R Studio numpy version

[lve0200]

Please refer to the following stack overflow message
https://stackoverflow.com/questions/51541031/numpy-version-is-not-satisfied-for-tensorflow
Error in py_dict_impl(keys, values, convert = convert) : Evaluation error: Required version of NumPy not available: installation of Numpy >= 1.6 not found
The issue seems to be here:
# check numpy version and provide a load error message if we don't satisfy it if (is.null(config$numpy) || config$numpy$version < "1.6") numpy_load_error <- "installation of Numpy >= 1.6 not found" else
Please note, that the version is found in R, there seems just to be a difficulty to interpret the version w/ 2 dots. Please see the follwing output from R
` np <- reticulate::import("numpy")

print(np$version$full_version)
[1] "1.15.4"
`

Hence there is no env/path issue!
I would compile a new version, if I would know, how to clone and compile in Windows for Windows 10. When I installed it, it did compile by itself (hence I have the necessary tools installed as cc, make etc. but I don't know, how to use them in windows.
Thanks

[kevinushey]

Can you please share what the output of the following is?

library(reticulate)
config <- py_config()
config$numpy

It's more likely that reticulate is unable to find or load numpy during initialization for some reason.

[kevinushey]

For reference I have:

> library(reticulate)
> config <- py_config()
> config$numpy
$path
[1] "/Users/kevin/Library/Python/2.7/lib/python/site-packages/numpy"

$version
[1] '1.15.4'

[lve0200]

Thanks for the question, things getting strange here...
That's what I get:

`> library(reticulate)

config <- py_config()
config$numpy
NULL
np <- reticulate::import("numpy")
print(np$version$full_version)
[1] "1.15.4"
print(config)
python: C:\Users\hk\Anaconda3\envs\r-tensorflow\python.exe
libpython: C:/Users/hk/Anaconda3/envs/r-tensorflow/python36.dll
pythonhome: C:\Users\hk\ANACON1\envs\R-TENS1
version: 3.6.8 |Anaconda, Inc.| (default, Feb 11 2019, 15:03:47) [MSC v.1915 64 bit (AMD64)]
Architecture: 64bit
numpy: [NOT FOUND]
tensorflow: C:\Users\hk\ANACON1\envs\R-TENS1\lib\site-packages\tensorflow_init_.p

python versions found:
C:\Users\hk\Anaconda3\envs\r-tensorflow\python.exe
C:\Users\hk\ANACON~1\python.exe
C:\Users\hk\Anaconda3\python.exe`

[lve0200]

just adding the call stack from R-Studio

`Error in py_call_impl(callable, dots$args, dots$keywords) :
Evaluation error: Required version of NumPy not available: installation of Numpy >= 1.6 not found.
In addition: Warning message:
In normalizePath(path.expand(path), winslash, mustWork) :

Hide Traceback

Rerun with Debug
Error in py_call_impl(callable, dots$args, dots$keywords) :
Evaluation error: Required version of NumPy not available: installation of Numpy >= 1.6 not found.
18.
stop(structure(list(message = "Evaluation error: Required version of NumPy not available: installation of Numpy >= 1.6 not found.",
call = py_call_impl(callable, dots$args, dots$keywords),
cppstack = NULL), class = c("Rcpp::eval_error", "C++Error",
"error", "condition")))
17.
py_call_impl(callable, dots$args, dots$keywords)
16.
tensorflow::tf$contrib$framework$is_tensor(x)
15.
keras_array(x)
14.
fit.keras.engine.training.Model(., x, y, epochs = 30, batch_size = 128,
validation_split = 0.2) `

[lve0200]

I see in
print (config$required_module_path)
[1] "C:\Users\hk\ANACON1\envs\R-TENS1\lib\site-packages\tensorflow\init.p"
is there eventually an "y" missing at the end, because the file in the directory is init.py !! ??

[lve0200]

sorry, if things are getting confused!
I did a complete fresh install of Keras and keras_install(method="conda")
Everything worked through, but I got a
Warning message:
In normalizePath(path.expand(path), winslash, mustWork) :
path[1]="C:\Users\hk\ANACON-1\envs\rstudio/python.exe": Das System kann die angegebene Datei nicht finden (not found)
I get the same warning, when executing
library(reticulate)

config <- py_config()
Warning message:
In normalizePath(path.expand(path), winslash, mustWork) :
path[1]="C:\Users\hk\ANACON-1\envs\rstudio/python.exe": Das System kann die angegebene Datei nicht finden
config$numpy
NULL
I then copied python.exe from ../tensorflow to ../rstudio
Now, it apeared working
library(reticulate)
config <- py_config()
config$numpy
$path
[1] "C:\Users\hk\ANACON-1\envs\R-TENS~1\lib\site-packages\numpy"

$version
[1] ‘1.15.4’

but just to get another error:
Error in py_call_impl(callable, dots$args, dots$keywords) :
AttributeError: 'Sequential' object has no attribute 'shape'
13.
stop(structure(list(message = "AttributeError: 'Sequential' object has no attribute 'shape'",
call = py_call_impl(callable, dots$args, dots$keywords),
cppstack = structure(list(file = "", line = -1L, stack = "C++ stack not available on this system"), class = "Rcpp_stack_trace")), class = c("Rcpp::exception",
"C++Error", "error", "condition")))
12.
layer(object)
11.
compose_layer.python.builtin.object(object, layer)
10.
compose_layer(object, layer)
9.
create_layer(keras$layers$Dense, object, list(units = as.integer(units),
activation = activation, use_bias = use_bias, kernel_initializer = kernel_initializer,
bias_initializer = bias_initializer, kernel_regularizer = kernel_regularizer,
bias_regularizer = bias_regularizer, activity_regularizer = activity_regularizer, ...
8.
layer_dense(., units = 256, activation = "relu", input_shape = c(12))
7.
function_list[i]
6.
freduce(value, _function_list)
5.
_fseq(_lhs)
4.
eval(quote(_fseq(_lhs)), env, env)
3.
eval(quote(_fseq(_lhs)), env, env)
2.
withVisible(eval(quote(_fseq(_lhs)), env, env))
1.
model %>% layer_dense(units = 256, activation = "relu", input_shape = c(12)) %>%
layer_dropout(rate = 0.4) %>% layer_dense(units = 128, activation = "relu") %>%
layer_dropout(rate = 0.3) %>% layer_dense(units = 10, activation = "softmax")

[microly]

Maybe this issue can be referenced to #367.

[microly]

I find the solution from #367 .

you can try:
Sys.setenv(PATH= paste("D:/DEVTOOLS/Anaconda2/envs/py37/Library/bin",Sys.getenv()["PATH"],sep=";"))
Sys.setenv(RETICULATE_PYTHON = "D:/DEVTOOLS/Anaconda2/envs/py37/python.exe")
library(reticulate)

Or

load R/Rstudio from a conda command prompt that has already set the PATH correctly.