EricLeer

Joined 5 years ago

Karma

Submissions

Member since

May 7

https://twitter.com/LeerEric

View on Hacker News

Recent submissions

Comment · 1 year ago

Jeah, the lab where I worked cleaned the machine on a 1-2 month basis. All the cleaned out metals where then sold to a recycler.

Comment · 1 year ago

This will probably wildly diver from country to country, but typically in elementary school where 50 kids, and for high school around 100-120.

Comment · 2 years ago

Yes there are many attempts in applying a transformers to timeseries forecasting. For instance (but there are many more): - Timegpt https://arxiv.org/abs/2310.03589 - Chronos https://github.com/amazon-science/chronos-forecasting

These kind of papers often talk the world, but often lack a proper baseline model. They only compare against very simple (naive forecast), or non tuned models. In my experience a gradient boosting model will probably solve 95% of your forecasting problems, and trying to get fancy with a transformer (or even just a simple neural net) is more trouble then it is worth.

Comment · 2 years ago

Funnily enough next to wind also solar is one of the major renewable sources in the Netherlands. So also day-night storage would be important.

The Netherlands actually is second place in terms of solar generation per capita in the world (only Australia has more).

Comment · 2 years ago

Any recommendations for a good AV/EPP for desktop linux (ubuntu)?

Comment · 3 years ago

The weather company claims to do this (they are also the main provider of weather data for apple).

Comment · 3 years ago

See for instance the pytorch geometric [1] package, which is the main implementation in pytorch. They also link to some papers there that might explain you more.

[1] https://pytorch-geometric.readthedocs.io/en/latest/

Comment · 3 years ago

I am in the power forecasting domain, where weather forecasts are one of the most important inputs. What I find surprising is that with all the papers and publications from google in the past years, there seems to be no way to get access to these forecasts! We've now evaluated numerous of the ai weather forecasting startups that are popping up everywhere and so far for all of them their claims fall flat on their face when you actually start comparing their quality in a production setting next to the HRES model from ECMWF.

Comment · 3 years ago

The wave length and pulse length aren't the same thing though. The short pulsed lasers actually consists of light that have many different wave lengths in them!

Comment · 3 years ago

Not really sure what you are looking for, but the easiest might be to just add lags of your input variables in the same linear model that you are using.

If you are looking for an actual timeseries method I would checkout either darts [0] or statsforecast [1]. They are currently the most mature timeseries packages.

[0] https://unit8co.github.io/darts/ [1] https://github.com/Nixtla/statsforecast

Comment · 3 years ago

Why not just startup simcity 4 then?

Comment · 3 years ago

I wonder if it would be possible to probe what a model is trained on by usage of prompts the reply to which can only be answered well with certain training data.

For instance if I have some body of text that can't be found elsewhere on the internet, if the reply of the model references the information in that text in some way you may be fairly certain it was used in training.

The hard part is probably finding such a body of text.

Comment · 3 years ago

Have been self-hosting this on a raspberry pi for the last 2 years, and it has been completely painless. It just keeps chugging along without needing to take a look.

Comment · 3 years ago

I would actually be fine with this. There is nothing special to the American social networks, they will just be replaced by European alternatives.