adamwespiser/deep-learning-koans

A series of educational Deep Learning Koans, using Julia and Flux.jl

deep-learning-koans's Issues

Typos and mistakes in Notebook #2

Typos:

• Multi-dimsional Array assignment
• matrixes
• diaganols
• Multi-dimsional Assignment
• multideminsional
• multidemisional (not sure if it is a running joke at this point)
• Indexies
• to a an array
• experssion
• whre
• in julia we can read a file in, line by line (capitalization and useless "in,")

Mistakes:

• @show a doesn't do much in the first code cell.
• "Initialize Arrays w/ Zero, Ones": already done in first notebook.
• The koans about "Random" don't work, for the same reason as in the first notebook.

Random seed koan not working

I changed the seed to 42 as indicated in the source file, but still fail the test. I'm guessing Julia has changed its randomness mechanism between 1.2 (the one you assume) and 1.5 (the one I use), so not sure an issue is in order.

using Random
aboutkoan = "set the myseed equal to something else, besides 1"
myseed = 42
Random.seed!(myseed)
@test rand(1:10^20) == 5590813852184710016

Test Failed at In[10]:5
  Expression: rand(1:10 ^ 20) == 5590813852184710016
   Evaluated: 2353793361580022217 == 5590813852184710016
There was an error during testing

Stacktrace:
 [1] record(::Test.FallbackTestSet, ::Union{Test.Error, Test.Fail}) at /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.5/Test/src/Test.jl:737
 [2] do_test(::Test.ExecutionResult, ::Any) at /Users/julia/buildbot/worker/package_macos64/build/usr/share/julia/stdlib/v1.5/Test/src/Test.jl:520
 [3] top-level scope at In[10]:5

Google Colab doesn't recognize Julia 1.2

I wasn't able to run the koans using the Google Colab link.

Environnement d'exécution "julia-1.2" non reconnu ; "python3" attribué par défaut
Translated: Execution environment "julia-1.2" not recognized ; "python3" used instead by default

Typos and mistakes in Notebook #1

I am scared of using git with notebooks, so I'm just going to post them here.

Typos:

• the concent of using IPython Notebeoks
• When you see a MethodError, it's a huge clue that you are (finishes in the middle of the sentence)
• resahpe
• evaluetes
• defintion

Mistakes:

• In "An Overloaded Function Koan", returns are reversed (Float64 returns 0 and Int64 returns 1)
• and their second dimension equal to 4. (third)
• We can create random arrays using the zeros or ones function (We can't)
• The struct in the last koan is useless and quite confusing. Why not just @test c == 55?