Відео

Thank you for sharing the course on UA-cam.

very nice lecture, thank you so much

Very informative! Much obliged.

fucking terrible

Hi! Thank you for the very informative lecture. What is the name of the annotation tool or where to find it? Thanks in advance.

and the best trick is the data structure used holding the language stats, that apparently gets keyed by current context, a vector of word embeddings????

I guess the RNN state has to also keep a whole English/German dictionary as well as a gigantic English language data structure that remembers all the occurrence/context statistics for every Ngram and every context Q-deep.....also for German.....and I doubt the RNN state is carrying around all that info from one step to the next..... so....how? maybe it'll get cleared up in future lectures...I hope so... RNN doesnt just <know> that it otta be <Haus> in German when it's <house> in English.....RNN is relying on a huge exhaustive statistical compilation that took place BEFORE we ever tried to translate anything....involving multi-BILLIONs of words of parallel text, then also English and also German text....musta needed MONTHs to acquire and weeks to execute...

Thank you so much i was looking for explanation about this topic glad i found whole series about it

I simply became fan of your videos on MT.. Thank you so much !!!

Thank you so much for making this high-quality course available for everyone! This is like a treasure for those who want to learn more about MT :)

it should be 5^3 not 4^3

In my thoughts outbof this lectur beside the technical issues face TM is language based on the research (English), there's no fixed mathematical rules the languages so we need a language helps us the challenge. I'm not linguist but after my mom language Somali English and Arabic are secondary and my thoughts are around it. English is more flexible language and well documented on the other hand not good when it comes Grammarly and using cases also it lacking on vocabulary, speakers put alot of effort on portability to other languages and it worked but still near full percent at last whiting is too poor we all know pronunciation is base of learning the language. where as Arabic have rigid language structuring and grammar understanding, vocabulary is rich enough as far My understanding portability to other language is to easy if you can speak, and write Arabic every language is be easy to learn other languages what language family it is, if learn how to read Arabic you can't miss pronounce. Somali is where between them,

This is exactly what i was looking for but the audio quality is making it really difficult to follow the explanation. Volume is maxed out but still inaudible at many places.

Good

35:51 literally crickets in the background lol appreciate the late night recording bro

useless lecture dont waste your time here

Thank you for making this publicly available!

Thanks, professor Philipp, very useful lectures. Could you recommend a book/resource on probability/maths that's necessary to follow the lectures for those who's been out of college for a long time?

Thanks.

Looking forward for the rest of the series. Thanks.