Failure on MacOS for Julia 1.3, 1.4: Intel MKL Error #212
Labels
Comments
|
Adding some considerations: commenting out the clustering tests is not enough, the regression and classification tests also fail horribly; there's something seriously wrong here. and more of the same I'm wondering whether there's not a clash between Conda versions somehow |
|
We should compare
cc @OkonSamuel |
|
Last commit ok: 61febaa on Feb 26 First commit nok: d9944ca on Feb 28 List of differences: Maybe relevant
Probably irrelevant
|
|
The openspecfun calls the compilersupportlibs, note that this change was made on February 26 which is also the date where we had our last working push... let's try putting a compat bound on that |
|
Ok this is almost certainly the culprit, you can see that the compat of that thing is Julia 1.3 which would also explain why it fails on 1.3/1.4... now where the heck is this loaded |
|
Steps to fix the problem:
in your terminal, go wherever that is, for me
Log (@v1.5) pkg> activate .
s Activating environment at `~/.julia/dev/MLJModels/Project.toml`
(MLJModels) pkg> status
Project MLJModels v0.9.3
Status `~/.julia/dev/MLJModels/Project.toml`
324d7699 CategoricalArrays v0.7.7
b4f34e82 Distances v0.8.2
31c24e10 Distributions v0.22.6
a7f614a8 MLJBase v0.12.3 `~/.julia/dev/MLJBase`
e80e1ace MLJModelInterface v0.2.1 `../MLJModelInterface`
6f286f6a MultivariateStats v0.7.0
efe28fd5 OpenSpecFun_jll v0.5.3+1 `~/.julia/dev/OpenSpecFun_jll`
bac558e1 OrderedCollections v1.1.0
d96e819e Parameters v0.12.0
ae029012 Requires v1.0.1
321657f4 ScientificTypes v0.7.1 `../ScientificTypes`
2913bbd2 StatsBase v0.32.2
bd369af6 Tables v1.0.3
b77e0a4c InteractiveUtils
37e2e46d LinearAlgebra
44cfe95a Pkg
9a3f8284 Random
10745b16 Statistics
[ Info: Precompiling MLJModels [d491faf4-2d78-11e9-2867-c94bc002c0b7]
[ Info: Model metadata loaded from registry.
[ Info: Precompiling RDatasets [ce6b1742-4840-55fa-b093-852dadbb1d8b]
[ Info: Training Machine{AffinityPropagation} @ 1…25.
Test Summary: | Pass Total
AffinityPropagation | 5 5
[ Info: Training Machine{AgglomerativeClustering} @ 1…41.
Test Summary: | Pass Total
AgglomerativeClustering | 4 4
[ Info: Training Machine{Birch} @ 9…06.
Test Summary: | Pass Total
Birch | 8 8
[ Info: Training Machine{DBSCAN} @ 9…63.
Test Summary: | Pass Total
DBSCAN | 4 4
[ Info: Training Machine{FeatureAgglomeration} @ 1…75.
Test Summary: | Pass Total
FeatureAgglomeration | 7 7
[ Info: Training Machine{KMeans} @ 2…15.
Test Summary: | Pass Total
KMeans | 6 6
[ Info: Training Machine{MiniBatchKMeans} @ 9…15.
Test Summary: | Pass Total
MBKMeans | 6 6
[ Info: Training Machine{MeanShift} @ 9…80.
Test Summary: | Pass Total
MeanShift | 5 5
[ Info: Training Machine{OPTICS} @ 5…86.
Test Summary: | Pass Total
OPTICS | 4 4
[ Info: Training Machine{SpectralClustering} @ 1…33.
Test Summary: | Pass Total
SpectralClustering | 4 4
Pass Total
SVM-infos | 7 7
Test Summary: | Pass Total
SVCs | 3 3
Test Summary: | Pass Total
SVRs | 3 3
Test Summary: | Pass Total
ARD | 3 3
Test Summary: | Pass Total
BayesianRidge | 3 3
Test Summary: | Pass Total
ElasticNet | 3 3
Test Summary: | Pass Total
ElasticNetCV | 3 3
Test Summary: | Pass Total
Huber | 3 3
Test Summary: | Pass Total
Lars | 3 3
Test Summary: | Pass Total
LarsCV | 3 3
Test Summary: | Pass Total
Lasso | 3 3
Test Summary: | Pass Total
LassoCV | 3 3
Test Summary: | Pass Total
LassoLars | 3 3
Test Summary: | Pass Total
LassoLarsCV | 3 3
Test Summary: | Pass Total
LassoLarsIC | 3 3
Test Summary: | Pass Total
LinReg | 5 5
Test Summary: | Pass Total
OMP | 3 3
Test Summary: | Pass Total
OMPCV | 3 3
Test Summary: | Pass Total
OMPCV | 3 3
Test Summary: | Pass Total
PassAggr | 3 3
Test Summary: | Pass Total
RANSAC | 3 3
Test Summary: | Pass Total
Ridge | 3 3
Test Summary: | Pass Total
RidgeCV | 4 4
Test Summary: | Pass Total
SGDReg | 3 3
Test Summary: | Pass Total
TheilSen | 3 3
Test Summary: | Pass Total
MTLassoCV | 4 4
Test Summary: | Pass Total
MTLassoCV | 3 3
Test Summary: | Pass Total
MTElNet | 3 3
Test Summary: | Pass Total
MTElNetCV | 3 3
Test Summary: | Pass Total
LogRegClf | 8 8
Test Summary: | Pass Total
LogRegCVClf | 8 8
Test Summary: | Pass Total
PAClf | 7 7
Test Summary: | Pass Total
PerceptronClf | 7 7
Test Summary: | Pass Total
RidgeClf | 7 7
Test Summary: | Pass Total
RidgeCVClf | 7 7
Test Summary: | Pass Total
SGDClf | 13 13
Test Summary: | Pass Total
GPRegressor | 3 3
Test Summary: | Pass Total
GPClassif | 8 8
Test Summary: | Pass Total
AdaBoostReg | 3 3
Test Summary: | Pass Total
BaggingReg | 3 3
Test Summary: | Pass Total
XTreeReg | 3 3
Test Summary: | Pass Total
GBReg | 3 3
Test Summary: | Pass Total
RFReg | 3 3
Test Summary: | Pass Total
AdaboostClf | 8 8
Test Summary: | Pass Total
BaggingClf | 8 8
Test Summary: | Pass Total
GradBoostClf | 8 8
Test Summary: | Pass Total
RForestClf | 8 8
Test Summary: | Pass Total
XTreeClf | 8 8
Test Summary: | Pass Total
LDA | 8 8
Test Summary: | Pass Total
QDA | 8 8
Test Summary: | Pass Total
DummyReg | 4 4
Test Summary: | Pass Total
DummyClf | 7 7
Test Summary: | Pass Total
GaussianNBClf | 8 8
Test Summary: | Pass Total
KNNReg | 5 5
Test Summary: | Pass Total
KNNClf | 8 8
Test Summary: | Pass Total
BernNBClf | 7 7
Test Summary: | Pass Total
MultiNBClf | 7 7
Test Summary: | Pass Total
ComplNBClf | 7 7This is on a Mac with v1.5 |
Closed
|
@tlienart looking at the METADATA file it seems OpenSpecFun_jll.jl is a depency only used by Distributions.jl and SpecialFunctions.jl |
|
I think possibly RMath? But either way Elliott has commented on a similar issue in the BinaryBuilder package, seems like |
|
I tried to reproduce this locally, and I haven't been able to so far. I installed |
|
Thanks for trying this Elliott, it may well be that the error now only shows on Travis (cf link in the other issue) though I’m a bit surprised if that’s really the case. My environment is far from pristine but I’ll try creating one from scratch and testing on my side. Either way I think it would be good to make sure we can control this so that no one encounters undue segfaults ! Thanks again! |
|
@staticfloat using MLJ
X, y = @load_boston;
train, test = partition(eachindex(y), .7, rng=333);
@load LGBMRegressor
mdl = LGBMRegressor()
mach = machine(mdl, X, y)
fit!(mach, rows=train)
@load ARDRegressor
mdl = ARDRegressor()
mach = machine(mdl, X, y)
fit!(mach, rows=train)
PyError ($(Expr(:escape, :(ccall(#= /Users/AZevelev/.julia/packages/PyCall/zqDXB/src/pyfncall.jl:43 =# @pysym(:PyObject_Call), PyPtr, (PyPtr, PyPtr, PyPtr), o, pyargsptr, kw))))) <class 'numpy.linalg.LinAlgError'>
LinAlgError('unrecoverable internal error.')
File "/Users/AZevelev/.julia/conda/3/lib/python3.7/site-packages/sklearn/linear_model/_bayes.py", line 577, in fit
sigma_ = update_sigma(X, alpha_, lambda_, keep_lambda, n_samples)
File "/Users/AZevelev/.julia/conda/3/lib/python3.7/site-packages/sklearn/linear_model/_bayes.py", line 562, in update_sigma
X[:, keep_lambda].T))
File "/Users/AZevelev/.julia/conda/3/lib/python3.7/site-packages/sklearn/externals/_scipy_linalg.py", line 99, in pinvh
s, u = decomp.eigh(a, lower=lower, check_finite=False)
File "/Users/AZevelev/.julia/conda/3/lib/python3.7/site-packages/scipy/linalg/decomp.py", line 474, in eigh
raise LinAlgError("unrecoverable internal error.")
pyerr_check at exception.jl:60 [inlined]
pyerr_check at exception.jl:64 [inlined]
_handle_error(::String) at exception.jl:81
macro expansion at exception.jl:95 [inlined]
#110 at pyfncall.jl:43 [inlined]
disable_sigint at c.jl:446 [inlined]
__pycall! at pyfncall.jl:42 [inlined]
_pycall!(::PyCall.PyObject, ::PyCall.PyObject, ::Tuple{Array{Float64,2},Array{Float64,1}}, ::Int64, ::Ptr{Nothing}) at pyfncall.jl:29
_pycall!(::PyCall.PyObject, ::PyCall.PyObject, ::Tuple{Array{Float64,2},Array{Float64,1}}, ::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at pyfncall.jl:11
(::PyCall.PyObject)(::Array{Float64,2}, ::Vararg{Any,N} where N; kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at pyfncall.jl:86
(::PyCall.PyObject)(::Array{Float64,2}, ::Vararg{Any,N} where N) at pyfncall.jl:86
fit!(::PyCall.PyObject, ::Array{Float64,2}, ::Vararg{Any,N} where N; kwargs::Base.Iterators.Pairs{Union{},Union{},Tuple{},NamedTuple{(),Tuple{}}}) at Skcore.jl:100
fit!(::PyCall.PyObject, ::Array{Float64,2}, ::Array{Float64,1}) at Skcore.jl:100
fit(::ARDRegressor, ::Int64, ::NamedTuple{(:Crim, :Zn, :Indus, :NOx, :Rm, :Age, :Dis, :Rad, :Tax, :PTRatio, :Black, :LStat),NTuple{12,Array{Float64,1}}}, ::Array{Float64,1}) at ScikitLearn.jl:157
fit!(::Machine{ARDRegressor}; rows::Array{Int64,1}, verbosity::Int64, force::Bool) at machines.jl:183
(::StatsBase.var"#fit!##kw")(::NamedTuple{(:rows,),Tuple{Array{Int64,1}}}, ::typeof(fit!), ::Machine{ARDRegressor}) at machines.jl:146
top-level scope at MLJ_FitMachine_Error.jl:18 |
|
Okay, I no longer fail on Sorry I don't have 1.4 installed just now. Can check later |
|
So it's only CI that fails this maybe due to the way that it installs SkLearn @staticfloat what do you think of the travis log here: https://travis-ci.com/github/alan-turing-institute/MLJModels.jl/jobs/320357986 ? |
|
This well-known issue not related to MLJ is being tracked. So closing |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Further to #211 :
I have introduce CI on this branch for julia 1.3 and julia 1.4, where testing in now failing for MacOS.
The error is triggered by testing of the wrapped scikit-learn clustering models. According to the travis logs, the conda installations for scikit-learn are the same for linux and macOS, excect that macOS has an additional package llvm-openmp-4.0.1 installed.
Any help at all on this one would be appreciated. In particular, should I regard this as julia 1.3/1.4 error?
Here is the tail of the stack trace
The text was updated successfully, but these errors were encountered: