That feeling when your code compiles and runs without error, but you still feel like there is a bug somewhere... lurking...
-
-
Replying to @kastnerkyle
@kastnerkyle _especially_ for ML where an optimiser will optimise any thing you give it (unlike say web dev where a button works/doesn't)1 reply 0 retweets 1 like -
Replying to @mat_kelcey
@mat_kelcey yup. Cost is decreasing - ship it! Bonus points if you only monitor training set2 replies 0 retweets 2 likes -
Replying to @kastnerkyle
@kastnerkyle@mat_kelcey in my exp implementing lots of RNNs, you can do things *very* differently from the papers and still get similar res2 replies 1 retweet 3 likes -
Replying to @fchollet
@fchollet@kastnerkyle & often that very similar result isn't even using an RNN ;)2 replies 0 retweets 2 likes
Replying to @mat_kelcey
@mat_kelcey @kastnerkyle the curse of DL research: when your experiment is successful, it's near impossible to link that to your hypothesis
6:13 PM - 19 Jan 2016
0 replies
2 retweets
5 likes
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.