Alexej Gossmann

> Ran model in fp16 and runs! Seems to hover around 15gb (standard, 16fp). Not scientific test though haha. My experience is similar: the (standard, 16fp) model runs for me...

@lokeshh > I am not familiar with dense/non-dense representations. What does Yale and List matrix representation mean? Is it the same as non-dense representation? Is Yale and List matrix a...

> I don't know if anyone uses the ruby interface, maybe @agisga, since he submitted #346 I have used `#permute_columns!` in the unmerged pull request #336 only. That's when the...

I just ran a regression model with the same data in R, since it's nowadays the standard in statistical computing. They seem to have the same solution as the one...

I think that maybe one way to start would be to look at how things are done in other tools such as R and Python. If it helps, I recently...

Interesting... I'll take a look sometime this week, and let you know if I can figure something out...

@lokeshh I just had another look at your benchmark code, and I noticed that you generate data from a _perfectly_ linear equation. When you generate the data according to a...

@lokeshh Thanks for trying it out! I guess the problem lies somewhere else. I will try to find some time and write a GLM solver with nmatrix-lapacke from scratch to...

Okay. So, I have written a simple gradient descent algorithm from scratch to fit GLMs with NMatrix. Here is the (quick and dirty) code: https://github.com/agisga/gradient_descent/blob/master/examples/logistic_regression.rb The benchmarks don't look good...

It seems to me that the algorithm is theoretically okay, because it gives correct results most of the time. Maybe it fails because it accumulates numerical error quickly, when the...