I have a fresh Windows 10 install of R, RStudio, Python 3.6.8, TensorFlow, and the tfestimators library. However, whenever I execute the feature_columns() function, for example:
feature_cols <- feature_columns(column_numeric("test_column_name"))
...I get this error message:
Error in py_get_attr_impl(x, name, silent) :
AttributeError: module 'tensorflow.python.feature_column.feature_column' has no attribute 'numeric_column'
Any idea how to solve this?
I've restarted RStudio and Windows several times, reinstalled Python (This error occurs using both versions I've tried, 3.7 and 3.6.8), reinstalled the tfestimators package, and reinstalled TensorFlow. Same error always occurs.
Related
I am currently working on installing the gpu version of tensorflow onto my computer; however, I am experiencing several difficulties. My initial problem was a relatively common problem where Tensorflow was not able to find the CUDA libraries during installation, and I found a very useful post here on StackOverflow that mentioned that by changing the 'CUDA_PATH' environment variable on my computer and adding the 'bin' folder to said variable's path, Tensorflow should be able to find the CUDA libraries.
This didn't work for me, and so I changed the name of the environment variable 'CUDA_PATH' to simply be 'PATH', since that was the environment variable name that was shown on the post I followed.
It didn't work either so I changed it back. Now, everytime I open RStudio, the following error message appears:
Error in Sys.setenv(PATH = new_path) : wrong length for argument
In addition: Warning message:
In system2(conda, c("run", in_env, run_args, shQuote(cmd), args), :
running command '"C:/Users/maria/AppData/Local/r-miniconda/condabin/conda.bat" run --prefix "C:\Users\maria\AppData\Local\r-miniconda\envs\r-reticulate" --no-capture-output "python" -c "import os; print(os.environ['PATH'])"' had status 1
I have tried uninstalling R, Rtools, Rstudio, miniconda and all my Nvidia GPU drivers, as well as CUDA and CUDANN. I then have re-downloaded every programme, and the error still appears whenever I start a session in Rstudio (before I can even do anything remotely related to tensorflow or before I even use R in any way).
If anyone has any suggestions I would greatly appreciate some help with this issue.
Thanks :)
P.S.: Please find below a screenshot of my RStudio upon opening the programme.
When loading
library(XLConnect)
I get the error message
Error: package or namespace load failed for ‘XLConnect’:
.onLoad failed in loadNamespace() for 'XLConnect', details:
call: if (twoFirst == "1.") {
error: argument is of length zero
This program has been working smoothly for months, this is out of the blue after no changes to the code. I have seen that many issues have come from Java, and have installed latest Java (V8 update 281) in 32 and 64-bit flavors. In addition, I have tried switching RStudio to 32 bit from 64 with no change in the result. Also have reinstalled the XLConnect package, rJava package, restarted, ensured R is running as administrator, all with still the same error.
What happened and why is it not loading?
Running R 4.0.4 and RStudio 1.4.1103 on Windows 10 x64.
I will probably switch to readxl after this, but that will take time.
One of my students was experiencing exactly the same problem mentioned above (in our case, with R version 4.0.3 in Windows 10, Java 1.8) and no solution was working for him. However, a few minutes ago, exploring his environment variables, we noticed that he had environment variable _JAVA_OPTIONS set to -Xmx512m. Once we have removed such variable, XLConnect is working properly.
I am facing error in this command though I have python3 x in my system and have anaconda as well. I tried all the option given in
Error: Unable to find conda binary. Is Anaconda installed? reticulate Rstudio
https://stackoverflow.com/posts/53434444/revisions, but nothing working for me.
In fact, I am getting error for H2o as well.
Warning in install.packages :
unable to move temporary installation ‘C:\Users\kumaashi\Documents\R\win-library\3.5\file124f052421e1\h2o’ to ‘C:\Users\kumaashi\Documents\R\win-library\3.5\h2o’
I got this error earlier also but used to get fixed with trace(utils:::unpackPkgZip, edit=TRUE) by changing sys.time - 2.5 but since i updated R that option is not coming at all
I'm getting a confusing error while using RStudio. When I use the command line, in the middle of typing a command (never while executing a command), R throws the following error:
Error in loadNamespace(name) : there is no package called ‘.GlobalEnv’
I am using R version 3.2.3 and RStudio version 0.99.491; I have tried to re-install RStudio and to delete settings.
Look at the tabs that open at startup.
In my case, it was because there was a debugging tab opened and RStudio was trying to load it at startup. I just closed it and that error stopped.
I have the following problem:
I have some development version of my package loaded in R session
I edited the source file.
I detach it and do:
system("R CMD check realizedvolatility")
system("R CMD build realizedvolatility")
system("R CMD install realizedvolatility_0.1.tar.gz")
library(realizedvolatility)
Until now, everything works. Now, if I try to bring up help file, for example ?realizedvolatility , the error
Error in fetch(key) : internal error -3 in R_decompress1
occurs.
The remedy is to restart entirely the R session, then everything works. Is there any way to make it work within one session? I tried detaching from namespace, installing with various options, nothing worked.
I am using Mac OS X Lion, and R 2.14.1
I asked the same question in R-help and one R core replied "this is by design" which means you cannot do anything about it except restarting R: https://stat.ethz.ch/pipermail/r-help/2011-July/283916.html (although I still do not understand why this design cannot refresh the cache database...)