Dennis Hoelgaard Bal

Plotting multivariate distributions fails

2

I love how easy it is to plot a PDF with Distributions and Makie. The following just works ``` julia> using Distributions julia> using GLMakie julia> plot(Normal(1, 2)) ``` However,...

The following page describes Julia usage on NextJournal: https://nextjournal.com/nextjournal/julia-environment?change-id=CozcTBwNndNtXtFSJxxokb&node-id=39e3f06d-60bf-4003-ae1a-62e835085aef The page contains the following block, with the only output being the single line error shown: I am not sure where...

DSP.jl still exports fftfreq

2

Despite this note  it seems like this package still exports fftfreq, causing a colision when using both DSP.jl and FFTW.jl: ``` julia> using Pkg; pkg"update DSP"; using DSP Updating...

Order of polynomial coefficients

2

I am generally confused around weather one should provide polynomial coeficcients for the lowest order term, or the higest order term, first. It varies from API to API. But in...

Question: `prob` string macro examples do not work.

4

Question on: https://github.com/TuringLang/Turing.jl/edit/master/docs/_docs/using-turing/guide.md The guide uses the following model: ``` @model function gdemo(x, y) s² ~ InverseGamma(2, 3) m ~ Normal(0, sqrt(s²)) x ~ Normal(m, sqrt(s²)) y ~ Normal(m, sqrt(s²))...

Question on: https://github.com/TuringLang/Turing.jl/edit/master/docs/_tutorials/05_linear-regression.md This tutorial makes use of MLDataUtils, which is archived and now read-only. The readme for MLDataUtils states that "This package has been discontinued. Most functionalities have been...

No error from negative learning rates

20

In using baysean optimisation hyperparameter tuning, the learning rate was estimated to be negative at some point. This raised no error, as demonstrated in the example below: Example ``` julia>...

Warn on `NaN` loss

1

Would it not be convenient, particularily for users with less experienced with artificial neural networks, to print a warning when trying to update parameters with a loss of NaN? This...

Easy to make mistake with `gpu()`

4

All of the following uses of `gpu()` result in an error: ``` begin # Directly on the return from traindata xtrain, ytrain = MNIST.traindata(Float32) |> gpu xtest, ytest = MNIST.testdata(Float32)...

60 minute blitz not wokring - trainimgs and valimgs not defined

11

From the 60-minute blitz: ```julia using Statistics using Flux, Flux.Optimise using Images: channelview using Metalhead using Metalhead: trainimgs, valimgs using Images.ImageCore using Flux: onehotbatch, onecold, flatten using Base.Iterators: partition ```...