Merge branch 'master' of https://github.com/BVLC/caffe#7
Merged
pmgysel merged 35 commits intopmgysel:masterfrom Dec 7, 2018
settle:master
Merged
Merge branch 'master' of https://github.com/BVLC/caffe#7pmgysel merged 35 commits intopmgysel:masterfrom settle:master
pmgysel merged 35 commits intopmgysel:masterfrom
settle:master
Conversation
Make caffe compile through cmake with CUDA 9.1
Since CUDA 9.0 doesn't support sm_20 and sm_21 anymore. This PR allows Caffe to compile with CUDA 9.0 and newer versions successfully. This addresses the discussion concerning [BVLC#6237](BVLC#6237) in [master (BVLC#6237)](BVLC@cb150ec#diff-2004a3d3e6b4ed2e2812bb0b4b998f18)
Minor correction concerning compilation compatibility with CUDA 9.0 (see BVLC#6237)
Fix for compatibility with CMake <3.7 (related conversation under 37e4289) * Fix cmake < v3.7 compatibility in Cuda.cmake * Fix version test variable naming in Cuda.cmake
revised guidelines for GitHub issues and caffe-users posts
Gaussian filler tests adjustments
Fix incorrect namespace for pycaffe submodule caffe_pb2
fix missing syncedmem constness fix issue BVLC#6387
drop unnecessary cast fix issue BVLC#6389
a sketch of `solver.step()` done out manually: 1. `solver.net.forward()` 2. `solver.net.backward()` 3. `solver.net.apply_update()` 4. `solver.net.clear_param_diffs()`
with update exposed it is important to increment the iteration when an update is made, whether by step or update alone. more fundementally, it's the update that defines an iterationa, so this is a natural place for the increment.
`solver.lr` is the effective learning rate in use while `solver.base_lr` is the configured learning rate at initialization. the solver parameter is now editable for setting fields that are in use throughout the lifetime of the solver, such as the maximum iteration.
[pycaffe] expose interface for manual, step-by-step optimization
Otherwise caffe allocates some memory on GPU#0
PoolingLayer customizable output shape rounding mode
minor lint fixes
Clip layer - resurrection
Added USE_HDF5 flag to build
Fix typo in inner_product_layer.cpp
fix typos and add passing by reference
python: Set gpu device id before setting gpu mode
Owner
|
Hi Sean Settle,
Thanks a lot for your pull request, and sorry for getting back to you so
late. Can you please shortly describe the changes which are involved in
your PR, apart from the official Caffe changes?
Best,
Philipp
…On Mon, Nov 12, 2018 at 3:57 AM settle ***@***.***> wrote:
------------------------------
You can view, comment on, or merge this pull request online at:
#7
Commit Summary
- Fix incorrect namespace for pycaffe submodule caffe_pb2 generated by
protobuf
- fix cuda 9.1 compilation
- PoolingLayer customizable output shape rounding mode
- Merge pull request #6237 from jasjuang/master
- Minor correction concerning compilation compatibility with CUDA 9.0
- Merge pull request #6336 from Coderx7/master
- Fix cmake < v3.7 compatibility in Cuda.cmake (#6338)
- Revised guidelines for GitHub issues (#6327)
- tweaked Gaussian filler tests for less false fails
- Merge pull request #6342 from Noiredd/gauss-fill-test-fix
- Cherry-picked USE_HDF5 from Android branch
- Merge pull request #6219 from uhfband/fix-caffe_pb2-module
- fix issue #6387.
- fix issue #6389
- Merge pull request #6388 from open-cv/fix-6387
- Merge pull request #6390 from open-cv/fix-6389
- [pycaffe] expose solver update to do manual solving
- increment iteration during update, not step
- [pycaffe] expose mutable solver parameter, base lr, and effective lr
- [pycaffe] test solver update
- Merge pull request #6238 from mitar/manual-sgd
- Update inner_product_layer.cpp
- fix typos and some minor fixes.
- python: Set gpu device id before setting gpu mode
- Merge pull request #6282 from Noiredd/pooling-mode
- Add clip layer
- test case fix for Clip layer gradient
- Clip layer documentation
- Merge pull request #6320 from Noiredd/clip
- Merge pull request #6346 from jerryz123/HDF5_config
- Merge pull request #6455 from lengly/patch-1
- Merge branch 'master' into patch_1
- Merge pull request #6461 from open-cv/patch_1
- Merge pull request #6499 from xerus/python_gpu
- Merge branch 'master' of https://github.com/BVLC/caffe
File Changes
- *M* .github/ISSUE_TEMPLATE.md
<https://github.com/pmgysel/caffe/pull/7/files#diff-0> (37)
- *M* CMakeLists.txt
<https://github.com/pmgysel/caffe/pull/7/files#diff-1> (3)
- *M* CONTRIBUTING.md
<https://github.com/pmgysel/caffe/pull/7/files#diff-2> (56)
- *M* Makefile <https://github.com/pmgysel/caffe/pull/7/files#diff-3>
(14)
- *M* Makefile.config.example
<https://github.com/pmgysel/caffe/pull/7/files#diff-4> (2)
- *M* cmake/ConfigGen.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-5> (12)
- *M* cmake/Cuda.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-6> (6)
- *M* cmake/Dependencies.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-7> (8)
- *M* cmake/Modules/FindMKL.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-8> (2)
- *M* cmake/ProtoBuf.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-9> (2)
- *M* cmake/Summary.cmake
<https://github.com/pmgysel/caffe/pull/7/files#diff-10> (2)
- *M* docs/tutorial/interfaces.md
<https://github.com/pmgysel/caffe/pull/7/files#diff-11> (2)
- *M* docs/tutorial/layers.md
<https://github.com/pmgysel/caffe/pull/7/files#diff-12> (1)
- *A* docs/tutorial/layers/clip.md
<https://github.com/pmgysel/caffe/pull/7/files#diff-13> (20)
- *A* include/caffe/layers/clip_layer.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-14> (75)
- *M* include/caffe/layers/pooling_layer.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-15> (1)
- *M* include/caffe/net.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-16> (6)
- *M* include/caffe/sgd_solvers.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-17> (5)
- *M* include/caffe/solver.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-18> (7)
- *M* include/caffe/syncedmem.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-19> (4)
- *M* include/caffe/util/hdf5.hpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-20> (2)
- *M* include/caffe/util/signal_handler.h
<https://github.com/pmgysel/caffe/pull/7/files#diff-21> (2)
- *M* python/caffe/_caffe.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-22> (22)
- *M* python/caffe/test/test_solver.py
<https://github.com/pmgysel/caffe/pull/7/files#diff-23> (11)
- *M* python/train.py
<https://github.com/pmgysel/caffe/pull/7/files#diff-24> (2)
- *M* src/caffe/layer_factory.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-25> (1)
- *A* src/caffe/layers/clip_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-26> (51)
- *A* src/caffe/layers/clip_layer.cu
<https://github.com/pmgysel/caffe/pull/7/files#diff-27> (67)
- *M* src/caffe/layers/hdf5_data_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-28> (2)
- *M* src/caffe/layers/hdf5_data_layer.cu
<https://github.com/pmgysel/caffe/pull/7/files#diff-29> (2)
- *M* src/caffe/layers/hdf5_output_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-30> (2)
- *M* src/caffe/layers/hdf5_output_layer.cu
<https://github.com/pmgysel/caffe/pull/7/files#diff-31> (2)
- *M* src/caffe/layers/inner_product_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-32> (2)
- *M* src/caffe/layers/pooling_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-33> (23)
- *M* src/caffe/net.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-34> (21)
- *M* src/caffe/proto/caffe.proto
<https://github.com/pmgysel/caffe/pull/7/files#diff-35> (23)
- *M* src/caffe/solver.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-36> (10)
- *M* src/caffe/solvers/sgd_solver.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-37> (16)
- *M* src/caffe/test/test_filler.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-38> (8)
- *M* src/caffe/test/test_hdf5_output_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-39> (2)
- *M* src/caffe/test/test_hdf5data_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-40> (2)
- *M* src/caffe/test/test_neuron_layer.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-41> (61)
- *M* src/caffe/test/test_syncedmem.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-42> (4)
- *M* src/caffe/util/hdf5.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-43> (2)
- *M* src/caffe/util/signal_handler.cpp
<https://github.com/pmgysel/caffe/pull/7/files#diff-44> (2)
Patch Links:
- https://github.com/pmgysel/caffe/pull/7.patch
- https://github.com/pmgysel/caffe/pull/7.diff
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#7>, or mute the thread
<https://github.com/notifications/unsubscribe-auth/AN-KWVY7htf_G-wY1KuSB6PO1t0GhqP3ks5uuOOsgaJpZM4YY3se>
.
--
*Philipp Gysel*
MSc Graduate, ECE
University of California, Davis
Homepage <http://lepsucd.com/?page_id=327>
|
Author
|
Hi Philipp, Apart from the official Caffe changes, the only changes I made were a few relative numbering of layer parameters in caffe.proto to match the number as if this repo was rebased on top of the official Caffe repo. Best, |
Owner
|
Ok sounds great thanks! |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
15 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.