Jeff Donahue [Tue, 26 Jan 2016 17:55:08 +0000 (09:55 -0800)]
Merge pull request #3496 from jeffdonahue/fix-testdatatransformer-leaks
TestDataTransformer: fix some memory leaks
Jeff Donahue [Tue, 26 Jan 2016 17:38:20 +0000 (09:38 -0800)]
Merge pull request #3581 from Austriker/draw-python3
Updated draw.py to make it work under Python 3
Hugo Serrat [Thu, 21 Jan 2016 13:34:01 +0000 (14:34 +0100)]
Updated import to make it work with pydotplus
Ronghang Hu [Sat, 23 Jan 2016 20:15:46 +0000 (12:15 -0800)]
Merge pull request #3593 from ronghanghu/matcaffe-versioning
show Caffe's version from MatCaffe
Ronghang Hu [Sat, 23 Jan 2016 09:22:04 +0000 (01:22 -0800)]
show Caffe's version from MatCaffe
Evan Shelhamer [Sat, 23 Jan 2016 02:54:17 +0000 (18:54 -0800)]
Merge pull request #3311 from lukeyeager/bvlc/versioning
Add versioning to build and mark version 1.0.0-rc3
Philipp Krähenbühl [Sat, 23 Jan 2016 00:59:01 +0000 (16:59 -0800)]
Merge pull request #3584 from graphaelli/python3-config
Add makefile config option for linking Python 3 libraries
Luke Yeager [Fri, 22 Jan 2016 23:30:35 +0000 (15:30 -0800)]
Version 1.0.0-rc3
Evan Shelhamer [Fri, 22 Jan 2016 18:49:48 +0000 (10:49 -0800)]
Merge pull request #3388 from mohomran/exponential_linear_units
Exponential Linear Units
Robbie Cooper [Thu, 21 Jan 2016 19:11:00 +0000 (14:11 -0500)]
Add makefile config option for linking Python 3 libraries
Jon Long [Wed, 20 Jan 2016 21:38:10 +0000 (13:38 -0800)]
Merge pull request #3536 from intelcaffe/im2col-speedup
Performance related update of im2col() and col2im() functions
Evan Shelhamer [Wed, 20 Jan 2016 18:09:59 +0000 (10:09 -0800)]
Merge pull request #3574 from tahtguymike/maxpool_bug
Substitute Caffe engine for cuDNN max pooling
as workaround for in-place after max pooling issue
thatguymike [Wed, 20 Jan 2016 01:01:34 +0000 (17:01 -0800)]
Workaround for inplace max pooling issue
Mariusz Moczala [Wed, 20 Jan 2016 08:28:35 +0000 (09:28 +0100)]
Performance related update of im2col() and col2im() functions
Jon Long [Fri, 8 Jan 2016 02:48:55 +0000 (18:48 -0800)]
Merge pull request #3525 from philkr/cmake_python3
CMake python version fix
Jon Long [Fri, 8 Jan 2016 02:42:40 +0000 (18:42 -0800)]
Merge pull request #2865 from philkr/top_bottom_names
Exposing layer top and bottom names to python
philkr [Wed, 6 Jan 2016 15:23:35 +0000 (07:23 -0800)]
CMake python version fix
Jeff Donahue [Wed, 6 Jan 2016 05:20:54 +0000 (21:20 -0800)]
Merge pull request #3519 from philkr/faster_solver
Speeding up ADAM (and maybe other solvers too)
philkr [Tue, 5 Jan 2016 20:45:52 +0000 (12:45 -0800)]
Speeding up the GPU solvers
philkr [Wed, 5 Aug 2015 18:54:08 +0000 (11:54 -0700)]
Exposing layer top and bottom names to python
Jeff Donahue [Wed, 30 Dec 2015 05:10:14 +0000 (21:10 -0800)]
TestDataTransformer: fix some memory leaks caused by use of 'new'
Jeff Donahue [Tue, 29 Dec 2015 06:27:34 +0000 (22:27 -0800)]
Merge pull request #3490 from fyu/spaceplus
remove extra space before +
Fisher Yu [Tue, 29 Dec 2015 03:46:49 +0000 (22:46 -0500)]
remove extra space before +
Jon Long [Mon, 28 Dec 2015 23:56:46 +0000 (15:56 -0800)]
Merge pull request #3487 from longjon/dilation
An early implementation of this functionality for Caffe was written by @gpapan, which was extended and improved by @tamakoji in a previous implementation of this branch.
Jonathan L Long [Mon, 28 Dec 2015 04:56:24 +0000 (20:56 -0800)]
enable dilated deconvolution
Since the underlying routines are shared, we need only upgrade
compute_output_shape.
Jonathan L Long [Sat, 26 Dec 2015 21:10:11 +0000 (13:10 -0800)]
add short description of dilation to caffe.proto
Fisher Yu [Sat, 26 Dec 2015 21:10:02 +0000 (13:10 -0800)]
disable dilated deconvolution
Fisher Yu [Sat, 26 Dec 2015 21:04:25 +0000 (13:04 -0800)]
add and improve tests for dilated convolution/im2col
Fisher Yu [Mon, 28 Dec 2015 04:48:30 +0000 (20:48 -0800)]
add support for N-D dilated convolution
Fisher Yu [Mon, 7 Dec 2015 01:04:43 +0000 (20:04 -0500)]
add support for 2D dilated convolution
Jon Long [Sun, 20 Dec 2015 23:49:58 +0000 (15:49 -0800)]
Merge pull request #3468 from mnogu/blobs_lr-to-lr_mult
Replace blobs_lr with lr_mult in readme.md.
Muneyuki Noguchi [Sun, 20 Dec 2015 10:12:09 +0000 (19:12 +0900)]
Replace blobs_lr with lr_mult in readme.md.
models/finetune_flickr_style/deploy.prototxt uses lr_mult now.
Evan Shelhamer [Sat, 12 Dec 2015 04:00:00 +0000 (01:00 -0300)]
Merge pull request #3439 from flx42/cudnn_v4
Add ifdef in CuDNNConvolutionLayer for cuDNN v4
Felix Abecassis [Thu, 10 Dec 2015 23:11:51 +0000 (15:11 -0800)]
Fix CuDNNConvolutionLayer for cuDNN v4
Add a macro to check the current cuDNN version
Evan Shelhamer [Wed, 9 Dec 2015 20:31:45 +0000 (12:31 -0800)]
Merge pull request #3432 from ianfhunter/patch-1
[docs] fix typo in interfaces.md
Ian Hunter [Wed, 9 Dec 2015 13:29:01 +0000 (13:29 +0000)]
Update interfaces.md
typo
Yangqing Jia [Wed, 9 Dec 2015 01:38:00 +0000 (17:38 -0800)]
Merge pull request #3090 from longjon/summarize-tool
A Python script for at-a-glance net summary
Jon Long [Fri, 4 Dec 2015 21:11:30 +0000 (13:11 -0800)]
Merge pull request #3395 from BonsaiAI/extract_features-compatibility
Fix compatibility issues with extract_features
Mohamed Omran [Thu, 26 Nov 2015 00:46:42 +0000 (01:46 +0100)]
ELU layer with basic tests
T.E.A de Souza [Sun, 29 Nov 2015 06:24:09 +0000 (14:24 +0800)]
Correct type of device_id; disambiguate shared_ptr
Jon Long [Wed, 2 Dec 2015 08:11:26 +0000 (00:11 -0800)]
Merge pull request #3404 from BonsaiAI/remove-hamming-dist
Remove hamming_distance and popcount
Tea [Wed, 2 Dec 2015 07:39:19 +0000 (15:39 +0800)]
Remove hamming_distance and popcount
Evan Shelhamer [Wed, 2 Dec 2015 06:17:36 +0000 (22:17 -0800)]
Merge pull request #3313 from gdh1995/master
don't divide by 0 duration when downloading model binary
Evan Shelhamer [Wed, 2 Dec 2015 06:05:37 +0000 (22:05 -0800)]
Merge pull request #3267 from cbalint13/cmake-cudnn
[build] Display and store cuDNN version numbers for CMake
Evan Shelhamer [Wed, 2 Dec 2015 06:04:34 +0000 (22:04 -0800)]
Merge pull request #3285 from longjon/cuda-dead-cpp
Remove dead preprocessor code for number of CUDA threads
Evan Shelhamer [Wed, 2 Dec 2015 05:30:35 +0000 (21:30 -0800)]
Merge pull request #3315 from shelhamer/layer-headers
Dismantle Layer Headers
Evan Shelhamer [Tue, 10 Nov 2015 08:22:58 +0000 (00:22 -0800)]
dismantle layer headers
No more monolithic includes: split layers into their own headers for modular inclusion and build.
Ronghang Hu [Sat, 28 Nov 2015 04:53:46 +0000 (20:53 -0800)]
Merge pull request #3378 from BonsaiAI/fix-temp-dir-creation-failure
Safely create temporary files and directories
Tea [Wed, 25 Nov 2015 03:43:45 +0000 (11:43 +0800)]
Secure temporary file creation
T.E.A de Souza [Tue, 24 Nov 2015 06:33:27 +0000 (14:33 +0800)]
Secure implementation of MakeTempDir
Ronghang Hu [Fri, 27 Nov 2015 19:29:15 +0000 (11:29 -0800)]
Merge pull request #3320 from BonsaiAI/disambiguate-dtype
Cast std::max args to Dtype
Ronghang Hu [Fri, 27 Nov 2015 19:25:52 +0000 (11:25 -0800)]
Merge pull request #2693 from AdamStelmaszczyk/patch-1
Deprecated OpenCV consts leading to compilation error
Ronghang Hu [Fri, 27 Nov 2015 18:50:52 +0000 (10:50 -0800)]
Merge pull request #3361 from BonsaiAI/avoid-snprintf
replace snprintf with a C++98 equivalent
Ronghang Hu [Thu, 26 Nov 2015 18:45:19 +0000 (10:45 -0800)]
Merge pull request #3352 from lukeyeager/bvlc/pytest-python-layer
Skip python layer tests if WITH_PYTHON_LAYER unset
Ronghang Hu [Thu, 26 Nov 2015 18:34:09 +0000 (10:34 -0800)]
Merge pull request #3389 from ronghanghu/remove-stepearly
Remove bogus stepearly in MNIST example
Ronghang Hu [Thu, 26 Nov 2015 05:02:02 +0000 (21:02 -0800)]
Remove bogus stepearly in MNIST example
This `examples/lenet/lenet_stepearly_solver.prototxt` is introduced in #190 by mistake, since stepearly is never actually merged.
Tea [Tue, 17 Nov 2015 09:05:56 +0000 (17:05 +0800)]
replace snprintf with a C++98 equivalent
Jeff Donahue [Mon, 23 Nov 2015 03:33:36 +0000 (19:33 -0800)]
Merge pull request #3296 from cdoersch/normalize_batch
Better normalization options for SoftmaxWithLoss layer
Carl Doersch [Fri, 6 Nov 2015 22:41:30 +0000 (14:41 -0800)]
Better normalization options for SoftmaxWithLoss layer.
Ronghang Hu [Fri, 20 Nov 2015 11:08:03 +0000 (03:08 -0800)]
Merge pull request #3363 from BonsaiAI/remove-opencv-include
Exclude core.hpp when building without OpenCV
Ronghang Hu [Fri, 20 Nov 2015 10:51:31 +0000 (02:51 -0800)]
Merge pull request #3362 from eelstork/patch-2
GetDB must return a value.
T.E.A de Souza [Fri, 20 Nov 2015 08:52:25 +0000 (16:52 +0800)]
Exclude core.hpp when building without OpenCV
T.E.A de Souza [Fri, 20 Nov 2015 08:36:29 +0000 (16:36 +0800)]
Function must return a value
Currently compilation will fail with some compilers when LevelDB and LMDB are disabled.
Very similar to a recently fixed issue.
Tea [Thu, 12 Nov 2015 07:15:22 +0000 (15:15 +0800)]
Convert std::max args to Dtype
Evan Shelhamer [Thu, 19 Nov 2015 20:07:32 +0000 (12:07 -0800)]
Merge pull request #3359 from ronghanghu/fix-accuracy-layer-top
Fix MaxTopBlobs in Accuracy Layer
Ronghang Hu [Thu, 19 Nov 2015 18:05:48 +0000 (10:05 -0800)]
Fix MaxTopBlobs in Accuracy Layer
Fix the typo "MaxTopBlos" to "MaxTopBlobs". This typo causes maximum top number to be incorrect.
Luke Yeager [Wed, 18 Nov 2015 18:38:32 +0000 (10:38 -0800)]
Skip python layer tests if WITH_PYTHON_LAYER unset
Evan Shelhamer [Mon, 16 Nov 2015 07:19:58 +0000 (23:19 -0800)]
Merge pull request #3321 from BonsaiAI/synced-mem-functions-will-return-a-value
In syncedmem.cpp, functions shall return a value
Evan Shelhamer [Mon, 16 Nov 2015 07:00:44 +0000 (23:00 -0800)]
Merge pull request #3261 from kli-nlpr/patch-1
fix plot_training_log.py.example usage text
Evan Shelhamer [Mon, 16 Nov 2015 06:51:51 +0000 (22:51 -0800)]
Merge pull request #3332 from alexlee-gk/master
[bug] fix outs and diffs being overwritten in forward_backward_all
Alex Lee [Sat, 14 Nov 2015 20:49:05 +0000 (12:49 -0800)]
Fix outs and diffs being overwritten in forward_backward_all.
Evan Shelhamer [Sat, 14 Nov 2015 19:53:26 +0000 (11:53 -0800)]
Merge pull request #3299 from kkhoot/fix_bn
Update BatchNormLayer: more numerical stability and backward with global stats
Evan Shelhamer [Sat, 14 Nov 2015 19:49:00 +0000 (11:49 -0800)]
Merge pull request #3323 from BonsaiAI/remove-unnecessary-includes
Remove un-necessary includes
Balint Cristian [Fri, 13 Nov 2015 11:58:49 +0000 (13:58 +0200)]
Display and store cuDNN version numbers during cmake.
Evan Shelhamer [Fri, 13 Nov 2015 08:25:12 +0000 (00:25 -0800)]
Merge pull request #3328 from w1res/patch-1
[doc] Add parentheses to backward_{cpu,gpu} method.
Adam Siembida [Thu, 12 Nov 2015 21:03:41 +0000 (16:03 -0500)]
Add parentheses to backward_{cpu,gpu} method.
Kang Kim [Thu, 12 Nov 2015 17:20:02 +0000 (02:20 +0900)]
Make backward pass work when global stats is active for BatchNormLayer
including minor code cleaning
Kang Kim [Sat, 7 Nov 2015 03:49:15 +0000 (12:49 +0900)]
Update computation of variance and global stats in BatchNormLayer
Tea [Thu, 12 Nov 2015 07:55:06 +0000 (15:55 +0800)]
Functions shall return a value in syncedmem
T.E.A de Souza [Thu, 12 Nov 2015 07:32:28 +0000 (15:32 +0800)]
Remove un-necessary includes
gdh1995 [Tue, 10 Nov 2015 14:41:55 +0000 (22:41 +0800)]
fix a bug that time duration may be 0 when downloading model binary
Evan Shelhamer [Wed, 11 Nov 2015 06:53:57 +0000 (22:53 -0800)]
Merge pull request #3127 from lukeyeager/install-lib-permissions
[build] Install libs as non-executable files
Evan Shelhamer [Wed, 11 Nov 2015 06:51:25 +0000 (22:51 -0800)]
Merge pull request #3235 from shelhamer/dlog-cudnn-workspace
cuDNN: only log conv workspace in debug mode
Evan Shelhamer [Wed, 11 Nov 2015 06:50:28 +0000 (22:50 -0800)]
Merge pull request #3290 from ernest-tg/master
[pycaffe] correct transposition & channel_swap in deprocess
Evan Shelhamer [Wed, 11 Nov 2015 06:48:23 +0000 (22:48 -0800)]
Merge pull request #3294 from panmari/simpler_diff
[example] switch order of two layers for simpler diff with untuned file
Evan Shelhamer [Wed, 11 Nov 2015 06:47:27 +0000 (22:47 -0800)]
Merge pull request #3295 from timmeinhardt/fix_issue_3274
[bug] fix issue #3274 -- shape argmax top carefully
Evan Shelhamer [Wed, 11 Nov 2015 06:44:46 +0000 (22:44 -0800)]
Merge pull request #3310 from gustavla/contrastive-doc-fix
[doc] Fix consistent typo in contrastive loss
Ronghang Hu [Tue, 10 Nov 2015 23:08:47 +0000 (15:08 -0800)]
Merge pull request #3305 from ronghanghu/display-ignore-layer
display 'ignore source layer' when initializing from existing parameters
Ronghang Hu [Tue, 10 Nov 2015 23:08:22 +0000 (15:08 -0800)]
Merge pull request #3300 from BonsaiAI/replace-unistd
Replace unistd functions with cross platform counterparts
Tea [Sat, 7 Nov 2015 06:09:59 +0000 (14:09 +0800)]
Replace unistd functions with cross platform counterparts
Gustav Larsson [Mon, 9 Nov 2015 20:32:37 +0000 (14:32 -0600)]
DOC: Fix consistent typo in contrastive loss
If a pair is similar, it should take the squared distance and not the
distance. This is clearly what the code is doing.
Ronghang Hu [Mon, 9 Nov 2015 18:15:26 +0000 (10:15 -0800)]
Merge pull request #3308 from cassinaj/master
minor typo fix
Jan Issac [Mon, 9 Nov 2015 11:42:56 +0000 (12:42 +0100)]
Merge pull request #1 from cassinaj/cassinaj-minor-typo-fix
minor typo fix
panmari [Fri, 6 Nov 2015 11:53:16 +0000 (12:53 +0100)]
Switched order of two layers for simpler diff with untuned file
Untuned file is in models/bvlc_reference_caffenet/train_val.prototxt.
Ronghang Hu [Sun, 8 Nov 2015 19:51:24 +0000 (11:51 -0800)]
Merge pull request #3303 from eelstork/patch-1
GetDB must return a value
Ronghang Hu [Sun, 8 Nov 2015 19:20:32 +0000 (11:20 -0800)]
display 'ignore source layer' when initializing from existing parameters
This helps in the case to see which layer is initialized from existing parameters, and which layer is ignored. This helps identify the cases where the user types a error mismatch layer name.
T.E.A de Souza [Sun, 8 Nov 2015 10:50:29 +0000 (18:50 +0800)]
GetDB must return a value
As noted by @danst18, when USE_LEVELDB and USE_LMDB are disabled, a compiler error is issued since GetDB no longer returns a value.
At runtime a fatal error would be issued anyways. However to help users who don't need a DB backend, NULL should be returned here.
Tim Meinhardt [Fri, 6 Nov 2015 13:51:46 +0000 (14:51 +0100)]
Fix ArgMaxLayer::Reshape for any num of bottom axes
ernest-tg [Thu, 5 Nov 2015 14:47:28 +0000 (15:47 +0100)]
Correct transposition & channel_swap in deprocess
The deprocess( ) function should invert the preprocess( ) function, however it only worked when the permutation of your channel_swap is of order 2 and the permutation of your transpose were of order 3. This is usually the case, which is why this bug went unnoticed for a long time.
To reproduce it (on former version), try to preprocess and then deprocess with
transformer.set_transpose('data', (0,2,1)) (or (1,0,2) or (2,1,0))
Or with
transformer.set_channel_swap('data', (2,0,1)) (or (1,2,0) )
Indeed, we had
L152 (in preprocess) caffe_in = caffe_in[channel_swap, :, :]
L181 (in deprocess) decaf_in = decaf_in[channel_swap, :, :]
So we applied [channel_swap,:,:] twice to the initial data => not always the identity
L154 (in preprocess) caffe_in = caffe_in.transpose(transpose)
L183 (in deprocess) decaf_in = decaf_in.transpose([transpose[t] for t in transpose])
The transposition [transpose[t] for t in transpose] is (tranpsose)² so we applied transpose[t] three times which is not always the identity.
Jon Long [Thu, 5 Nov 2015 08:27:56 +0000 (00:27 -0800)]
Merge pull request #3287 from longjon/travis-install-whitespace
Fix whitespace in travis_install.sh