Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror

Comment Evaluating an unreleased model ... (Score 2) 12

ycombinator.com says:

The news here is that gpt2-chatbot is confirmed to be an anonymized, unreleased model. See yesterday's discussion here: https://news.ycombinator.com/i... The https://chat.lmsys.org/ page now currently says this:

> *gpt2-chatbot is currently unavailable.* See our model evaluation policy [here].

That link takes you to https://lmsys.org/blog/2024-03... , which is this post's URL. The page was edited yesterday (when gpt2-chatbot was added) to include this section:

> Evaluating unreleased models: We collaborate with open-source and commercial model providers to bring their unreleased models to community for preview testing.

> Model providers can test their unreleased models anonymously, meaning the models' names will be anonymized. A model is considered unreleased if its weights are neither open, nor available via a public API or service. Evaluating an unreleased model consists of the following steps:

> 1. Add the model to Arena with an anonymous label. i.e., its identity will not be shown to users.

> 2. Keep it until we accumulate enough votes for its rating to stabilize or until the model provider withdraws it.

> 3. Once we accumulate enough votes, we will share the result privately with the model provider. These include the rating, as well as release samples of up to 20% of the votes. (See Sharing data with the model providers for further details).

> 4. Remove the model from Arena.

Comment Silly (Score 1) 2

Canada, lacking First Amendment,

This is profoundly ignorant. "Lacking first amendment (of the US constitution?)" What are you even talking about. Are you implying that Canadians don't have freedom of speech in our constitution? So ignorant. But this is just silly:

In the past Bill C-16 criminalized misgendering.

No it didn't.

Comment Re:Results from five different groups (Score 2) 114

The average warming per decade over the last two (since 2004) has been: (source)

Land station measurements:
BEST: slope = 0.24C per decade
GISS: slope = 0.32C per decade

Satellite measurements:
UAH: slope = 0.23C per decade
RSS: slope = 0.26C per decade

Roughly one Celsius every four decades if the rate of increase doesn't continue to accelerate. I'm somewhat less worried about hitting 2C than I am about the velocity we'll have achieved when we do.

Comment Re:shouldn't every year be a new record? (Score 2) 114

shouldn't every year be a new record?

Not quite. There are a number of oscillations superimposed upon the long term warming trend. For example, here are a few of the factors that contribute to global average temperature. Add those together (plus volcanic activity and other factors) and you get a steady upward trend, but with oscillations that bring any given year somewhat higher or lower than that ever rising mean.

Submission + - "Artificial creativity" music software for Commodore Amiga unearthed (breakintochat.com)

Kirkman14 writes: Josh Renaud of breakintochat.com has recovered two early examples of "artificial creativity" software for the Commodore Amiga that generate new music by recombining patterns extracted from existing music.

Developed by cartoonist Ya'akov Kirschen and his Israeli software firm LKP Ltd. in 1986-87, "Computer Composer" demo and "Magic Harp" were early attempts at AI-like autonomous music generation. Kirschen's technology was used to help score a BBC TV documentary in 1988, and was covered by the New York Times and other major newspapers. None of the Amiga software was ever sold, though the technology was ported to PC and published under the name "The Music Creator" in 1989.

Comment Re:Details [Re:Rabid fans incoming] (Score 2) 227

It's a fair point, but I don't see too much to quibble with in the Slashdot summary - except perhaps "If annual averages reach above 1.5 degrees Celsius, the effects of global warming could become irreversible"

There's nothing particular about 1.5C. The warming that has been locked in due to emissions to date is very close to the amount of warming that has occurred to date (ATTP). That has been true for a while and will remain true beyond 1.5C.

Comment Re:There's a better design out there.... (Score 1) 203

It is the only thing that matters to humans. As I pointed out to the other guy, this is what we're talking about. We already know climate change is bad. We are talking about something more specific. Stick to the topic.

As it happens, I am a human. For me, a livable climate is more important than all the money in the world.

Comment Re: We have no choice who we punish/reward (Score 3, Informative) 347

Free will is often thought to mean: "Could we have made a different choice under the exact same circumstances?"

The choices we make are deterministic, based on the input parameters. If we ran the simulation 100 times with the exact same parameters, we would make the exact same choice each time.

Comment Re:Sample size of 1 - no control group. (Score 1) 2

Even more curiously, the peer reviews are public.

The reviewers called out the fact that the focus was too narrow and recommended revision.

In response, the authors acknowledged that there are various factors, including "changes in human population distribution, fuel breaks, land use, ignition patterns, firefighting tactics, forest management strategies, and long-term buildup of fuels," but that "Accounting for changes in all of these variables and their potential interactions simultaneously is very difficult. This is precisely why we chose to use a methodology that addresses the much cleaner but more narrow question of what the influence of warming alone is on the risk of extreme daily wildfire growth."

Comment Sample size of 1 - no control group. (Score 1) 2

This is a bit of a strange one. All Nature did was publish a paper that this author wrote. A paper that he says he stands by. They didn't reject the paper he wishes he'd written.

It seems to me that this confession tells more about the author than it does about Nature.

I'm not sure that he's wrong. Possibly the paper he wished he'd written would have been rejected. Maybe even for the reasons he imagines. But other than the authors 'feels', there's nothing to show that this is the case.

Comment wildfire insurance is estimated to see costs rise (Score 1) 226

The statement by State Farm doesn't mention climate at all, though it does look like there's a link between climate and wildfires. According to the California Climate Change Assessment of 2019:

By 2100, if greenhouse gas emissions continue to rise, one study found that the frequency of extreme wildfires would increase, and the average area burned statewide would increase by 77 percent. In the areas that have the highest fire risk, wildfire insurance is estimated to see costs rise by 18 percent by 2055.

A Fourth Assessment review of forest health literature provides further scientific backing to the State’s Forest Carbon Plan to increase forest restoration and treatment, such as prescribed fire, to an average of 35,000 acres a year by 2020.

Slashdot Top Deals

Pascal is a language for children wanting to be naughty. -- Dr. Kasi Ananthanarayanan

Working...