Evan Shelhamer [Tue, 19 Apr 2016 00:51:52 +0000 (17:51 -0700)]
Merge pull request #4007 from lukeyeager/bvlc/docs-typo
[docs] typo in docs/installation.md
Luke Yeager [Mon, 18 Apr 2016 15:54:21 +0000 (08:54 -0700)]
Typo in docs/installation.md
Evan Shelhamer [Fri, 15 Apr 2016 08:02:02 +0000 (01:02 -0700)]
Merge pull request #3993 from shelhamer/fix-crop
Fix Crop layer dimension checking to only check cropped dimensions
Evan Shelhamer [Fri, 15 Apr 2016 05:31:38 +0000 (22:31 -0700)]
CropLayer: groom comments
Evan Shelhamer [Fri, 15 Apr 2016 05:16:07 +0000 (22:16 -0700)]
[fix] CropLayer: check dimension bounds only for cropped dimensions
check only the dimensions to be cropped for compatible sizes and offsets
Evan Shelhamer [Fri, 15 Apr 2016 05:56:37 +0000 (22:56 -0700)]
[test] CropLayer: test dimensions check to reveal bounds checking bug
Evan Shelhamer [Thu, 14 Apr 2016 09:19:27 +0000 (02:19 -0700)]
Merge pull request #3988 from shelhamer/install-docs
bring install docs up-to-date and better highlight included details
Evan Shelhamer [Thu, 14 Apr 2016 01:53:28 +0000 (18:53 -0700)]
[docs] install: include more lab tested hardware
Evan Shelhamer [Thu, 14 Apr 2016 01:52:50 +0000 (18:52 -0700)]
[docs] install: be more firm about compute capability >= 3.0
Evan Shelhamer [Wed, 13 Apr 2016 23:43:39 +0000 (16:43 -0700)]
[docs] install: include latest versions and platforms, highlight guides
Caffe runs on Ubuntu, OS X, and RHEL (+ company) in master
with branches for OpenCL and Windows.
Docker is a nice route to out-of-the-box brewing.
Evan Shelhamer [Wed, 13 Apr 2016 23:40:30 +0000 (16:40 -0700)]
[docs] install: CUDA 7+ and cuDNN v4 compatible
Latest CUDA versions are all compatible, and
Caffe has been compatible with cuDNN v4 since PR #3439
Evan Shelhamer [Wed, 13 Apr 2016 01:18:26 +0000 (18:18 -0700)]
Merge pull request #3982 from mnogu/fix-typo-model-option
[docs] typo in help text for "-model" option
Muneyuki Noguchi [Tue, 12 Apr 2016 14:19:27 +0000 (23:19 +0900)]
Fix typo in help text for "-model" option
Jeff Donahue [Fri, 8 Apr 2016 22:07:20 +0000 (15:07 -0700)]
Merge pull request #3937 from emaggiori/exp
Issue with the exp layer when the base is e
emmanuel maggiori [Fri, 8 Apr 2016 08:25:12 +0000 (10:25 +0200)]
Solving issue with exp layer with base e
Jeff Donahue [Tue, 5 Apr 2016 17:15:16 +0000 (10:15 -0700)]
Merge pull request #3942 from jeffdonahue/propagate-down-true
Set propagate_down=true to force backprop to a particular bottom
Jeff Donahue [Wed, 27 Jan 2016 20:55:41 +0000 (12:55 -0800)]
Net: setting `propagate_down: true` forces backprop
Jeff Donahue [Mon, 4 Apr 2016 18:36:15 +0000 (11:36 -0700)]
test_net.cpp: add TestForcePropagateDown
Jon Long [Fri, 1 Apr 2016 20:30:54 +0000 (13:30 -0700)]
Merge pull request #3891 from danielgordon10/pycaffe-multi-instantiation-fix
Use lazy initialization to reuse ordered dict/list creations to save…
Daniel Gordon [Wed, 30 Mar 2016 21:27:19 +0000 (14:27 -0700)]
Use lazy initialization to reuse orderd dict/list creations to save time on repeated calls.
Evan Shelhamer [Tue, 29 Mar 2016 21:01:00 +0000 (14:01 -0700)]
Merge pull request #3821 from jreniecki/mkl
[bulid] correct MKL_DIR to MKLROOT
JacekR [Tue, 15 Mar 2016 09:43:34 +0000 (10:43 +0100)]
Update Makefile: Changed MKL_DIR to MKLROOT
MKLROOT variable is set by MKL scripts, so it also should be used in Makefile.
Evan Shelhamer [Wed, 9 Mar 2016 20:42:24 +0000 (12:42 -0800)]
Merge pull request #3797 from shelhamer/travis-rm-conda
Fix Travis-CI conda/boost issue by workaround: remove conda dir
Andy Feng [Wed, 9 Mar 2016 18:39:31 +0000 (18:39 +0000)]
[build] travis: remove existing conda dir
there seems to be a caching issue at the moment;
this is a temporary fix for #3786
Evan Shelhamer [Mon, 7 Mar 2016 21:55:48 +0000 (13:55 -0800)]
Merge pull request #3292 from dgolden1/log-parse-no-test
parse_log.py bugfix: don't attempt to write CSV if there are no lines to write
Jon Long [Sun, 6 Mar 2016 03:32:14 +0000 (19:32 -0800)]
Merge pull request #3590 from junshi15/GPUUtilities
Add functions to check and grab GPU
Jon Long [Sun, 6 Mar 2016 03:17:14 +0000 (19:17 -0800)]
Merge pull request #3588 from junshi15/P2psyncPrepare
Refine P2PSync
Evan Shelhamer [Sat, 5 Mar 2016 19:35:56 +0000 (11:35 -0800)]
Merge pull request #3613 from longjon/py-coord-map
Python/net spec coordinate map and crop offset computation
Evan Shelhamer [Sat, 5 Mar 2016 19:34:50 +0000 (11:34 -0800)]
Merge pull request #3570 from BlGene/crop-nd
ND Crop layer
Jun Shi [Fri, 22 Jan 2016 17:58:37 +0000 (09:58 -0800)]
split p2psync::run()
Evan Shelhamer [Fri, 4 Mar 2016 10:20:13 +0000 (02:20 -0800)]
Crop: more tests and test tuning.
Changes are:
reduce test blob dims for speed
use standard Gaussian filler,
polish formatting and rename tests,
test HW crop and 5D crop,
standard gradient checks
max argus [Mon, 29 Feb 2016 11:24:25 +0000 (11:24 +0000)]
Crop: fixes, tests and negative axis indexing.
max argus [Tue, 19 Jan 2016 18:35:04 +0000 (18:35 +0000)]
Extend Crop to N-D, changed CropParameter.
Jonathan L Long [Sat, 27 Dec 2014 09:44:36 +0000 (01:44 -0800)]
add CropLayer: crop blob to another blob's dimensions with offsets
configure offset(s) through proto definition.
Jun Shi [Fri, 22 Jan 2016 17:50:31 +0000 (09:50 -0800)]
add check and find GPU device utilities
Evan Shelhamer [Fri, 4 Mar 2016 00:39:18 +0000 (16:39 -0800)]
[pycaffe] test coord_map
- test known mappings: conv-pool-deconv stack, ReLU and 1x1 conv
- test effects of padding
- test rectangular/anisotropic coordinate mapping, test N-D
- catch error cases: negative crop, scale mismatch, tops that are not
spatially connected
Evan Shelhamer [Sun, 28 Feb 2016 19:51:52 +0000 (11:51 -0800)]
[pycaffe] align coord_map and #3570 Crop layer
- crop -> offset
- adjust crop axis by 1
Evan Shelhamer [Sun, 28 Feb 2016 07:57:42 +0000 (23:57 -0800)]
[pycaffe] document, style, and complete coord_map
- document by docstring and comment
- pep8
- add latest layers and alphabetize
- respect default crop params
- handle graphs with compositions of crops by walking only the
first, cropped bottom of Crop layers
- make python3 happy by replacing arg tuple unpacking
Evan Shelhamer [Sat, 5 Mar 2016 02:53:21 +0000 (18:53 -0800)]
Merge pull request #3770 from BlGene/lint_fix
Removed lint script reference to non-existant caffe_memcpy function.
Jonathan L Long [Sat, 30 Jan 2016 02:41:12 +0000 (18:41 -0800)]
[pycaffe] add coord_map.py for computing induced coordinate transform
This provides a framework for automatically aligning different layers of
a net despite up/downsampling, padding, and output size rounding.
Jon Long [Fri, 4 Mar 2016 21:58:03 +0000 (13:58 -0800)]
Merge pull request #3773 from longjon/travis-protobuf3-fix
Travis: force protobuf 3.0.0b2 for Python 3
Jonathan L Long [Fri, 4 Mar 2016 20:53:49 +0000 (12:53 -0800)]
[travis] force protobuf 3.0.0b2 for Python 3
This is temporary measure to avoid an apparent upstream issue with
protobuf 3.0.0b2.post1.
max argus [Fri, 4 Mar 2016 01:59:48 +0000 (01:59 +0000)]
Removed lint script reference to non-existant caffe_memcpy function.
Evan Shelhamer [Fri, 4 Mar 2016 00:42:00 +0000 (16:42 -0800)]
Merge pull request #3575 from errord/fix-boost-shared_ptr-caffe-Blob-float-no-register
[pycaffe] fix boost 1.60 compatibility issue; close #3494
Jon Long [Thu, 3 Mar 2016 00:23:14 +0000 (16:23 -0800)]
Merge pull request #3716 from ttdt/master
Use six library to ensure pycaffe.py python3 compliance
Jeff Donahue [Wed, 2 Mar 2016 05:11:22 +0000 (21:11 -0800)]
Merge pull request #3759 from vivkul/patch-1
minor mistakes removed
Viveka Kulharia [Wed, 2 Mar 2016 05:07:56 +0000 (10:37 +0530)]
minor mistakes removed
Evan Shelhamer [Tue, 1 Mar 2016 23:41:33 +0000 (15:41 -0800)]
[example] groom multilabel notebook title, order
Evan Shelhamer [Tue, 1 Mar 2016 20:18:57 +0000 (12:18 -0800)]
Merge pull request #3755 from shelhamer/fix-upgrade-proto
Fix Upgrade Net Tools
Evan Shelhamer [Tue, 1 Mar 2016 18:29:15 +0000 (10:29 -0800)]
Merge pull request #3471 from beijbom/clean-datalayer-tutorial
[example] tutorial on python data layers and multilabel classification
Evan Shelhamer [Tue, 1 Mar 2016 18:00:33 +0000 (10:00 -0800)]
Merge pull request #3756 from intelcaffe/mkl-cosmetic-imprvmnts
[build] MKL support improvements
Jacek Czaja [Tue, 1 Mar 2016 10:32:21 +0000 (11:32 +0100)]
- doc and cmake update MKL related
- cosmetic change to mkl related doc
Evan Shelhamer [Tue, 1 Mar 2016 05:18:50 +0000 (21:18 -0800)]
refuse to upgrade net with layer/layers inconsistency
die loudly if a net definition (prototxt) mixes proto formats by
defining both `layer` and `layers` fields instead of complaining but
discarding and continuing.
fix #3381
Evan Shelhamer [Tue, 1 Mar 2016 05:17:21 +0000 (21:17 -0800)]
fix input field -> input layer net upgrade: only convert full defs
convert inputs in legacy definitions (prototxt), but simply strip inputs
from legacy weights (caffemodel).
fix #3750
Evan Shelhamer [Tue, 1 Mar 2016 05:15:56 +0000 (21:15 -0800)]
check all net upgrade conditions
check all conditions all the time; V0 -> V1 and V1 -> V2 do not suffice.
Evan Shelhamer [Tue, 1 Mar 2016 05:13:27 +0000 (21:13 -0800)]
output all logging from upgrade net tools
output info, warnings, and errors for fuller description of the upgrade
Oscar Beijbom [Sat, 27 Feb 2016 03:17:21 +0000 (19:17 -0800)]
Finalized tutorial. Removed asyncronous layer.
Evan Lezar [Thu, 18 Feb 2016 10:50:15 +0000 (11:50 +0100)]
Refactor and improve code style.
Fix some typos. Correct imports. Refactor data layers. Apply PEP8 formatting.
Oscar Beijbom [Mon, 21 Dec 2015 07:02:42 +0000 (23:02 -0800)]
Added tutorial on how to use python datalayers and multilabel classification.
Evan Shelhamer [Mon, 29 Feb 2016 17:13:01 +0000 (09:13 -0800)]
Merge pull request #3747 from Nerei/master
[pycaffe] NetSpec: set tops by attribute or dictionary
Anatoly Baksheev [Mon, 29 Feb 2016 15:01:07 +0000 (18:01 +0300)]
NetSpec: allow setting blob names by string
Thibault Deregnaucourt [Mon, 29 Feb 2016 09:21:07 +0000 (10:21 +0100)]
Use 'six' library to ensure python3 compliance.
Use '//' instead of '/' for entire division.
Jeff Donahue [Sun, 28 Feb 2016 08:31:17 +0000 (00:31 -0800)]
Merge pull request #3725 from shaibagon/drop_nd_blobs
supporting N-D Blobs in Dropout layer Reshape
shai [Thu, 25 Feb 2016 10:09:45 +0000 (12:09 +0200)]
supporting N-D Blobs in Dropout layer Reshape
fixing lint errors
Evan Shelhamer [Sat, 27 Feb 2016 20:12:24 +0000 (12:12 -0800)]
Merge pull request #3740 from shelhamer/fix-docker-flags
Fix flags for nvidia-docker from #3518
Evan Shelhamer [Sat, 27 Feb 2016 20:10:16 +0000 (12:10 -0800)]
fix flags in #3518 for nvidia-docker
nvidia-docker requires long args with equal sign as of docker 1.10:
see https://github.com/BVLC/caffe/pull/3518#issuecomment-
189576419
Evan Shelhamer [Sat, 27 Feb 2016 20:08:39 +0000 (12:08 -0800)]
Merge pull request #3518 from zalando/feature/docker_images
[build] Add docker images for running caffe out-of-the-box (caffe:cpu, caffe:gpu)
Evan Lezar [Tue, 5 Jan 2016 15:27:28 +0000 (16:27 +0100)]
Add Dockerfiles for creating Caffe executable images.
These can be used as direct replacements for the Caffe executable.
Evan Shelhamer [Sat, 27 Feb 2016 07:20:39 +0000 (23:20 -0800)]
Merge pull request #3211 from shelhamer/input-layer
Add Input Layer to Replace `input`s
Evan Shelhamer [Sat, 27 Feb 2016 06:20:50 +0000 (22:20 -0800)]
Deprecate ForwardPrefilled(), Forward(bottom, loss) in lieu of dropping
Relax removal of `Forward()` variations by deprecating instead.
Evan Shelhamer [Fri, 4 Dec 2015 03:39:19 +0000 (19:39 -0800)]
[examples] switch examples + models to Input layers
Evan Shelhamer [Thu, 3 Dec 2015 02:31:26 +0000 (18:31 -0800)]
collect Net inputs from Input layers
Restore the list of net inputs for compatibility with the
pycaffe and matcaffe interfaces and downstream C++.
Evan Shelhamer [Sat, 17 Oct 2015 04:11:32 +0000 (21:11 -0700)]
drop Net inputs + Forward with bottoms
Drop special cases for `input` fields, the `Net` input members,
and the `Net` interface for Forward with bottoms along with
Forward() / ForwardPrefilled() distinction.
Evan Shelhamer [Fri, 16 Oct 2015 23:44:45 +0000 (16:44 -0700)]
deprecate input fields and upgrade automagically
Evan Shelhamer [Fri, 16 Oct 2015 23:44:23 +0000 (16:44 -0700)]
add InputLayer for Net input
Create an input layer to replace oddball Net `input` fields.
Jeff Donahue [Thu, 25 Feb 2016 19:17:19 +0000 (11:17 -0800)]
Merge pull request #3612 from kashefy/tied_weights_ip_transpose
Tied weights with transpose flag for InnerProduct layer
Evan Shelhamer [Thu, 25 Feb 2016 07:21:44 +0000 (23:21 -0800)]
Merge pull request #3715 from olesalscheider/master
CMake: Do not include "${PROJECT_BINARY_DIR}/include" with SYSTEM option
Evan Shelhamer [Thu, 25 Feb 2016 01:37:56 +0000 (17:37 -0800)]
Merge pull request #3703 from shaibagon/pycaffe_nd_blobs
[pycaffe] make pycaffe support N-D blobs throughout
Evan Shelhamer [Thu, 25 Feb 2016 00:41:12 +0000 (16:41 -0800)]
Merge pull request #3719 from shelhamer/new-tutorial-notebooks
[example] Improve Tutorial Notebooks on Classification, Learning LeNet, and Fine-tuning
Niels Ole Salscheider [Wed, 24 Feb 2016 16:00:58 +0000 (17:00 +0100)]
CMake: Do not include "${PROJECT_BINARY_DIR}/include" with SYSTEM option
This is important for the include order. Without this patch, a
previously installed caffe.pb.h might be included instead of the one
that is generated during the build.
Jeff Donahue [Wed, 24 Feb 2016 07:42:11 +0000 (23:42 -0800)]
[example] improve brewing logreg notebook
- create solvers inline through python protobuf
- drop manually written solver prototxt
- remove ordering prefix, since there is no real sequencing constraint
for this example
Jeff Donahue [Wed, 24 Feb 2016 07:43:31 +0000 (23:43 -0800)]
[example] improve fine-tuning notebook
- add headings and text detail
- make nets by net spec
- define solvers inline through python protobuf
- do two-stage fine-tuning (first last layer alone, then end-to-end)
- show sample results
Jonathan L Long [Wed, 24 Feb 2016 07:46:49 +0000 (23:46 -0800)]
[example] improve learning LeNet notebook
- add subheadings and list steps for structure
- edit text and comments for clarity
- switch paths and use chdir for idempotency of scripts [shelhamer]
- title accuracy plots, rename ip -> fc to fit common naming, and
rename output layer -> score [shelhamer]
- add experimentation section [shelhamer]
Jonathan L Long [Fri, 5 Feb 2016 03:41:55 +0000 (19:41 -0800)]
[data] get_mnist.sh rewrite; prevents prompt in tutorial notebooks
Jonathan L Long [Wed, 24 Feb 2016 07:34:46 +0000 (23:34 -0800)]
[example] improve classification notebook
- add subheadings and list steps
- edit text, add comments, and try to make the code more understandable
- add new section for summary and encouragement to try your own image
shai [Tue, 23 Feb 2016 08:42:54 +0000 (10:42 +0200)]
removing all references to Blob.num property (that assumes Blob is 4D). Replacing it with accessing Blob.shape[0] - for Blobs with num_axes() != 4
Evan Shelhamer [Sun, 21 Feb 2016 06:42:05 +0000 (22:42 -0800)]
Merge pull request #3696 from shelhamer/fix-osx-install-name
fix library install name on OS X for @rpath linking
Evan Shelhamer [Sun, 21 Feb 2016 06:39:30 +0000 (22:39 -0800)]
Merge pull request #3695 from shelhamer/fix-osx-elcap-cuda-link
Fix OS X El Capitan linking for CUDA
Mohamed Ezz [Fri, 5 Feb 2016 00:54:31 +0000 (01:54 +0100)]
Fix OSX El Capitan CUDA incompatibility, by adding lib to rpath
Evan Shelhamer [Sat, 20 Feb 2016 23:45:31 +0000 (15:45 -0800)]
fix library install name on OSX for relative path linking
for linking of the caffe tools, tests, etc. the library install name
needs to include the @rpath set for the executables and interfaces
this was broken on OSX by #3311
Youssef Kashef [Fri, 29 Jan 2016 18:21:48 +0000 (19:21 +0100)]
tranpose parameter added to IP layer to support tied weights in an autoencoder. Arguments to matrix multiplication function are conditioned on this parameter, no actual transposing takes place.
test ip gradient computation with transpose on
Jeff Donahue [Thu, 18 Feb 2016 07:30:12 +0000 (23:30 -0800)]
Merge pull request #3676 from flx42/leveldb_include_fix
Remove useless LevelDB include
Jeff Donahue [Thu, 18 Feb 2016 04:54:02 +0000 (20:54 -0800)]
Merge pull request #3687 from pra85/patch-2
Fix a typo in docs
Prayag Verma [Thu, 18 Feb 2016 04:43:34 +0000 (10:13 +0530)]
Fix a typo in docs
`peformance` → `performance`
Felix Abecassis [Tue, 16 Feb 2016 00:52:32 +0000 (16:52 -0800)]
Remove useless LevelDB include
The tests could not compile with USE_LEVELDB=0 and LevelDB missing from the system
Evan Shelhamer [Thu, 11 Feb 2016 21:19:50 +0000 (13:19 -0800)]
Merge pull request #3451 from intelcaffe/cmake-clang-fix
Fix to cmake build for clang on linux
Jeff Donahue [Tue, 9 Feb 2016 18:27:35 +0000 (10:27 -0800)]
Merge pull request #3650 from intbots/cmake-matlab-fix
bugfix for incorrect behaviour in `caffe_parse_linker_libs`
Abhijit Kundu [Tue, 9 Feb 2016 07:45:46 +0000 (02:45 -0500)]
bugfix for incorrect behaviour in caffe_parse_linker_libs function while extracting libflags from absolute library path with multiple (dots)
Jeff Donahue [Wed, 3 Feb 2016 06:05:07 +0000 (22:05 -0800)]
Merge pull request #3624 from drnikolaev/bvlc-print-gpu-names
Nicely prints gpu names
Sergei Nikolaev [Tue, 2 Feb 2016 21:48:18 +0000 (13:48 -0800)]
Nicely prints GPU names