comparison doc/v2_planning/datalearn.txt @ 1371:98d4232df1d8

comment on OD idea
author Razvan Pascanu <r.pascanu@gmail.com>
date Mon, 15 Nov 2010 16:28:02 -0500
parents 5785cbac3361
children decee534c78d
comparison
equal deleted inserted replaced
1370:5785cbac3361 1371:98d4232df1d8
474 constants=['corruption_layer_1'])))) 474 constants=['corruption_layer_1']))))
475 This would be one way to have automatic lazy function cache / compilation 475 This would be one way to have automatic lazy function cache / compilation
476 while still letting the user specify for which parameters a new function needs 476 while still letting the user specify for which parameters a new function needs
477 to be compiled when their value changes. 477 to be compiled when their value changes.
478 478
479 RP comment : What about the same trick that Theano uses, namely, if you want
480 a non "default" behaviour you wrap the input in a dictionary. You would
481 write tranform1( input_data,
482 corruption_layer_1= In(value = c1, fixed = True)) ?
483 I started to like this approach of passing extra info about an argument :).
484 Other that this it sounds good to me.
485
479 486
480 Discussion: Helper Functions 487 Discussion: Helper Functions
481 ---------------------------- 488 ----------------------------
482 489
483 James: Another syntactic option for iterating over datasets is 490 James: Another syntactic option for iterating over datasets is