Demonstration of the new tfruns package by RStudio

[@fssski]

Can I use egpu with caret package on rstudio for machine learning?

[@rickeycarter]

I do not believe any of the routines called or wrapped by caret are optimized for GPUs. The neural networks employed by caret are relatively simple architecture. I suspect you could write your own wrappers for caret or use their helper tools to segment the data for cross validation easily enough.

[@Tom-sy8yt]

Hi thanks for the post! I'm kinda stuck with first input value of a file for the training_run. Do you save your rfile, kerasmodel or something else?

[@rickeycarter]

tfruns captures most objects created in the projects as they execute. Console text and objects are sent the the /runs directory. However, prior to running in tfruns, I get some basic frameworks that seem promising. If you are interested, I posted a more detailed blog post on my much neglected webpage. www.rickeyecarter.com/blog/2018/3/23/tflearns-applied-to-mnist-data

[@priyamsinha9685]

Can you provide the file

[@rickeycarter]

PRIYAM SINHA I can’t share the image data. I did write a generic blog post that shows how to use the tfruns package. www.rickeyecarter.com/blog/2018/3/23/tflearns-applied-to-mnist-data