#introductions Hey everyone. I am a PhD Student in Computational Linguistics. My main research interest is computational linguistics for under resourced languages. I am particularly interested in computational linguistics for Bantu languages because of the extremely complex morphology that they bring to the table.
I am also doing research in sentiment analysis on stance detection and irony/sarcasm detection. Most of my research is on machine learning approaches.
#introductions
Oh I also work at the LINGUIST List linguistlist.org, the linguistics community website.
#introductions Hi, I am a PhD student from Wroclaw University of Science and Technology, Poland. My research fileds are discrete optimization, cloud computing and software engineering
Does any other field value conference publications to the extent that computational linguistics does?
I feel like it shapes the whole conference submission process. We don't submit abstracts or short papers, typically you submit what's going in the proceedings instead of an abstract.
Maybe I just think it's so different because Linguistics conferences typically do abstracts to determine who gets accepted.
Camera ready version of my Luyia pos tagging paper has been submitted!
I can feel my stress levels dropping.
I wonder if I can post my paper on my website. I imagine I can since all acl papers are creative Commons licensed.
Car stuff Show more
Hello! I’m a newly-minted CS PhD (UC Santa Cruz) and Assistant Prof at Pomona College. My research area connects videogames, AI, verification, and specification recovery. I’m also really interested in GPGPU and other forms of energy-efficient computation. I’m on a mission to help my students replace the games industry and Silicon Valley with more humane alternatives.
I want to learn more about equitable instructional, assessment, and evaluation design!
The splunk tutorial has a child's laugh for admins who get a new report.
I think this is supposed to be funny but it honestly scared me.
there’s a special pain when you see a flawed and easily refuted argument made in support of a conclusion you agree with for other reasons
gargron told me if this post gets ten thousand boosts he'll give me the website and i will become the new gargron
The final release of the TuBa-D/Z corpus was just released. They've now got almost two million words annotated with a whole host of different grammatical markups.
http://www.sfs.uni-tuebingen.de/en/ascl/resources/corpora/tueba-dz.html
Rdp with kde is so much smoother than vnc.
Oof... Indiana University is going to start requiring two factor authentication for sshing onto the compute clusters.
Don't get me wrong, 2fa is good. I just do not like duo (which is the client they chose). It's frequently caused errors when my cell signal or WiFi signal is low where it says the request has timed out in the app but the ldap server is still waiting for a response to its push.
I get in situations often where I have enough service to check my email but not enough to Auth email
I propose a new ideology: #antifatalism
It's about the affirmative belief that humans can change the things that can in principle be changed by humans: Climate change, War, Poverty, etc. We don't have to be resigned to the idea that misery is neverending.
There's no natural law (not even entropy) that says we cannot fix these things. There's only us.
(Note that antifatalism does not imply belief in free will.)
#car opinion Show more
Does anyone know if you can link a pixelfed to mastodon yet?
We've released #Krita 4.1! Check out the annoucement https://krita.org/en/item/krita-4-1-0-released/ and the release notes https://krita.org/en/krita-4-1-release-notes/ , download and have fun!
Debating whether it's worth it to get two e5-2680v2's for one of the nodes.
On the one hand, most mpi workloads do not perform well with asymmetric performance between nodes. You waste a lot of time waiting for the slower node to return if your problem can't be scheduled to take advantage of the fact that one node is faster. I think all_gather and all_reduce allow for modified scheduling but I can't quite remember.
But the the question becomes how often am I going to write mpi workloads...
This quanta windmill is a workhorse. One of my friends has been running an expectation-maximization algorithm he developed on the working node for a week now and all 16 cores are at 450 mhz over base clock with full utilization.
Definitely worth buying a new pwm generator for my second node.
I think a more qualitative approach may be better for my examination of why Tiriki works better than Bukusu. If I look through the words that the Tiriki models got right and the Bukusu models got wrong, what do I notice?
Got to put on my #Bantu hat.
personal Show more