Skip to content

Fix Dense init bug where initialisation of kernel weights with a numpy array raises a TypeError #3592

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 4 commits into
base: master
Choose a base branch
from
Open

Fix Dense init bug where initialisation of kernel weights with a numpy array raises a TypeError #3592

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
bd526b9
Merge pull request #1 from Microsoft/master
delzac Jul 19, 2018
51f5ba3
Merge branch 'master' of https://github.com/delzac/CNTK
delzac Sep 13, 2018
bf0df64
Merge branch 'master' of https://github.com/Microsoft/CNTK
delzac Mar 10, 2019
7fe3782
allow init in dense to be directly initialised by numpy array
delzac Mar 10, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion bindings/python/cntk/layers/layers.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -135,7 +135,11 @@ def Dense(shape, activation=default_override_or(identity), init=default_override
infer_input_rank_to_map = map_rank # infer W to use all input dims except the first static 'map_rank' ones

# parameters bound to this Function
init_weights = _initializer_for(init, Record(output_rank=output_rank))
if isinstance(init, np.ndarray):
init_weights = init
else:
init_weights = _initializer_for(init, Record(output_rank=output_rank))

W = Parameter(input_shape + output_shape, init=init_weights, name='W')
b = Parameter( output_shape, init=init_bias, name='b') if bias else None

Expand Down