[NLPL Task Force (A)] [NLPL Board] Associate member request: Department of Linguistics, Stockholm University

[Robert Östling]

Thanks Stephan!

On 10/17/2018 01:24 PM, Stephan Oepen wrote:
> dear robert,
> 
> on behalf of the NLPL steering group, i am happy to confirm that we much 
> welcome your involvement as an associate partner!
> 
> may i suggest that both you and marathan request access following the 
> procedures described on the NLPL wiki:
> 
> http://wiki.nlpl.eu/index.php/Infrastructure/resources
> 
> please apply for both Abel and Taito at the same time, as currently our 
> harvesting of user information (for a low-traffic NLPL-internal 
> announcement list) is restricted to the user database on Abel.  and, 
> actually, Abel gpu resources have been lightly subscribed in recent 
> months :-).  NLPL load on Taito is comparatively high lately, so the 
> more you can put on Abel rather than on Taito the better.

Will do!

> 
> —out of curiosity: how was your experience with the estonian system 
> under the dellingr pilot?

We didn't use it a lot in fact, mostly because I have been doing less 
computationally intensive stuff for a while. And the investment in time 
needed for setting up a new environment was a problem too, especially 
since the allocation was only for a limited time.

Robert


> 
> best wishes, oe
> 
> 
> On Tue, Oct 16, 2018 at 2:23 PM Robert Östling <robert at ling.su.se 
> <mailto:robert at ling.su.se>> wrote:
> 
>     Hi!
> 
>     Stephan and Jörg suggested we should apply to become associate members
>     of NLPL.
> 
>     Users here at the department would be:
> 
>     - Robert Östling (assistant professor)
>     - Murathan Kurfali (PhD student)
> 
>     Expected types of data to be used:
> 
>     - Parallel (OPUS) and monolingual text (web corpora)
>     - Pre-trained word vectors
> 
>     Expected types of computation:
> 
>     - GPUs for neural network training (mainly on Taito I guess)
>     - CPUs for preprocessing, word alignment, etc. (Taito/Abel both OK)
> 
>     We expect to contribute resources related to our research on highly
>     multilingual models with the Bible corpus (word alignments and word
>     embeddings for 1400+ languages). If people are interested we would need
>     a storage allocation for that.
> 
>     Don't hesitate to ask if you need more information from or about us.
> 
>     Robert
>