Knut Jägersberg
Knut Jägersberg
Have you considered working on meta embeddings and embedding imputation? I think fse might practically challenge some deep learning architectures, especially when taking knowledge graph embeddings into account.
From this research, Sentence Embedding via Semantic Subspace Analysis (S3E) allows us to construe sentence embeddings which are competitive with sentence bert. [[Sentence Embedding via Semantic Subspace Analysis](https://github.com/BinWang28/Sentence-Embedding-S3E)](https://arxiv.org/abs/2002.09620) Would be...
Integrating OnlineStats (its online learning algorithms) and giving it an easy to use hyperparameter tuning context makes Julia even more useful for quick ML on real big data.
Hi, sometimes using the contentscraper function on a longer vector of urls, it may fail in between and then everything is lost. working around it with wrapping the function with...
Not always I use the crawler, actually more often a content scraper following a predefined link path, so I can be sure the complete site is scraped. The content scraper...
Hi, a nice feature would be an integration with ip shuffeling services. There are numerous, i was thinking about something like https://proxycrawl.com/ It is not hard to implement it yourself,...
would be very handy. XLM-R is not supported yet by most nlu libs. i.e. https://github.com/huggingface/transformers/issues/1769
I like the pak package, for parallelized install and downloads. Is it possible to use that as backend for pacman?
Is it possible to monitor RAM usage and automatically adjust disk frame on the fly, i.e. with regards to workers used? I sometimes have the issue that I run out...
That would be handy, too. For now we can reshape data using map, but it would feel more natural to do that in well known tidyr syntax. think of gather...