Prodigy: A new tool for radically efficient machine teaching • Hacker News Board

Re: Prodigy: A new tool for radically efficient machine teaching

#71

Post by nl » Sat, Aug 05, 2017, 9:58 AM UTC

blueyes wrote: post #24

theincredulousk wrote: post #17

Guess the radical efficiency didn't carry over to their web server

> spaCY, the leading open-source NLP tool? Sounds like marketing BS. what about OpenNLP and Stanford's for NLP?

OpenNLP is.... well I've never heard of anyone using it (except once in a ensemble). I think NLTK is more widely used.

Stanford CoreNLP give good accuracy and is pretty much the benchmark in English for accuracy. BUT it isn't great software. It falls over if you pass large amounts of text to it, the code is dreadful, it's hard to integrate (even in Java because of its own wacky config system), various parts aren't integrated (eg, SUTime), it doesn't have an embedding representation and it is pretty slow.

Having said all that I still use it sometimes. But Spacy is much nicer to use, and 99% (probably more) of the time the slightly lower accuracy is offset by things like the easy availability of word embedding right with the word tags.

I think it's pretty fair to say Spacy is the leading open-source NLP tool.

Re: Prodigy: A new tool for radically efficient machine teaching

#72

Post by carbocation » Sat, Aug 05, 2017, 1:25 PM UTC

user5994461 wrote: post #61

carbocation wrote: post #28

Earlier quoted context omitted.

Turn off KeepAlive, if it's on.

Bad advice. That's likely to make it a lot worse.

I disagree with you based on experience, but you don't have to take my word for it. 'patio11 also has had some experience here: http://www.kalzumeus.com/2010/06/19/running-apache-on-a-memo...

Re: Prodigy: A new tool for radically efficient machine teaching

#73

Post by user5994461 » Sat, Aug 05, 2017, 1:51 PM UTC

carbocation wrote: post #72

user5994461 wrote: post #61

Earlier quoted context omitted.

Bad advice. That's likely to make it a lot worse.

I disagree with you based on experience, but you don't have to take my word for it. 'patio11 also has had some experience here: http://www.kalzumeus.com/2010/06/19/running-apache-on-a-memo...

Also disagree with you based on experience ;)

patio11 blog is HTTP only but this blog is HTTPS.

HTTPS without keepalive is likely to kill any cheap VPS, establishing HTTPS connection is intensive.

That being said, the core of the issue is that they should use nginx (or apache in mpm-events). And they should have cloudflare.

Re: Prodigy: A new tool for radically efficient machine teaching

#74

Post by carbocation » Sat, Aug 05, 2017, 10:52 PM UTC

user5994461 wrote: post #73

carbocation wrote: post #72

Earlier quoted context omitted.

I disagree with you based on experience, but you don't have to take my word for it. 'patio11 also has had some experience here: http://www.kalzumeus.com/2010/06/19/running-apache-on-a-memo...

Also disagree with you based on experience ;) patio11 blog is HTTP only but this blog is HTTPS. HTTPS without keepalive is likely to kill any cheap VPS, establishing HTTPS connection is intensive. That being said, the core of the issue is that they should use nginx (or apache in mpm-events). And they should have cloudflare.

I guess we're going to just wait here until OP delivers the answer for why the machine couldn't answer enough requests :)

Re: Prodigy: A new tool for radically efficient machine teaching

#75

Post by patio11 » Sun, Aug 06, 2017, 1:40 AM UTC

user5994461 wrote: post #73

carbocation wrote: post #72

Earlier quoted context omitted.

I disagree with you based on experience, but you don't have to take my word for it. 'patio11 also has had some experience here: http://www.kalzumeus.com/2010/06/19/running-apache-on-a-memo...

Also disagree with you based on experience ;) patio11 blog is HTTP only but this blog is HTTPS. HTTPS without keepalive is likely to kill any cheap VPS, establishing HTTPS connection is intensive. That being said, the core of the issue is that they should use nginx (or apache in mpm-events). And they should have cloudflare.

establishing HTTPS connection is intensive

More intensive than adding numbers together, sure, but computers are pretty fast. If you're doing 100k connections a second you might have to give some thought to that. Meanwhile, if you have KeepAlive on, 2~5 clients per second will kill you.