wrote: post #56

Earlier quoted context omitted.

> what the slope of the beach is today (many people don't even know this changes drastically daily and why it is important So why is it important? As far as I know the slope changes AFTER the weather not before as a prediction mechanism but happy to learn

It changes constantly on a daily cycle and also on a larger seasonal cycle. Slope defines the water depth and what kind of beach it is - long flat beach means shallow water, steep beach means deep water - it affects if you can you fish there, route lagoon systems and seasonal floodwater, do sports like surfing and skimboarding which are highly dependent on it to the point you can only do it some times of day only par…

Ok but none of that is what was mentioned. I lived near the sea all my life I'm aware of winter vs summer differences in slope and waves in different beaches. You were saying that if for example your 24h case happened, the weather around that beach would be affected. What I know of and what you describe are not that.

You were saying something like, if the slope changes the weather will change. But ended up saying many things can change the slope. The order is reversed.

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#137

Post by Angostura »

dylan604 wrote: post #78
Angostura wrote: post #72

Earlier quoted context omitted.

30% more time to know exactly where to board up etc. seems very significant

Wouldn't it be 50% more time?

Yes. My incompetence in maths strikes again. I knew it wasn't right as I typed it

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#138

Post by xnx »

WarmWash wrote: post #60
tcumulus wrote: post #32

Everything in AI seems to be focused on LLMs lately. But in my opinion, powerful problem-specific models like this are even more interesting. The SOTA AI models used in weather forecasting are already outperforming the classic NWP models while being orders of magnitude more efficient (inference). Most are based on multi scale (hierarchical) Graph Neural Networks, an architecture which is not often talked about. The o…

Rumor is that part of the disruption at GDM these past few months also involved people not wanting to be bound to strictly LLM research.

Interesting. Both the Gemini 3.5 Pro delay and the staff shakeup?

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#139

Post by slashdave »

tcumulus wrote: post #32

Everything in AI seems to be focused on LLMs lately. But in my opinion, powerful problem-specific models like this are even more interesting. The SOTA AI models used in weather forecasting are already outperforming the classic NWP models while being orders of magnitude more efficient (inference). Most are based on multi scale (hierarchical) Graph Neural Networks, an architecture which is not often talked about. The o…

Um... cofolding?

Re: DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

#140

Post by purplemoonx »

vasco wrote: post #136
purplemoonx wrote: post #130

Earlier quoted context omitted.

It changes constantly on a daily cycle and also on a larger seasonal cycle. Slope defines the water depth and what kind of beach it is - long flat beach means shallow water, steep beach means deep water - it affects if you can you fish there, route lagoon systems and seasonal floodwater, do sports like surfing and skimboarding which are highly dependent on it to the point you can only do it some times of day only par…

Ok but none of that is what was mentioned. I lived near the sea all my life I'm aware of winter vs summer differences in slope and waves in different beaches. You were saying that if for example your 24h case happened, the weather around that beach would be affected. What I know of and what you describe are not that. You were saying something like, if the slope changes the weather will change. But ended up saying man…

No, all I’m saying is weather apps should be able to show the slope of the beach that day (without having to go there and measure it.

Maybe a symbol like this:

    \
With

    Beach advisory. Shorebreak inclines can cause dangerous riptides.
Or even being able to predict that it will rain in Hayes Valley San Francisco 10 days from now. That sounds like magic, but I think it’s possible with enough training data.

Then you asked for clarity on the slope of beach part - the main thing is it’s a metric that matters to a lot of people and it would be useful if weather apps can accurately forecast it.

I think maybe with AI it can. That’s all really

Post reply on HN