Skip to content

Add public predict method for vector of tensors#396

Merged
nyanp merged 3 commits into
tiny-dnn:masterfrom
reunanen:add-public-predict-method-for-vector-of-tensors
Nov 12, 2016
Merged

Add public predict method for vector of tensors#396
nyanp merged 3 commits into
tiny-dnn:masterfrom
reunanen:add-public-predict-method-for-vector-of-tensors

Conversation

@reunanen

Copy link
Copy Markdown
Contributor

For completeness. Want to fprop multiple samples in parallel (in particular because of how fully_connected_op_custom and conv2d_op_custom are parallelized at the moment).

@edgarriba

Copy link
Copy Markdown
Member

@reunanen Could you provide a small test for this?

@edgarriba edgarriba self-assigned this Nov 11, 2016
@reunanen

Copy link
Copy Markdown
Contributor Author

@edgarriba Ok, will try to find the time to do this.

@reunanen

Copy link
Copy Markdown
Contributor Author

@edgarriba Added a test.

@edgarriba

Copy link
Copy Markdown
Member

LGDT

@nyanp review also and merge

@nyanp

nyanp commented Nov 12, 2016

Copy link
Copy Markdown
Member

Looks good.

@reunanen
Thanks again for your contribution! :)

@nyanp nyanp merged commit 94eb590 into tiny-dnn:master Nov 12, 2016
xuyuan added a commit to xuyuan/tiny-cnn that referenced this pull request Nov 16, 2016
# By nyanp (59) and others
# Via GitHub (3) and others
* new_dnn: (145 commits)
  Typo fix (tiny-dnn#404)
  remove warnings (tiny-dnn#387)
  add a workaround to mitigate memory consumption problem
  Update author list
  Fix error
  update comments&docs
  Add some tests & update docs (tiny-dnn#381)
  Fix bugs in test implementation (tiny-dnn#397)
  Add public predict method for vector of tensors (tiny-dnn#396)
  Minor optimization in load_weights_fullyconnected: get dst in and out sizes only once (tiny-dnn#395)
  rename custom -> internal (tiny-dnn#392)
  fix several warnings (tiny-dnn#391)
  fix tests bugs by adding device_t NONE and enable core tests (tiny-dnn#390)
  rename tiny_dnn backend to custom (tiny-dnn#385)
  refactor max_pool layer to op (tiny-dnn#376)
  Fix another compiler warning, this time when building 64-bit (so sizeof size_t > sizeof cnn_size_t) (tiny-dnn#384)
  Make result::print_detail a const function (tiny-dnn#383)
  Fix an unnecessary compiler warning (tiny-dnn#382)
  update README.md with GTest
  Fix modification of CMAKE_CXX_FLAGS_RELEASE and CMAKE_CXX_FLAGS_DEBUG in CMakeLists.txt (tiny-dnn#379)
  ...
pansk pushed a commit to pansk/tiny-cnn that referenced this pull request Nov 17, 2016
* Add public predict method for vector of tensors

* Add test for parallel prediction

* Fix some unnecessary compiler warnings by adding explicit typecasting
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants