dear Avi and Bert,
I think I got my answer. I will just run it with a
small sample and check the execution time and extrapolate from that. By the way,
LDA (I am using topicmodels package) cannot be parallelized, right? Thanks
in advance.
Thanking you,
Yours sincerely,
AKSHAY M KULKARNI
________________________________
From: R-help <r-help-bounces at r-project.org> on behalf of Avi Gross via
R-help <r-help at r-project.org>
Sent: Sunday, January 30, 2022 4:15 AM
Cc: r-help at r-project.org <r-help at r-project.org>
Subject: Re: [R] progress of LDA algorithm...
I agree with Bert that this is way off topic and one few here know (or care)
about.
Generally, if a package has functionality with manual pages, it may have
abilities defined such as setting verbose=TRUE or to various levels of output
that may satisfy the request or they may make a copy of code including their
print or logging statements and so on.
If the request is more general such as how to run a program under some debugging
method and set checkpoints at which some reporting is done, that too is a bit
outside the normal uses of this forum.
The usual suggestion here is to contact the package maintainer, with no
guarantee of getting any useful response, or find a forum way more specific than
R HELP just because part of the package is in R.
As it happens, the lda() function being discussed may (or may not) be in the
MASS package. Looking at the documentation, I saw no obvious hook to show it as
it makes progress. Of course Akshay can do some external testing using standard
R timing mechanisms to see how long it takes to do just some of the news
categories without going in to the details of the function called and that might
partially answer his question. Asking how to do that might fit the parameters
here.
-----Original Message-----
From: Bert Gunter <bgunter.4567 at gmail.com>
To: akshay kulkarni <akshay_e4 at hotmail.com>
Cc: R help Mailing list <r-help at r-project.org>
Sent: Sat, Jan 29, 2022 3:34 pm
Subject: Re: [R] progress of LDA algorithm...
I presume this is in some specialized package that you have not told
us about -- topicmodels maybe? It is therefore off topic here. In any
case, this is the sort of question for which you should contact the
package maintainer (?maintainer).
As your question may also intersect with high performance computing
considerations, you might want to post it on the R-Sig-HPC list,
https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
Bert Gunter
"The trouble with having an open mind is that people keep coming along
and sticking things into it."
-- Opus (aka Berkeley Breathed in his "Bloom County" comic strip )
On Sat, Jan 29, 2022 at 8:27 AM akshay kulkarni <akshay_e4 at hotmail.com>
wrote:>
> dear members,
> I want to run LDA(latent Dirichlet allocation) on
certain news articles. i have the following questions:
>
>
> 1. Is there any way to know the progress of the execution of the LDA
algorithm?
> 2. I read in SO that if you have more memory, faster is the execution
time of LDA. I am using AWS z1d instance with 48 cores and about 325 GB RAM. I
have multiple categories of news, but one of them is much larger than others,
containing about 25000 articles. Is it preferable to send those categories
individually to different processors, and whether R frees up the memory after
running on the smaller categories so that the largest category can run with more
memory? Or is it preferable to first run the smaller sets, finish the job, and
then run the largest category?
>
> Thanking You,
> Yours sincerely,
> AKSHAY M KULKARNI
>
______________________________________________
R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
[[alternative HTML version deleted]]