greato
greato on Hacker News
Posted by
greato on Monday April 02, 2018 @06:36PM
from the show-hn dept.
Posted by
greato on Saturday July 22, 2017 @08:03PM
from the rickyhan.com dept.
Does this mean ETC is compromised too?
Houellebecq's first book Whatever is about the unfruitfulness of programming.
Unless you want to execute the structure on multiple GPUs.
In neural translation seq2seq, using while_loop in the decoder RNN saves a lot of GPU time because it can quit early when a sentence ends.
Using tf.scan is a
bad idea.
scan implements strict semantics so it will always execute the same number of timesteps no matter what the accumulator is (nan).
while_loop implements dynamic execution (quit once cond is not met) and at the same time allows parallel execution when some ops are not dependent on accumulator.
If you read the code for `dynamic_rnn` and contrib.legacy Seq2seq model you'll find while_loop. I have yet to see tensorflow library code using tf.scan anywhere!
Also, internally, scan is defined using while_loop. In my code, I find scan lacking in RNN and always have to fall back to while_loop.
Here is video of a talk by the RNN/Seq2Seq author himself:
https://youtu.be/RIR_-Xlbp7s?t=16m3s
I read the article and seems to be well-written though lacking.
For even more customized RNNs such as attention mechanism, beam search as in Seq2Seq, you'll need to skip the tf.dynamic_rnn abstraction and use a symbolic loop directly: tf.while_loop
He was against people calling Deep Blue AI when it was not. I doubt he would say the same thing about AlphaGo.
This is one of the many unconstructive comments
Thanks for the heads up. This will be free and open source. Also checkout my main project txtpen thanks.
Posted by
greato on Monday June 05, 2017 @12:51PM
from the show-hn dept.
https://github.com/TXTPEN/hnA minor bug has been fixed. Now it should be ready for production.
It's been fixed. It was broken because it was querying a sub comment(not root comment).
hn.algolia.com is aggressively cached. It's updated every 3 minutes.
Posted by
greato on Sunday June 04, 2017 @11:11AM
from the show-hn dept.
Posted by
greato on Tuesday May 23, 2017 @04:38AM
from the ask-hn dept.
Is it even possible to create a faster data structure server? Redis is written in C and made the design choices to be very, very fast.
Posted by
greato on Monday May 22, 2017 @12:55PM
from the show-hn dept.
MVP is minimum measurable product. When you have analytics set up, you are ready to launch. (mailing list)
Posted by
greato on Monday May 08, 2017 @01:39AM
from the rickyhan.com dept.
Posted by
greato on Tuesday January 24, 2017 @10:37PM
from the fordite.com dept.
You should talk the users and iterate your product. When you have enough traction, talk to investors. It'll make your life and theirs easier.
YC partner Justin Kan has a sketch of magic mushrooms framed in his living room. I saw it on his Snapchat when he showed his followers around his house.
just another study on ketosis
Posted by
greato on Sunday April 26, 2015 @08:40PM
from the panzi.github.com dept.
The contributors of the site all have Spanish last names. The neutrality of the site is questionable.
Or you can just write a small script that syncs Google Tasks to trello
Posted by
greato on Tuesday October 28, 2014 @01:47PM
from the show-hn dept.