That sounds like a really messed up set of incentives. These are Norwegians, they have a right by birth to live in Norway and do business with their countrymen, and their government is saying "go find another tribe to join, we don't want you." That's a really fucked up thing to do, kick people out of their nation because they are financially better off than the average just to improve the government's quarterly report on wealth inequality.
AFAIK the Medieval Warm Period was indeed a time with high temperatures but this was mostly local (restricted to few areas), while the current increase in temperature is global and is also clearly linked to human activites
There are reports from the 90s where the medieval warm period was hidden conveniently. Where some politicians wanted the data manipulated to pull money from Siemens (this is just two examples of real, factual manipulation that happened, that you will find in the first link below).
Also, the models used were "toucehd" to fit wanted predictions, the models chosen also were (and still are I think) the most catastrophic.
I am NOT saying the climate change is not mostly due to humans (I think so though). What I am saying is that I have very reasonable doubts.
I guess, even if you vote negative, that I have reasons to at least doubt about this as a whole, given the high level of manipulation that there is around it, since what I see here is a huge effort to do an exercise of cherry-picking and bias confirmation: some scientists are discarded or hidden, the data is used in the most catastrophist way, manipulated at convenience, information omitted... it looks certainly suspicious to me.
* The medieval warm period was a great deal less global than many climate skeptic bloggers present it to be. It's never been "hidden" in any legit modelling where it's been relevant.
* The models were "touched" - I'm tempted to ask you show me on a globe where they were touched and how that made you feel .. but seriously; 1/ All gathered raw instrument data for whatever purpose is inevitably normalised, leveled, scaled, transformed in some manner or another for completely legitimate, numerically sound, physically sensible reasons, and 2/ "the (climate) models" are clusters of families of models that bracket parameters in order to tease out { worst case | best case } probability spread scenarios - this is normal and expected.
* ClimateGate ... ffs, seriously? Cherry picking internal emails to expose the worst parts that individuals say to other individuals is something that rips the lid of a giant climate conspiracy?? You don't actually by into that, btw, do you?
For what it's worth I've been in global scale exploration geophysics for energy and mineral resources for decades ... I've done all this big scale data work for the people least likely to support AGW theories .. and the AGW case is extremely sound.
We've significantly increased the insulation properties of the atmosphere at large and it's gradually causing more and more energy to remain trapped at the land|sea interface where hilarity ensures; climbing mean tempretures, more extreme storms, cold cells, and hot cells being driven by that surface layer heat engine, etc.
If you want to read, you have the links with the evidence tha manipulation on purpose happened sustained in time (and still happens!) such as making people out that contradicted the human warming theory, data manipulation, cherry-picking and replacing data at convenience. These are real things that have been happening, independently of our positions on the topic.
Take the time to read through if you are genuinely interested. You can use a translator. There are literal quotes of the mails sent back and forth.
Whether you are skeptical or not, that is a position I respect. Just understand that I have my doubts: it is a fact that climate changes.
For me what it is controversial is whether it is humans who do it mostly and I have a lot of reasons to think that people who expose a different opinion are systematically pushed out of the media and mainstream.
Not only that, there have been periods with 5 to 10 times more CO2, CO2 is used for bigger harvest (double to triple sometimes!), yet, until I took the time to investigate, it all looks headed towards one single idea: that CO2 is absolutely bad (and they are interested in taxing it of course because it is the only and true cause) and the rest is censored or half-censored. I suspect that the reason for this is getting money, since I am not sure one can conclude that the cause ofwarming is exactly CO2 and humans. They would lose their justification for taxing.
This makes me suspicious that there is a big interest in twisting the truth and I think it is reasonable to think so.
I am not saying I am right. Just that it looks like that to me.
Hope that, after linking (partial) evidence you understand why I do not believe official versions right away.
Obviously you know more about the raw topic than me. But there is nothing close to consensus as far as my understanding goes, even from experts.
I'm over sixty years old, I started in applied mathematic and geophysical modelling (gravity, magnetics, radiometrics) over forty years ago and have read along to the climate debate in scientific literature and the popular press all that time.
Claims of "data manipulation" are smoke and leaves tossed in the air by people that either don't get into the weeds or the very few that have reasonable points to ask (are stations affected by urban heat islands?) that have been addressed in detail.
> There are literal quotes of the mails sent back and forth.
And? Does any of that chatter amoung a few people at one institute in change any of the actual hard data and the interpretation tens of thousands of other qualified people about the globe have reached?
There's plenty to read about the Climategate fiasco, the breathless pearl clutching by AGW deniers screaming "Look! Look! See the way people have tried to bend the PR to warn us deniers! It's a smoking Gun I tell you!!", and the cold calm collected analysis pointing out that a bunch of cherry picked emails means fuck all against the actual case.
> Not only that, there have been periods with 5 to 10 times more CO2,
Sure .. a very long time ago .. and once the Earth had a largely molten surface also.
Right now in the timescale of humans spread across the planet raising crops and having numbers over a billion .. that all came about in a comparatively stable climate - and it is very clearly human activity that has changed the current atmospheric makeup causing mean trapped energy levels and the land|sea layer to rise.
> But there is nothing close to consensus as far as my understanding goes, even from experts.
Incorrect.
There is overwhelming consensus within the community of people that can read the papers with the exception of a very few outliers who also largely "agree" but with weird issues (Lindzen (Sr and Jr) et al).
The consensus within those qualified is covered here:
And then there's a whole body of self declared "experts" that flounder but otherwise make good money blogging and denying while being paid to do so by the usual suspects. You can follow that money outwards from the Koch brothers, Lord Monckton, et al.
Its understandable that people are confused by all this, there's been a good 50 years now of actively running public media campaigns to sow confusion and to question the science.
Mine presented as a very high fever along with loss of appetite and unwillingness to drink.
The first and last of those combined to create severe problems related to dehydration, such as when I briefly passed out in the kitchen and had an uncontrolled fall into the furniture there.
So my advice tends to be focused on that problem: force yourself to drink as much as you can (I was given a target of two liters a day which proved to be less than would have been ideal, but it was about all I could do to drink that much in a day), stock water or, better, sports drinks close to your bed so that you don't have to get up before rehydrating, and ideally have some people around you who can check on you and bring you things.
When I did not appear to be getting any better after three days, and a splitting headache persisted through a fairly aggressive schedule of tylenol and ibuprofen, I visited an emergency room and was given antibiotics. That's when I learned I had pneumonia.
Medical advice on loss of appetite was: don't worry about it; whether you eat or not doesn't matter at all (assuming you and your appetite recover on a reasonable time frame).
Back on topic, this is true in general of formerly healthy people. Lack of food will not cause any problems in the present, because you were not suffering from it in the recent past. And unless it persists a long way into the future, it also won't cause any problems in the future, because you will regain access to food and start eating again.
So true. I still own a Sony ereader and although the battery is a little worn off, the whole ease of use with epub or pdf is remarkable (its nearly 10 years old now). I really wonder why they stopped producing them
Regarding number 4) this seems very similar to the board game Stratego [1] where part of the squares can not be stepped on. Also you only see your opponents pieces when you attack!
Nice work! I was wondering if you noticed changes in the output coherence during training?
I fine-tuned it on the corpus of The Office quotes [1] and I noticed that a loss of around 0.9 gives me the most 'humorous' outputs. This may be subjective but I think for comedy the surprise plays a huge role and for longer training (and loss around 0.4) it feels overly unsurprising and therefore less funny. I also tried sampling with temperatures >1 but then it just goes crazy (e.g. some outputs are completely in Latin).
I get a lot of Latin and Spanish in mine but I think that's because they actually are represented in the poetry corpus. Not too surprising that the regular GPT-2s are also exposed to a lot of foreign language, as Reddit is not a strictly anglophone website, and that it'll remember despite some finetuning (there are so many parameters in it, after all).
I do look at the training samples but I've never noticed a worsening of 'coherence' in the samples, so to speak. I wonder if that what overfitting looks like? My PG corpus is so large that the GPT-2s struggle to converge, much less overfit, so I don't know what overfitting would look like. You could try using the new pseudo-validation loss checking feature nshepperd added to see if there's any connection between the validation loss and your perception of coherence.
Hi! Thanks for taking your time doing this. Are there projects in the intersection between deep learning and neuroscience? As mentioned there are people with neuroscience background but from the papers mentioned, I dont see something really neuroscience related.
GB (and Google more broadly) are very interested in this intersection. So yes, there are opportunities to do awesome work here, albeit actually working in a wet-lab is more difficult.
