2
93%
master: 93%

Ran 25 Apr 2016 02:55AM UTC

Files 47

Run time 2s

Badge

Embed ▾

Committed 25 Apr 2016 02:50AM UTC coverage: 93.649%. Remained the same

Job # TESTS=blocks FLOATX=float32 DB=sqlite

Build Type

push

travis-ci

Committed by

dwf

Commit Message

Use tensor.nnet.relu for ReLU calculations. (#1070)

* Use tensor.nnet.relu for ReLU calculations.

We punt the decision of how to implement max(0, x) or max(alpha * x, x) most efficiently to Theano, via `tensor.nnet.relu`. This is what Lasagne and Keras do, which between them have > 10x as many GitHub stars, so I am willing to treat this solution as "battle-tested".

(Finally) closes #960.

N.B. This was done via a quick edit on the GitHub web interface so the branch is on `mila-udem`, I'll delete it when merged.

* Fix LeakyRectifier test sensitivity.

Tests were failing with this new implementation due to an unduly high sensitivity.

Run Details

4158 of 4440 relevant lines covered (93.65%)

0.94 hits per line

mila-udem / blocks / 3374 / 2
93%
master: 93%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 3374.2 (TESTS=blocks FLOATX=float32 DB=sqlite)

mila-udem / blocks / 3374 / 2 93% master: 93%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 3374.2 (TESTS=blocks FLOATX=float32 DB=sqlite)

mila-udem / blocks / 3374 / 2
93%
master: 93%

README BADGES
x