This is an automated email from the ASF dual-hosted git repository.
wangwei pushed a change to branch SINGA-422
in repository https://gitbox.apache.org/repos/asf/incubator-singa.git.
at 0d6b1ae SINGA-422 ModuleNotFoundError: No module named '_singa_wrap'
This branch includes the following new commits:
new 95b1e6d migrating code from github to apache git repo; init commit +
test
new b2dc51d Transfer code from nusinga repo to singa apache repo. New
commuinication framework is implemented to unify the frameworks of existing
distributed deep learning systems. Communication is now implmented using
ZeroMQ. APIs are general to replace the implementation using MPI. Tested on
single node using examples in example/ folder. Todo test with other cluster
configurations (e.g., multiple nodes).
new 50ed157 add automake toolkits files
new 05e3d51 Add configure script and install guidance
new f883d6f Add install sh
new 561a3a8 Add install sh
new 9e18833 Add aclocal m4
new 8a2fe00 add config sub
new 87a550f add config guess
new 9baa90a fix bugs in configure ac
new 894845f edit README md
new 1aeb64a add config h in
new 1895139 add auxiliary files
new ab9fd55 reorganise automake files
new 6778b5c add Makefile in
new e900cd4 delete *.pb.cc & *.pb.h in make clean
new 86977fb add output information
new da3b5dd replace lmdb with DataShard as data provider for cifar10
example
new 7993a78 merge new configure
new 831efef add PrefetchLayer; Prefetching is done by moving DataLayer
and ParserLayer as sublayers/members of PrefetchLayer, whose ComputeFeature
function launches a thread to call the ComputeFeature functions of all
DataLayer and ParserLayer.
new 654d733 remove compile warnings
new 48b8fea update example of training MLP on MNIST dataset (replacing
lmdb with DataShard)
new 9ca7417 add FAQ
new 39969f2 Working on data partition within one group running on a
single node. TODO 1. update the performance collection by reporting performance
to the stub. 2. let workers pass requests to the stub without copying data
(passing addr or param id). messages to servers are then generated by the stub
which can aggregate gradients of shared parameters from all workers and collect
the updated parameters for them.
new 06f85e2 fix bugs from zmq sockets: move socekts creation to sub
threads.
new b5b943c 1. move functions in pm_server (pm_worker) into server
(trainer) to simplify the logics. now workers send simple messages to the stub
thread which construct the real update/get/put requests. the stub thread also
handles the responses from servers. E.g., the get/update response is handled by
the stub now. the workers then wait until its param's version is updated in the
collect function. avoid deadlocks for param_dealer_ and layer_dealer_ 2. tested
data partition in singl [...]
new 0d47ec5 update automake
new f29d93f merge code for data partition within one group on single node
to the upstream
new 9061245 fix bugs from gflags by moving definition of flag_sleep into
worker.cc
new 679573a fix compile bugs about gflags
new 6b1e65e add ClusterRunTime component definition
new e2b8abe Add cluster runtime component implementation based on
zookeeper
new 011823d add checking on test/validation frequency for TestNow and
ValidateNow functions.
new bd1d91c release callback vector in ZKClusterRT destructor
new c318a98 detect worker status through ClusterRuntime which calls
zookeeper. stop servers when all workers finish; stop stub when all workers and
servers finish
new a617e6c move performence display from worker to stub: worker send
performance as msgs to stub; fix bugs from multi threads call losslayers() of
neuralnet by init members of neuralnet in main thread. Then sub threads only
read these members and no conflicts from concurrent write.
new cd9fc79 try to fix bugs from zsock_connect by binding the router
before connecting to it; todo test the changed code.
new 96121ba add retry for worker connecting to zk, when its subscribed
server group is not up yet
new af23545 fix bug from zsock_connect by binding router firstly and then
connecting dealers to it.
new 9827f4f add scripts for running singa job; it start/stop zookeeper
service automatically;
new ee18746 download dependencies individually & add zookeeper
new d7849a9 delete download.sh
new f1ecd78 Merge branch 'master' of
https://git-wip-us.apache.org/repos/asf/incubator-singa
new 76b4119 add start/stop singa indicators in singa-run.sh
new 3921c29 Merge branch 'master' of
https://git-wip-us.apache.org/repos/asf/incubator-singa into feature-zookeeper
new 09c1b4e merge pull request 3; bug in compile using Makefile generated
by configure
new 5e111c3 add zookeeper
new 9dfb9b9 add zookeeper check in configure script
new 921f927 update singa-run.sh to fix wired bug which happens for
display_freq=30 and exec singa as ./singa -model. need to investigate it further
new 6bcaaaa fix bugs for early server termination
new 7be1adc Merge branch 'master' of github.com:nudles/incubator-singa
new b9680a2 Merge branch 'master' of
https://git-wip-us.apache.org/repos/asf/incubator-singa
new 806826e implement shared memory hogwild. update param at worker side
new 56fe4b8 ensure the callback function used in ZKClusterRT.sWatchGroup
cannot be NULL
new 40a86ab SINGA-7 Add support for shared memory hogwild training in
single process (node). hogwild training is enabled by setting share_memory to
true (i.e., worker groups share memory space for parameter values) in
cluster.conf. Depending on the configurations of cluster.conf, there are two
places to do the update: 1. if sever_update is true (default) and there is at
least one server group, workers send gradients to servers which conduct the
update based on sending worker's runni [...]
new eaaf5b8 SINGA-6 Update the implementation of Singleton to make it
thread-safe (using C++11's static construction). The ASingleton (constructor
with Argument) was used for sharing the Mshadow::Random among Layers and
Params. To make it thread-safe 1. we change it to TSingleton (Thread specfic
singlton). 2. we add a construtor without arguments for Mshadow::Random which
uses a seed generated based on the system clock. 3. we replace the rand/srand
in Mshaodw::Random<cpu> with C++11 [...]
new f13e3a7 add zookeeper check in configure script
new 3ba7155 SINGA-6 Update the implementation of Singleton to make it
thread-safe (using C++11's static construction). The ASingleton (constructor
with Argument) was used for sharing the Mshadow::Random among Layers and
Params. To make it thread-safe 1. we change it to TSingleton (Thread specfic
singlton). 2. we add a construtor without arguments for Mshadow::Random which
uses a seed generated based on the system clock. 3. we replace the rand/srand
in Mshaodw::Random<cpu> with C++11 [...]
new e990442 SINGA-6
new 856fc1f SINGA-6 Implement thread-safe singleton Merge branch
'fixbug-singleton'
new d5f9b9c SINGA-7 Implement shared memory Hogwild algorithm
new bfc09c1 Merge branch 'pull4'
new bd87c7a SINGA-7 Implement shared memory Hogwild algorithm
new 4df2bb5 Merge branch 'master' of
https://git-wip-us.apache.org/repos/asf/incubator-singa
new a1e449e SINGA-13 Separate intermediate files of compilation from
source files Modify Makefile.am to support advanced compilation. Add
corresponding guidance in README.
new d98c9ef SINGA-13 Separate intermediate files of compilation from
source files (optional)
new 280d5d6 fix bugs
new 24960c9 Merge branch 'master' of
https://github.com/apache/incubator-singa
new c12dc91 SINGA-15 Fixg a bug from ConnectStub function which gets
stuck for connecting layer_dealer_
new 6019905 Merge branch 'fixbug-layer_dealer'
new ceaa962 SINGA-14 Update layer API to be general for training
different models
new 9d07f3c SINGA-14 Update layer API to be general for training
different models Merge branch 'update-layerapi'
new 017b042 Merge branch 'master' of
https://github.com/apache/incubator-singa
new da6e4dc SINGA-16 Runtime Process id Management
new 51a924d SINGA-16 Runtime Process id Management Merge branch
'feature-procsregister'
new 65afb1a add root permission while install in default path
new f1c8491 SINGA-18 update API for displaying performance metric Update
the API of TrainOneBatch and TestOneBatch to support different ways of
displaying performance metric
new 69cf6cb support installing czmq in default path while zeromq library
path is specified
new 1318722 Merge branch 'master' of
https://github.com/apache/incubator-singa
new 81f7a12 Merge branch 'SINGA-17'
new a32102e SINGA-18 update API for displaying performance metric Merge
branch 'update-metricapi'
new 0f3a8ff SINGA-21 Code Review
new 767bad2 SINGA-21 Code Review
new dca8a23 set OpenBLAS default install path as /usr/local/
new 2bbed5f SINGA-22 add Q&A3 to clarify the environment settings
new 07596d9 SINGA-22 redress Q&A3 and add -fPIC option when installing
gflags
new 1c124e3 SINGA-22 redress Q&A3 and add -fPIC option when installing
gflags
new e0a52a6 SINGA-19 Slice large Param objects for load-balance Tested
with single worker, two worker group and two worker groups TODO test with
multiple servers and server groups for distributed hogwild and allreduce.
new c635cc6 SINGA-19 Slice large Param objects for load-balance
new 56d32e8 SINGA-19 Slice large Param objects for load-balance
new b0483f2 SINGA-21 Code review
new b2d7332 SINGA-21 Code review
new 2586e14 SINGA-21 Code review
new b0a5832 SINGA-21 Code review
new b6f2950 SINGA-21 Code review
new aefc2d4 SINGA-21 Code review
new 28ac509 SINGA-21 Code review
new acfc409 SIGNA-21 Code review-2
new cfde471 SINGA-21 Code review - 2
new 7d39f88 SINGA-21 Code Review-2
new 8133ac5 Merge branch 'master' of
https://github.com/apache/incubator-singa
new ffa2d66 SINGA-21 fix aclocal version problem
new 8bf975c SINGA-21 Code review
new a019958 add uncommited file in last commit, i.e., Makefile.example
new f437011 SINGA-8 Implement distributed Hogwild The program is simply
tested using two processes. TODO 1. read process endpoints from the zookeeper
instead of hard-coding them. 2. split large parameters to avoid load-balance
issue among server groups. currently, server groups are assigned (almost) equal
number of param objects. but these objects may be quite different in terms of
memory space.
new 51d4c2a SINGA-8 Implement distributed Hogwild Have replaced hard-code
enpoints with RegistPocs() and GetProcHost() implemented with the help of
zookeeper. TODO slice large Param objects in a separate branch.
new ad13d03 fixbugs; todo wait master to merge feature-sliceparam
new 884b9d7 SINGA-8 Implement distributed Hogwild
new 4956d6a SINGA-8 Implement distributed Hogwild
new c51f926 SINGA-8 Implement distributed Hogwild
new 9dc0567 SINGA-23 fix libtool version mismatch error
new 14ce5d9 SINGA-24 Implement Downpour training framework
new 90c2694 SINGA-25 - Setup glog output path
new f234c49 SINGA-25 - Setup glog output path
new 132eaa3 SINGA-27 Generate python modules for proto objects
new 27b863c SINGA-27 Generate python modules for proto objects
new 3c5b569 SINGA-27 Generate python modules for proto objects
new 327ab9b SINGA-27 Generate python modules for proto objects
new b87c21d SINGA-27 Generate python modules for proto objects
new 9d2d859 SINGA-27 Generate python modules for proto objects
new 903e036 SINGA-26 Run distributed training in a single command
new 064f95e SINGA-26 Run distributed training in a single command
new 3af32e5 SINGA-26 Run distributed training in a single command
new 7d9c0fb SINGA-26 Run distributed training in a single command
new cce3aeb Merge branch 'master' of
https://github.com/apache/incubator-singa
new ea7cfea SINGA-28 Fix a bug from toplogy sort of Graph
new 9a6e09f SINGA-29 Update NeuralNet class to enable customizing layer
partition type
new acb96d4 SINGA-33 Automatically launch a number of processes in the
cluster
new 72e73cc fixbug from the model.conf of cifar10 example due deletion of
partition_type field in modelproto
new 3819e59 SINGA-34 Support external zookeeper service
new 209ba1f SINGA-34 Support external zookeeper service
new a14c682 SINGA-34 Support external zookeeper service
new 7954a87 SINGA-34 Support external zookeeper service
new 585e275 SINGA-32 Implement Synchronous training frameworks
new 96bedb2 SINGA-32 Implement synchronous training framework
new 729a5c4 SINGA-12 Supprt Checkpoint and Restore
new 4502f41 SINGA-37 Enable users to set parameter sharing in model
configuration
new 189261f SINGA-12 Supprt Checkpoint and Restore
new 5bf1c92 SINGA-37 Enable users to set parameter sharing in model
configuration
new f746b99 SINGA-38 Support concurrent jobs
new 1fa5d5f SINGA-38 Support concurrent jobs
new 29de863 SINGA-38 Support concurrent jobs
new 0478e8c SINGA-36 Clean ModelProto, ClusterProto, JobProto and driver
program
new c3a248a SINGA-36 Clean ModelProto, ClusterProto, JobProto and driver
program
new e8e592f SINGA-39 Avoid ssh in scripts for single node environment
new 4afa468 SINGA-9 Add Support for Restricted Boltzman Machine (RBM)
model
new 1b03a9b SINGA-9 Add Support for Restricted Boltzman Machine (RBM)
model
new 4c28974 Merge branch 'rbm'
new 0616395 SINGA-12 Supprt Checkpoint and Restore
new a92a1c7 SINGA-42 Issue when loading checkpoints
new d5b6a30 SINGA-49 Fix a bug in HandlePutMsg func that sets param
fields to invalid values
new 4446889 SINGA-44 A bug when reseting metric values
new 6b34ff4 SINGA-46 Fix a bug in updater.cc to scale the gradients
new a8c8211 SINGA-44 A bug when reseting metric values
new 538736c SINGA-46 Fix a bug in updater.cc to scale the gradients
new 7a61a68 SINGA-47 Fix a bug in data layers that leads to out-of-memory
when group size is too large
new 2da5e23 SINGA-45 Set openblas num threads in job configuration
new 2c7edd7 SINGA-43 Remove Job-related output from workspace
new da844af SINGA-45 Set openblas num threads in job configuration
new db44012 SINGA-48 Fix a bug in trainer.cc that assigns the same
NeuralNet instance to workers from diff groups
new 975f0cf SINGA-53 Add lmdb compiling options
new 9d43056 SINGA-53 Add lmdb compiling options
new 039de8b SINGA-48 Fix a bug in trainer.cc that assigns the same
NeuralNet instance to workers from diff groups
new f0071a5 SINGA-52 Remove Python dependency in bash scripts
new 539fcee SINGA-56 No automatic launching of zookeeper service
new 1b574f3 SINGA-54 Refactor job configuration to move fields in
ModelProto out
new 4dee7b9 SINGA-54 Refactor job configuration to move fields in
ModelProto out
new d269b67 SINGA-54 Refactor job configuration to move fields in
ModelProto out
new b33e50d SINGA-50 Improve the code of ParseUpdateMsgs function
new a584da6 SINGA-54 Refactor job configuration to move fields in
ModelProto out
new fcd377a SINGA-58 Fix fan-in dimension of weight matrix
new 3ec12b9 SINGA-55 Refactor main.cc and singa.h
new 75b2e04 SINGA-54 Refactor job configuration to move fields in
ModelProto out
new f3cc20a SINGA-53 Add lmdb compiling options update README and
simplify singa_LDFLAGS
new 2498ff1 SINGA-55 Refactor main.cc and singa.h
new f2b0aef SINGA-59 Remove dependency on gflags
new e28b039 SINGA-21 Code review 3
new b24f0a3 SINGA-21 Code review 3
new 04b24e7 SINGA-21 Code review 3
new d0438b4 SINGA-21 Code review 3
new 14be101 SINGA-62 Remove building scrips and auxiliary files
new 97141e2 SINGA-61 Support user defined classes
new 6afa895 SINGA-60 Make learning rate and param init modular
new ef4de79 SINGA-9 Add Support for Restricted Boltzman Machine (RBM)
model This is to implement RBM in SINGA. To training RBM models, the
Contrastive Divergence (CD) algorithm is implemented. We have implemented a
BPWorker to run the Back-Propagation algorithm. To implement the CD algorithm,
we follow the same way to create a CDWorker whose RunOneBatch function controls
the logic of the CD algorithm, including positive phase, negative phase and
computing gradient phase. RBM's layer [...]
new fbbcaaf SINGA-9 Add Support for Restricted Boltzman Machine (RBM)
model
new f99246e SINGA-21 Code review 4
new 134c891 SINGA-21 Code review 4
new 53de92b SINGA-21 Code review 4
new 63adbda SINGA-21 Code review 4
new ae20303 SINGA-21 Code review 4
new 50deedd SINGA-66 Fix bugs in Worker::RunOneBatch function and
ClusterProto
new bc400a7 SINGA-67 Add singatest into build targets Modified
Makefile.am to add a target named "singatest"i. Now "make" will generate 5
files/libs: libsinga.la libgtest.la singa singatool singatest. User can also
specify the target he wants to generate, for example, "make singa".
new e0de439 SINGA-69 Add debug option in compilation configuration Two
updates: 1.User can use "--enable-singatest" option while running configure
script to generate test binary file and corresponding library. 2.User can use
"--enable-debug" option while running configure script to compile with "-g"
parameter. The optimization level is adapted to "-O2".
new 5d7d53f SINGA-69 Add debug option in compilation configuration Build
in a clean mode. Change "--enable-singatest" to "--enable-test".
new 6d59eec SINGA-51 Improve the convolution and pooling operations
new 077d380 SINGA-64 Add the test module for utils/common
new e769142 SINGA-51 Improve the convolution and pooling operations
new d5d817e SINGA-51 Improve the convolution and pooling operations
new ed9e373 SINGA-57 Improve Distributed Hogwild
new 7486647 SINGA-68 Preparation for Release -- License etc.
new be035df SINGA-68 Preparation for Release -- License etc.
new e1e8838 SINGA-68 Preparation for Release -- License etc.
new 9cded5e SINGA-68 Preparation for Release -- License etc.
new 1791442 SINGA-68 Preparation for Release -- License etc.
new 13b1c08 SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new ad86f72 SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new 3dc1eee SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new e8e07f1 SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new c1c6a2e SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new ba3b1a5 SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new 79a241c SINGA-10 Add Support for Recurrent Neural Networks (RNN)
new 3161175 SINGA-21 Code review 5
new d3e1fca SINGA-21 Code review 5
new f50d293 SINGA-21 Code review 5
new 366e6a8 SINGA-21 Code review 5
new 0c6e5c6 SINGA-21 Code review 5
new 4e15c34 SINGA-73 Refine the selection of available hosts
new 8e7c6cc SINGA-72 Minor updates to be consisten with documentation
new 3d1b0dc SINGA-72 Minor updates to be consisten with documentation
new 6bb1a8a SINGA-72 Minor updates to be consisten with documentation
new 243f210 SINGA-72 Minor updates to be consisten with documentation
new ab984da SINGA-72 Minor updates to be consisten with documentation
new 321ef96 SINGA-70 Refactor API of Layer, Worker, Server and Driver
new 88e2ba3 SINGA-72 Minor updates to be consisten with documentation
new 2160ab2 SINGA-72 Minor updates to be consisten with documentation
new 45e5026 SINGA-72 Minor updates to be consisten with documentation
new 2f66537 SINGA-72 Minor updates to be consisten with documentation
new 207e05f SINGA-77 Integrate with Apache RAT
new aad22fc SINGA-77 Integrate with Apache RAT
new 5900111 SINGA-79 Fix bug in singatool that can not parse -conf flag
new 71d10fb SINGA-79 Fix bug in singatool that can not parse -conf flag
new 958dfeb Preparing for v0.1.0-incubating.
new d99b24c SINGA-82 Refactor input layers using data store abstraction
new 5f010ca SINGA-82 Refactor input layers using data store abstraction
new 4afc7b8 SINGA-82 Refactor input layers using data store abstraction
new dc7f199 SINGA-82 Refactor input layers using data store abstraction
new b0b41f8 SINGA-68 Preparation for Release
new 239ed21 SINGA-84 Header Files Rearrange
new e469503 SINGA-84 Header Files Rearrange
new d452a2f SINGA-84 Header Files Rearrange
new f90e4dc SINGA-84 Header Files Rearrange
new e928ebb SINGA-85 Add functions for extracting features and test new
data
new 584a849 SINGA-85 Add functions for extracting features and test new
data
new 288a2ab SINGA-85 Add functions for extracting features and test new
data
new 595606c SINGA-85 Fix compilation errors in examples.
new b1da6e6 SINGA-90 Miscellaneous trivial bug fixes
new 021f41e SINGA-90 Miscellaneous trivial bug fixes
new f2899d5 SINGA-90 Miscellaneous trivial bug fixes
new 497dafc SINGA-90 Miscellaneous trivial bug fixes
new ef9a111 SINGA-91 - Add SoftmaxLayer and ArgSortLayer
new 1f513ec SINGA-89 Add Docker support
new f0e1629 SINGA-11 Start SINGA on Apache Mesos
new 0a228c8 SINGA-91 - Add SoftmaxLayer and ArgSortLayer
new f66bf46 [SINGA-93] Remove the asterisk in the log
tcp://169.254.12.152:*:49152
new 4abee35 SINGA-94 Move call to google::InitGoogleLogging() from
Driver::Init() to main().
new 96c440f SINGA-11 Start SINGA on Apache Mesos
new 4567ed8 SINGA-89 Add Docker support
new 809798a SINGA-90 Trivial README.md documentation fixes
new 5277935 SINGA-89 Add Docker support
new e5e679f SINGA-94 Move call to google::InitGoogleLogging() from
Driver::Init() to main()
new 316c65b SINGA-94 Move call to google::InitGoogleLogging() from
Driver::Init() to main()
new 2f5db2a SINGA-11 Start SINGA on Apache Mesos
new d6c3963 SINGA-89 Add Docker support
new 5d076e5 SINGA-11 Start SINGA on Apache Mesos
new 7c50e03 SINGA-89 Add Docker support
new 152e473 SINGA-95 Add make test after building
new 84c7f90 SINGA-95 Add make test after building
new 45d6880 (SINGA-96) Add Momentum to Cifar10 Example
new 3f80cca SINGA-78 Use Doxygen to generate documentation.
new 059ae9a SINGA-87 Replace exclude field to inlcude field for layer
configuration (still support exclude field).
new e1a6f14 SINGA-87 Replace exclude field to inlcude field for layer
configuration (still support exclude field).
new 903b8d8 SINGA-87 Replace exclude field to inlcude field for layer
configuration (still support exclude field).
new e137784 SINGA-68 Preparation for Release -- License etc. Add README.md
new 06cdf36 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new b99de6c SINGA-80 New Blob Level and Address Level Math Operation
Interface
new bbb7dbc SINGA-80 New Blob Level and Address Level Math Operation
Interface
new c13e037 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 4728f7c SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 01d91af SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 641eb31 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 4b84dbe SINGA-80 New Blob Level and Address Level Math Operation
Interface
new d3379cb SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 98f5256 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new d333cba SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 38dc9a1 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 21cfc21 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 32e0921 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 86284f1 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new a65a953 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 364c885 SINGA-80 New Blob Level and Address Level Math Operation
Interface Merge branch 'SINGA-80'
new 771ff32 SINGA-104 Add Context Class
new 35de4f9 SINGA-104 Add Context Class
new 9aff30a SINGA-104 Add Context Class
new 3841bc5 SINGA-104 Add Context Class
new e3bda08 SINGA-104 Add Context Class
new b2cfa17 SINGA-104 Add Context Class
new 71819ab [SINGA-101] Add ll (ls -l) command in .bashrc file when using
docker
new 4664b6b SINGA-106 Add dummy layer for test purpose
new bd2e345 SINGA-109 Refine bridge layers
new e8d01dc SINGA-110 Add Layer member datavec_ and gradvec_
new f16b1be SINGA-107 Error from loading pre-trained params for training
stacked RBMs
new 0dd2d5c SINGA-105 Update GUN make files for compiling cuda related
code
new 9b68af1 SINGA-108 Add Python wrapper to singa
new 0578d82 SINGA-108 Add Python wrapper to singa
new b8104ef SINGA-108 Add Python wrapper to singa
new ef5b655 SINGA-108 Add Python wrapper to singa
new 3e58760 SINGA-108 Add Python wrapper to singa
new 94ff71c SINGA-108 Add Python wrapper to singa
new 28e48a6 SINGA-111 Add slice, concate and split layers
new 3ea1eb6 SINGA-111 Add slice, concate and split layers
new 7e414e5 SINGA-111 Add slice, concate and split layers
new 7cdb22f SINGA-111 Add slice, concate and split layers
new be0f179 SINGA-108 Add Python wrapper to singa
new 9aaba61 SINGA-108 Add Python wrapper to singa
new 1981874 Merge branch 'feature-python'
new af1bf50 SINGA-100 Implement layers using CUDNN for GPU training
new 49293a6 SINGA-100 Implement layers using CUDNN for GPU training
new 5d35ef2 SINGA-100 Implement layers using CUDNN for GPU training
new 2ed18a5 SINGA-100 Implement layers using CUDNN for GPU training
new 6e56334 SINGA-100 Implement layers using CUDNN for GPU training
new 8cacd83 SINGA-100 Implement layers using CUDNN for GPU training
new f31ba64 SINGA-100 Implement layers using CUDNN for GPU training
new eb97097 SINGA-100 Implement layers using CUDNN for GPU training
new 8174760 SINGA-100 Implement layers using CUDNN for GPU training
new 5728c24 SINGA-100 Implement layers using CUDNN for GPU training
new f8be9af SINGA-100 Implement layers using CUDNN for GPU training
new 15b23a6 SINGA-100 Implement layers using CUDNN for GPU training
new 05680dd SINGA-100 Implement layers using CUDNN for GPU training
new 6f81adb SINGA-100 Implement layers using CUDNN for GPU training
new f9d4268 SINGA-100 Implement layers using CUDNN for GPU training
new 372245f SINGA-100 Implement layers using CUDNN for GPU training
new f3b47c7 SINGA-100 Implement layers using CUDNN for GPU training
new 8ac511c SINGA-100 Implement layers using CUDNN for GPU training
new b003353 SINGA-100 Implement layers using CUDNN for GPU training
new 1f977b1 SINGA-115 - Print layer debug information in the neural net
graph file
new c8cb913 SINGA-115 - Print layer debug information in the neural net
graph file
new 138599f SINGA-115 - Print layer debug information in the neural net
graph file
new 2d38cb3 SINGA-113 Model/Hybrid Partition Support
new 8af565c SINGA-113 Model/Hybrid Partition Support
new 82563f6 SINGA-113 Model/Hybrid Partition Support
new bf4cd3b SINGA-113 Model/Hybrid Partition Support
new 4a0db51 SINGA-113 Model/Hybrid Partition Support
new cf6ef82 SINGA-116 Fix a bug in InnerProductLayer caused by weight
matrix sharing
new 0cb6cc2 SINGA-116 Fix a bug in InnerProductLayer caused by weight
matrix sharing
new aea0f1a SINGA-114 Remove short logs in tmp directory
new b8ac2b8 SINGA-118 Make protobuf LayerType field id easy to assign
new 1bc5007 SINGA-118 Make protobuf LayerType field id easy to assign
new 74a28dc SINGA-121 Not linking math_kernel.o when generate libsinga.so
new 7d43e27 SINGA-81 Add Python Helper, which enables users to construct
a model (JobProto) and run Singa in Python
new ceb0665 SINGA-81 Add Python Helper, which enables users to construct
a model (JobProto) and run Singa in Python
new 94435eb SINGA-81 Add Python Helper, which enables users to construct
a model (JobProto) and run Singa in Python
new 8b69cad SINGA-81 Add Python Helper, which enables users to construct
a model (JobProto) and run Singa in Python
new 3c12730 SINGA-81 Add Python Helper, which enables users to construct
a model (JobProto) and run Singa in Python
new 8914750 SINGA-81 Add Python Helper
new d47ed4b SINGA-81 Add Python Helper
new 2a6645c SINGA-81 Add Python Helper
new 9ff176c SINGA-81 Add Python Helper
new 8a07a29 SINGA-97 Add HDFS Store
new 374f11d SINGA-97 Add HDFS Store
new 9fbc8ee SINGA-97 Add HDFS Store
new 4cfe813 SINGA-97 Add HDFS Store
new aada365 SINGA-97 Add HDFS Store
new befe5ca SINGA-97 Add HDFS Store
new 714fd2c SIGNA-97 Add HDFS Store
new fe86b02 SINGA-98 Add Support for AlexNet ImageNet Classification Model
new 6e815db SINGA-98 Add Support for AlexNet ImageNet Classification Model
new bb75a0b SINGA-98 Add Support for AlexNet ImageNet Classification Model
new ddf4e79 SINGA-120 - Implemented GRU and BPTT \n 1) Added the
implementation of the GRU model; \n 2) Added a test for GRU functions
new 473c985 SINGA-120 - Implemented GRU and BPTT: 1) Updated Driver.cc to
register GRU; 2) Updated job.proto to include configuration of GRU; 3) Updated
configure.ac to fix some compliation errors
new 777dfb6 SINGA-120 - Implemented GRU and BPTT: 1) Implemented the
unrolling function for BPTT; 2) Added tests for unrolling
new c440e4a SINGA-120 - Implemented GRU and BPTT: Fixed bugs regarding
the Map function
new d43af9f SINGA-120 - Implemented GRU and BPTT: Fixed bugs of returning
null
new b4b6a38 SINGA-120 - Implemented GRU and BPTT: Made changes for debug
tokenizer
new 1f03f9d SINGA-120 - Implemented GRU and BPTT: Updated dummy layer for
fixing bugs
new 959ef70 SINGA-120 - Implemented GRU and BPTT
new 6a4c996 SINGA-120 - Implemented GRU and BPTT
new a2f4e46 SINGA-120 - Implemented GRU and BPTT
new 96202a8 Added 1-2 sentences about optional dependencies.
new 247fabd Add Apache License to several files.
new d14ee19 Add description on hdfs and gpu support compilation in
README.md.
new 27a48c2 Add further explanations.
new ae1b33a add release note
new 60a17bb Prepare RAT check
new 8ad73e6 Passed RAT check
new b6591d1 singa-0.2.0-rc1 update release notes to include features
about gru, bptt and alexnet
new 4ebf0bc Add license for python files to pass 'make rat'
new c45ff1c SINGA-120 - Implemented GRU and BPTT
new c72ef0f SINGA-120 - Implemented GRU and BPTT
new d5e991c SINGA-120 - Implemented GRU and BPTT
new 9705f39 SINGA-120 - Implemented GRU and BPTT
new 24a1be9 SINGA-120 - Implemented GRU and BPTT
new 198a049 Update LICENSE file and headers of source files.
new f495ff3 Update Readme file to specify tested OS systems
new e385d2a singa incubating 0.2.0 RC2; update license header for
common.cc
new 4662dc3 SINGA-125 Improve Python Helper
new 1cfdac6 SINGA-125 Improve Python Helper
new a1553f8 SINGA-125 Improve Python Helper
new f32c78b SINGA-125 Improve Python Helper
new 5526c51 SINGA-139 Fixed a bug in built-in convolution layer
new c1444b7 SINGA-112 Clean GNU configure and make code
new 8d4953a SINGA-143 The compilation cannot detect libsingagpu.so file
new 122d362 SIGNA-141 Undesired Hash collision when locating process id
to workers and servers
new e32e70c SINGA-145 New SGD based optimization Updaters: AdaDelta,
Adam, AdamMax
new 6814007 SINGA-145 New SGD based optimization Updaters: AdaDelta,
Adam, AdamMax
new ec0878b SINGA-145 New SGD based optimization Updaters: AdaDelta,
Adam, Adammax
new 348c5e5 SINGA-149 Docker build fail
new 259422c SINGA-141 Merge pull request fixing hash collision
new a6eea9c SINGA-150 Mesos Docker container failed
new 9679417 SINGA-148 Race condition between Worker threads and Driver
new 77d5c5c SINGA-148 Race condition between Worker threads and Driver
new b00dc32 SINGA-148 Race condition between Worker threads and Driver
new adeced6 SINGA-155 Remove zookeeper for single-process training
new 0233049 SINGA-155 Remove zookeeper for single-process training
new d8dffdf SINGA-156 Remove the dependency on ZMQ for single process
training
new 247002d SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 8ade7d7 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 65b8c8d SINGA-156 Remove the dependency on ZMQ for single process
training
new d452c1f SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 42f5253 SINGA-156 Remove the dependency on ZMQ for single process
training
new 914c1e7 SINGA-156 Remove the dependency on ZMQ for single process
training
new a3c82ca SINGA-134 Extend SINGA to run over a GPU cluster
new 8b7d1e0 SINGA-134 Extend SINGA to run over a GPU cluster
new 8130b7e SINGA-126 Python Binding for Interactive Training
new 5a8fc37 SINGA-126 Python Binding for Interactive Training
new afc50a9 SINGA-136 Support cuDNN v4
new 1c8e0dc SINGA-126 Python Binding for Interactive Training
new 4b4ad05 SINGA-136 Support cuDNN v4
new c97b970 SINGA-126 Python Binding for Interactive Training
new a91e82f SINGA-131 Implement and optimize hybrid training using both
CPU and GPU
new 040cbb2 SINGA-131 Implement and optimize hybrid training using both
CPU and GPU
new 9c2c297 SINGA-136 Support cuDNN v4
new efdf6d7 SINGA-136 Support cuDNN v4
new e5f111c SINGA-157 Change the priority of cudnn library and install
libsingagpu.so
new 061f8e3 SINGA-157 Change the priority of cudnn library and install
libsingagpu.so
new 5f67e57 Add license to Dockerfile_gpu
new a0bdd0b SINGA-130 Data prefetching layer
new 4344ab7 SINGA-130 Data prefetching
new 3b2e895 SINGA-136 Support cuDNN v4
new 369d879 SINGA-136 Support cuDNN v4
new 8329aa0 SINGA-80 New Blob Level and Address Level Math Operation
Interface
new 991c6ab SINGA-130 Data Prefetching
new cf4be5a SINGA-130 Data prefetching
new 04cfe2d Update license, release notes and readme for
incubating-v0.3-RC1
new fb5b218 Prepare v0.3.0-RC1
new fde87ed SINGA-159 Rewrite safe_queue.h to avoid the license issue
new d547a86 Prepare for v0.3.0-RC2
new dd1e4af SINGA-163 - Reorganize the project folder layout
new 193e9df SINGA-165 Add cross-platform timer API to singa Add timer
class with cmake compilation. How to test: (in singa main folder) mkdir build
&& cd build cmake .. make ./bin/timer
new bdbffdc SINGA-165 Add cross-platform timer API to singa
new dc5aa6e SINGA-164 - Add the base Tensor class
new 94633f4 SINGA-166 light built-in logging replacing glog
new 9bccfb6 SINGA-166 light built-in logging replacing glog
new 4a2ffb7 SINGA-167 - Add Tensor Math function APIs
new e36bc92 SINGA-169 Add base Layer class for V1.0
new 02851fa SINGA-167 - Add Tensor Math function APIs
new 99e0d24 SINGA-170 Add Dropout layer and CudnnDropout layer
new c3a0558 SINGA-170 Add Dropout layer and CudnnDropout layer
new 3a87201 SINGA-172 Add CMake supporting for Cuda and Cudnn libs
new b491875 SINGA-170 Add Dropout layer and CudnnDropout layer
new 282712c SINGA-171 - Create CppDevice and CudaDevice
new 0b4b2e2 SINGA-171 - Create CppDevice and CudaDevice
new e3da6a5 SINGA-172 - Add CMake supporting for Cuda and Cudnn libs
new 9d1bcb4 SINGA-171 - Create CppDevice and CudaDevice
new 611554f SINGA-177 Add fully cmake supporting for the compilation of
singa_v1
new 72923b1 SINGA-177 Add fully cmake supporting for the compilation of
singa_v1
new d680079 SINGA-176 - Add loss and metric base classes
new 668ae16 SINGA-181 Add NVCC supporting for .cu files
new a1c3437 SINGA-176 - Add loss and metric base classes
new 3171459 SINGA-176 - Add loss and metric base classes
new fbd5219 SINGA-182 Clean math function APIs and implementations
new 870d1a9 SINGA-182 Clean math function APIs and implementations
new c6ae786 SINGA-185 Add CBLAS and GLOG detection for singav1
new 152056d SINGA-178 Add Convolution layer and Pooling layer
new 7d149ec SINGA-178 Add Convolution layer and Pooling layer
new 2dac380 SINGA0-183 Add the base classes for optimizer, constraint and
regularizer
new 3e2507b SINGA-180 Add Activation layer and Softmax layer
new efd7b62 SINGA-184 Add Cross Entropy loss computation
new 73d4a34 SINGA-188 Add Dense layer
new 64ea206 SINGA-188 Add Dense layer
new eadd3f9 SINGA-174 Add Batch Normalization layer and Local Response
Normalization layer.
new fa2ea30 SINGA-174 Add Batch Normalization layer and Local Response
Normalization layer.
new da23e57 SINGA-189 Generate python outputs of proto files
new 7a19165 Merge SINGA-189 with the latest code.
new 04e23d1 SINGA-162 Transfer the codebase for SINGA v1.0 into dev branch
new 01aaf49 SINGA-191 Add "autotune" for CudnnConvolution Layer
new 683b3a7 SINGA-175 Add memory management APIs and implement a subclass
using CNMeM
new 178db01 SINGA-192 Implement optimization algorithms for v1
new 5afd81b SINGA-190 - Add prelu layer and flatten layer
new 07c49da SINGA-168 Implement Cpp Math functions APIs
new 564c88a SINGA-182 Clean math function APIs and implementations
new 6d69047 SINGA-182 Clean math function APIs and implementations
new 5784bff SINGA-192 Implement optimization algorithms for v1
new 58be3f8 SINGA-190 - Add prelu layer and flatten layer
new 26df5ac SINGA-190 - Add prelu layer and flatten layer
new ec17aca SINGA-184 Add Cross Entropy loss computation
new a2a8e34 SINGA-195 Channel for sending training statistics
new 21e4b2d SINGA-184 Add Cross Entropy loss computation
new a4fc4ea SINGA-195 Channel for sending training statistics
new b167dfa SINGA-195 Channel for sending training statistics
new 272100a SINGA-192 Implement optimization algorithms for v1
new 0cd9663 SINGA-192 Implement optimization algorithms for v1
new 4d596dd SINGA-192 Implement optimization algorithms for v1
new f488070 SINGA-196 Rename class Blob to Block
new a4d9aab SINGA-190 - Add prelu layer and flatten layer
new ce3e6dc SINGA-197 Add CNMem as a submodule in lib/
new 74f0214 SINGA-198 - Change Layer::Setup API to include input Tensor
shapes
new b91002b SINGA-180 Add Activation layer and Softmax layer
new 8900543 SINGA-187 Add popular parameter initialization methods
new 2a582df SINGA-186 Create Python Tensor class
new 24833fa SINGA-186 Create Python Tensor class
new 254343c SINGA-186 Create Python Tensor class
new 396f9bf SINGA-186 Create Python Tensor class
new 077d13e SINGA-175 Add memory management APIs and implement a subclass
using CNMeM
new 5651383 changed all device pointer to shared pointer
new 1ca8c63 SINGA-201 Error when running Mesos
new bef1db0 SINGA-200 - Implement Encoder and Decoder for data
pre-processing
new 67ee3c5 SINGA-203 Add OpenCV detection for cmake compilation
new 13d60b0 SINGA-200 - Implement Encoder and Decoder for data
pre-processing
new 833f461 SINGA-200 - Implement Encoder and Decoder for data
pre-processing
new 913f45c SINGA-202 Add reader and writer for binary file
new 14d31a4 SINGA-200 - Implement Encoder and Decoder for data
pre-processing
new 9abd791 SINGA-205 - Enable slice and concatenate operations for
Tensor objects
new dd08f41 Merge PR #165 for CnMeM
new d826b2e SINGA-204 Support the training of feed-forward neural nets
new cf1d841 SINGA-204 Support the training of feed-forward neural nets
new 71eb059 SINGA-204 Support the training of feed-forward neural nets
new fce5d81 SINGA-204 Support the training of feed-forward neural nets
new 97648a6 SINGA-204 Support the training of feed-forward neural nets
new 4db968c SINGA-204 Support the training of feed-forward neural nets
new d3c1bae SINGA-211 Add TextFileReader and TextFileWriter for CSV files
new 62c6603 SINGA-210 Enable checkpoint and resume for v1.0
new f026b60 SINGA-212 Disable the compilation of libcnmem if USE_CUDA is
OFF
new 7cfdb99 SINGA-186 Create Python Tensor class
new 50f6722 SINGA-186 Create Python Tensor class
new d83f5d5 SINGA-186 Create Python Tensor class
new bcda737 SINGA-214 Add LMDBReader and LMDBWriter for LMDB
new cde7dcf Merge PR #185 for more py tensor functions. remove Tensor
member func to_numpy, and keep global to_numpy
new f0bc228 Merge PR #192 for lmdb reader and writer.
new dc013f3 SINGA-194 Add a Platform singleton
new 164fc2e SINGA-186 Create Python Tensor class
new e857bc3 SINGA-214 add jenkins for CI support ------ to support CUDNN
CI
new c502552 Merge PR #195
new 37e6ad2 SINGA-217 build python package with setup.py
new 7333517 SINGA-199 Implement Python classes for SGD optimizers
new 790b7b4 SINGA-193 Add Python layers
new dbc720d SINGA-221 Travis-CI configuration
new 245f60e SINGA-222 Fixed bugs in IO
new 7444f0a SINGA-213 Implement Encoder and Decoder for CSV
new 4931922 SINGA-213 Implement Encoder and Decoder for CSV
new 7561653 SINGA-223 Use Sphinx to create the website.
new 7e050c1 SINGA-215 Implement Image Transformation for Image
Pre-processing
new 722a59e SINGA-225 Documentation for installation and Cifar10 example
new 95091dc SINGA-225 Documentation for installation and Cifar10 example
new a18a019 SINGA-225 Documentation for installation and Cifar10 example
new 8c2a0e8 SINGA-225 Documentation for installation and Cifar10 example
new 8e3d3df SINGA-223 Use Sphinx to create the website.
new d45715d SINGA-226 Add parallel training on a single machine for singa
v1.0
new 0184fac SINGA-226 Add parallel training on a single machine for singa
v1.0
new 4e7f3c1 SINGA-226 Add parallel training on a single machine for singa
v1.0
new 35c8930 SINGA-173 OpenCL device support and implementation
new 57f3a1e SINGA-173 OpenCL device support and implementation.
new f9219cc SINGA-173 OpenCL device support and implementation: Fixed a
missing header include in device.cc, and ifdef switches for OpenCL.
new 3f6b5e3 SINGA-173 OpenCL device support and implementation: Removed
errant constructor in Platform class.
new ed98149 Singa-228 Add Cpp Version of Convolution and Pooling layer
new 464dcda SINGA-172 OpenCL device support and implementation
new f07e354 Merge PR#224 for CPP version of convolution and pooling layers
new 96ed638 SINGA-174 Add Batch Normalization layer and Local Response
Nomalization layer.
new db5478e SINGA-174 Add Batch Normalization layer and Local Response
Nomalization layer.
new e39be3e SINGA-229 Complete install targets
new bc3b74b SINGA-231 Batchnormlized VGG model for cifar-10
new 34d3ae6 SINGA-232 Alexnet on Imagenet
new 8051720 SINGA-232 Alexnet on Imagenet
new 28678ae SINGA-231 Batchnormlized VGG model for cifar-10
new c51f944 SINGA-218 Implementation for RNN CUDNN version
new 8e0b108 SINGA-218 Implementation for RNN CUDNN version
new dfc422e SINGA-218 Implementation for RNN CUDNN version
new 889abf8 SINGA-233 A new communication framework for SINGA
new 90a1cd5 SINGA-233 Fix a minor bug in processing lost connections
new d308e06 SINGA-233 Fix bugs in sending large messages
new 1565e65 SINGA-233 Add the support for the detection of endpoint
timeout
new 4353ce9 SINGA-233 Some code cleaning
new 541ad68 SINGA-218 Implementation for RNN CUDNN version
new 45620d5 SINGA-233 New communication interface for SINGA
new 17bfb19 Merge PR #230 of new communcation interface for distributed
training.
new 53639b7 Merge PR #232 for training AlexNet over ImageNet
new 05720c2 SINGA-235 - Unify the engines for cudnn and singa layers
new a0af465 SINGA-229 Complete install targets
new 055ff17 SINGA-174 Add Batch Normalization layer and Local Response
Nomalization layer.
new bf81f25 SINGA-217 build python package with setup.py - Singa python
binding, package with pip or create wheel - In the cmake file, specify
USE_PYTHON ON, then cmake and make. After that run "<sudo> pip install ."
under "build/python" path User can run command 'singa' in a model workspace
to serve and train - dependent on numpy>=1.11.0; protobuf>=2.5.0,<3;
flask>=0.10.1 Users should also install openblas - Models are sharing on
github gists. - Todo, optimize [...]
new 6b2ff3c SINGA-217 build python package with setup.py
new 3299b0c SINGA-237 New documentation files for SINGA v1.0
new 94ffe55 Merge PR #235 which fixed a bug in batchnorm.cc
new a91bf2a SINGA-235 - Unify the engines for cudnn and singa layers
new 230230c SINGA-229 Complete install targets
new e1a524d SINGA-238 RBM on mnist
new 8cd5530 SINGA-237 New documentation files for SINGA v1.0
new dffae6b Merge PR #228 which updates cmake files to enable 'make
install'.
new 5b332a4 SINGA-238 RBM on MNIST
new 1db2784 Merge PR #240 for training RBM againt MNIST
new d3a57cf SINGA-237 New documentation files for SINGA v1.0
new bc822cd SINGA-237 New documentation files for SINGA v1.0
new 33992c9 SINGA-237 New documentation files for SINGA v1.0
new cdd718e SINGA-237 New documentation files for SINGA v1.0
new f80fedb SINGA-210 Enable checkpoint and resume for v1.0
new 410f238 SINGA-237 New documentation files for SINGA v1.0 - add
python installation instruction, remove flask dependency
new 30731ee SINGA-237 New documentation files for SINGA v1.0
new 2c049d6 Merge commits from wenfeng for the installation page.
new 9c71bd6 SINGA-237 New documentation files for SINGA v1.0
new 5db7eb6 Merge commits for updating layer identifers and commits for
updating the installation page.
new 6d4539e Fixed the bug leading to wired accuracy (nan), which was
caused by forgeting to average the gradient over the whole mini-batch. That is
why we need a lower learning rate and could not use momentum. Update the lr in
optimzier.py to time the multiplier Fix the bug from mis-setting the pooling
type of alexnet.py (max-->avg)
new 5d20d35 Merge commits for debugging the gradient average error and
commits for documentation.
new e963363 SINGA-237 New documentation files for SINGA v1.0
new 31ae6bd SINGA-223 Use Sphinx to create the website.
new c2173b3 SINGA-237 New documentation files for SINGA v1.0
new 72d736a Merge branch commits for doc layout change and commits for
other doc and code changes
new 0a76425 Minor updates to pass tests and run examples
new a144a61 SINGA-237 New documentation files for SINGA v1.0
new 22889bc fix minor bug
new d2300ae Merge updates for docs and bug-fix for pool.cc and
tensor.to_numpy()
new cdc5ffd Merge PR #245 for fixing the error in test_snapshort.cc due
to data type problems.
new 7ebea53 SINGA-227 Add Split and Merge Layer and add ResNet
Implementation
new a54c889 SINGA-227 Add Split and Merge Layer and add ResNet
Implementation
new b3566e4 update results of rat check.
new 1c5ca22 SINGA-240 Add license for singa source
new 9ee16e0 SINGA-240 Add license for singa source
new ed9587c Preparing for V1.0 RC0.
new f9c6d5c Merge branch 'v1-rc0'
new 68278c4 SINGA-241 Fix bugs in Imagenet example
new 9b9dc10 fix license issues
new 65bf582 Merge PR #248 for removing the fixed numbers of training and
test images in Imagenet example
new 40c9123 SINGA-242 Compile all source files into a single library.
new 546c38b SINGA-242 Compile all source files into a single library.
new 59f56e6 fix bug for python build in mac
new c5e6309 fix bug
new 2008446 Merge pull request #1 from aaronwwf/test
new 3d18111 SINGA-242 Compile all source files into a single library
new 61ec2ec Merge PR #249 for compiling all singa code into a single
libsinga.so
new 595302a SINGA-243 ViennaCL backend for OpenCL support
new 5380b3a Update installation.md
new 1996e8a Merge PR #251 for updating opencl support using ViennaCL
new 0416a0f fix minor erros from installation.md and
examples/cifar10/predict.py
new d76caea SINGA-244 Separating swig interface and python binding files
- move swig interface files to src/api - move python to root folder - move
python related cmake functions to python cmake files - use add_library OBJECT
command to build singa_objects, then be used by singa shared library and
python wraper library, avoiding build twice - todo, add java binding
new 32ba40d SINGA-244 Separating swig interface and python binding files
- add java binding cmake files - todo: add test code
new 76cd806 SINGA-245 Float as the first operand can not multiply with a
tensor object
new 0ebce1a SINGA-245 Float as the first operand can not multiply with a
tensor object
new 5439e3a SINGA-244 Separating swig interface and python binding files
Remove dependencies on copy_protobuf target.
new 92677fa Merge branch 'master' of
https://github.com/apache/incubator-singa into SINGA-244
new 7a85c6d SINGA-246 Imgtool for image augmentation - add Imgtool
class in singa.utils - provide resize,crop,flip,color_cast,enhance methods
new 809a592 SINGA-246 Imgtool for image augmentation - put Imgtool
class into a separate file - some improvements
new 26d9cd4 SINGA-246 Imgtool for image augmentation
new f4fae37 fix some bugs and reduce training memory
new 8732021 SINGA-246 Imgtool for image augmentation
new bc699f4 udpate docs for v1 release.
new 392a9c3 add release note for v1
new 897f5d5 Fixed a bug reported by hacker99 from layer.py (SINGA-248)
new e321281 SINGA-243 ViennaCL backend for OpenCL support
new f3665e5 SINGA-243: ViennaCL Backend: Removed errant FindOpenCL find.
new e3df3bd SINGA-230 OpenCL Convolution and Pooling
new 538bdac SINGA-252 Use the snapshot methods to dump and load models
for pysinga
new 7c12f40 SINGA-252 Use the snapshot methods to dump and load models
for pysinga
new deb187b SINGA-252 Use the snapshot methods to dump and load models
for pysinga
new 064f3fb SINGA-251 - Implement image loader for pysinga
new 5716105 SINGA-254 Implement Adam for V1
new 17ac160 SINGA-251 - Implement image loader for pysinga
new 3a64342 Merge pull request 262 into master
new 8cf18e5 update the docs of schedule for v1.1 and installation for a
FAQ entry
new 8055515 update docs for image_tool.py and data.py
new 350bdf4 SINGA-247 Add windows support for singa
new d0317a7 SINGA-247 Add windows support for singa
new 37d3b39 SINGA-247 Add windows support for singa
new af30903 SINGA-247 Add windows support for singa.
new 6c9e0b4 SINGA-247 Add windows support for singa.
new 8531af5 SINGA-247 Add windows support for singa
new 901e885 SINGA-247 Add windows support for singa
new e9b2c96 SINGA-247 Add windows support for singa
new c967169 SINGA-259 Add maven pom file for building java classes
new 77405ec SINGA-253 Net converter for caffe model
new 32b4368 SINGA-253 Net converter for caffe model
new f536975 SINGA-266 Add Rafiki python toolkits - add rafiki python
package - firt version of rafiki agent module
new 9eabb95 SINGA-266 Add Rafiki python toolkits - fix bug, remove
debug comments
new 61faa84 SINGA-267 Add spatial mode in batch normalization layer
new fac3af9 SINGA-267 Add spatial mode in batch normalization layer
new 932a0cd SINGA-255 Compile mandatory depedent libaries together with
SINGA code
new a98446f SINGA-255 Compile mandatory depedent libaries together with
SINGA code
new 47e38c8 SINGA-255 Compile mandatory depedent libaries together with
SINGA code
new f2c0b84 SINGA-267 Add spatial mode in batch normalization layer
new 4826d40 SINGA-255 Compile mandatory depedent libaries together with
SINGA code
new 1117a16 Merge PR #272 for compiling depedent libs together with SINGA
new beb5e72 Reorganize the installation doc page
new e5a3e52 SINGA-266 Add Rafiki python toolkits - fix bug, sleep a
while when agent stop
new 22d98bb SINGA-269 - Fix a bug from tensor multiplication and a typo
new f35d217 SINGA-264 Extend the FeedForwardNet to accept multiple inputs
new 16f3bf6 SINGA-271 Add Concat and Slice layers
new d84af80 SINGA-271 Add Concat and Slice layers
new 5afd81c SINGA-271 Add Concat and Slice layers
new 0bf1475 SINGA-261 Add version ID into the checkpoint files
new 7da4455 SINGA-268 Add IPython notebooks to the documentation
new 0a8dbcd SINGA-268 Add IPython notebooks to the documentation
new e43ea69 SINGA-261 Add version ID into the checkpoint files
new d917007 SINGA-273 Improve license and contributions
new 15a619b Merge PR #281
new d1110c0 SINGA-275 - Add Cross Entropy Loss for multiple labels
new 8481111 Check and fix cudnn engine for concat and slice layer
new 6ef4bbe SINGA-268 Add IPython notebooks to the documentation
new 48d43c1 SINGA-268 Add IPython notebooks to the documentation
new 4d7a8ee SINGA-268 Add IPython notebooks to the documentation
new 0971268 SINGA-268 Add IPython notebooks
new cb81caa SINGA-278 Convert trained caffe parameters to singa
new c7ceb07 SINGA-278 Convert trained caffe parameters to singa
new aae9d3e SINGA-278 Convert trained caffe parameters to singa
new bd41fd1 SINGA-278 Convert trained caffe parameters to singa
new 5953e7b SINGA-261 Add version ID into the checkpoint files
new ee93b7d SINGA-279 Unit Test test_layer failed when CUDNN version is
4.0
new 3b8cb94 SINGA-280 Jenkins CI support
new 33cf546 SINGA 280 Jenkins CI support
new 82c12c2 SINGA-271 Add Concat and Slice layers
new a43f291 SINGA-282 Add grayscale to load method in ImageTool class
new 7956019 SINGA-283 Fix a bug of net.py backward function for layers
with multiple inputs
new 4dfee52 SINGA-261 Add version ID into the checkpoint files
new eca642b SINGA-278 Convert trained caffe parameters to singa
new 458b0f6 SINGA-287 - Add memory size check for cudnn convolution
new 3f23c0d Merge PR #286
new e44658a SINGA-284 - Add python unittest into Jenkins and link static
libs into whl file
new 761c463 SINGA-284 - Add python unittest into Jenkins and link static
libs into whl file
new 330c879 SINGA-288 - Publish wheel of PySINGA generated by Jenkins to
public servers
new f94ec89 SINGA-289 Update SINGA website automatically using Jenkins
Add Dockerfile for generating html files from doc/. Add a shell script
(jenkins_doc.sh) to build the documentation and udpate svn repo. Update the
tool/jenkins/README.md with instructions to set up the Jenkins job for
documentation update. Move doc/notebook into doc/en/docs/notebook as the
notebooks depend on the APIs in doc/en/docs/. to add docker file for runtime
new d98ad69 SINGA-266 Add Rafiki python toolkits - change agent api to
support genernal post request
new d2a8d77 SINGA-266 Add Rafiki python toolkits
new 781f740 update docs of installation via wheel and debian package
new 65dad40 Add instructions for creating Debian packages in installation
page.
new 515ed0e SINGA-274 Create Debian packages
new f4e4a78 SINGA-274 Create Debian packages
new 734009e SINGA-274 Create Debian packages
new 771d0ee add ca-certificates
new a3446b5 update README.md of the repo. update the docs for debian
pacakge fix a bug in build.sh (the sed command) fix the notebook link
new 45efd35 SINGA-292 Update CMakeLists.txt for compiling OpenBlas and
Protobuf
new 8621f58 SINGA-293 - Bug from compiling PySINGA on Mac OS X with
multiple version of Python
new 8dac86c SINGA-274 Create Debian packages
new c3ce75e SINGA-276 Create docker images
new ab7221e SINGA-276 Create docker images
new 9a0a3fd SINGA-276 Create docker images
new 9ea8b53 SINGA-276 Create docker images
new 1ec625f replace markdown table with html table for the installation
page as recommonmark does not support github flavor markdown syntax
new 6b5ec2a add border for the tables in the installation page.
new f647d68 SINGA-276 Create docker images
new 3d40706 SINGA-296 - Add sign and to_host function for pysinga tensor
module
new 2d5f696 re-organize the installation page and update the index page
new 6b70dfc Patches imported from SINGA-257 to update the OpenCL API.
new 5d3da92 Enabled python API for Singa-OpenCL.
new 8101f00 Added FindOpenCL.cmake retrieved from viennacl-dev to enable
linking of OpenCL libs on some computers.
new d190fa8 SINGA-295 - Add an example of image classification using
GoogleNet
new 5d220cd prepare v1.1-rc1; add release notes; add license for some
files
new 50a8997 Update license date to 2017
new 5a6c303 update release notes for v1.1-rc1
new b2e21c6 preparing for v1.1-rc1: remove outdated files; fix links
new 2a1b988 add googlenet into model zoo page
new 3a35635 preparing v1.1-rc1: updating version No. from v1.0.1 to v1.1.0
new 43497bc preparing v1.1-rc1: remove prototxt files; fix license
issues; update package version to 1.1.0
new 59ca44a preparing v1.1-rc1: update the path to numpy.i in LICENSE
new 6a4a0f7 SINGA-299 - Remove glog from travis.yml
new c0317d1 SINGA-300 - Add residual networks for imagenet classification
new 45ec92d SINGA-300 - Add residual networks for imagenet classification
new 9b2d161 update docs for the release of v1.1.
new 026db89 SINGA-300 - Add residual networks for imagenet classification
new 8ddfab5 SINGA-300 - Add residual networks for imagenet classification
new f2ad933 SINGA-300 Add residual networks for imagenet classification
new 8cc453a SINGA-301
new f6cf8f5 SINGA-301 Hardcode OpenCL source code as static strings
new cde701a SINGA-274 Create Debian packages
new 6fbc4e9 SINGA-302 - Extend the travis config for building SINGA on OSX
new 5d17266 SINGA-303 Create conda packages
new dde8d14 develop and test precision for multi-label metric
new de8a654 Finish precision and recall for multi-label metrics
new f7caf46 fix a bug when provided metric is None
new 3417dd9 SINGA-303 Create conda packages
new 493fc65 SINGA-303 Create conda package
new ca0be88 SINGA-303 Create conda packages
new a3f34cb SINGA-303 Create conda packages
new e8e4660 update installation page for instructions of using conda to
install singa export the protobuf path when compiling singa on osx using travis
new 97781d0 Merge PR #309; test passed for non-python version;
new 3206904 remove duplicated precision
new 0be3e03 Merge PR #312
new a6e7690 To install all headers via 'make install'
new 2247f11 SINGA-306 - Write the definition of USE_GLOG into
singa_config.h
new 5144bcf SINGA-305 - Add jupyter notebooks for SINGA V1 tutorial
new d238042 remove ADD_DEFINITION(DUSE_GLOG) in Dependencies.cmake to
avoid compilation warning
new 0695daa Merge PR #314
new 0d4b97e Merge branch 'SINGA-306'
new 5d259e1 Documentation: Add logos and links for Github, Docker, AWS
Marketplace and Jupyter.
new 1e30a8b Merge branch 'SINGA-263' from PR #316
new 2aa1d22 fix bugs in layer::param_names() and tensor::sum_columns()
(device was not set)
new e0cae3c fix a typo
new 5d50f22 SINGA-274 Improve Debian packaging with CPack
new 85dbad7 Merge PR #317 for creating debian package
new 38da789 SINGA-309 Update the layer setting/config dynamically
new 4fd44ed SINGA-311 Add unittest for Adam updater
new f5cb70c SINGA-310 Create conda packages on CentOS
new b35e03f Merge PR #319
new 475d264 SINGA-223 Update Apache Incubator logo
new 2608d41 Fix broken Markdown headings
new e527db9 Enable configuration of cudnn workspace size limit in PySINGA
new 2d1dd42 Fix a typo for cudnn_prefer and change the default workspace
limit size to 1024MB
new 0815391 SINGA-313 Add L2 norm layer
new fa4f631 SINGA-312 Rename layer parameters
new ea078dc SINGA-315 Reduce memory footprint by Python generator for
parameter gradient
new be093f1 SINGA-316 Add SigmoidCrossEntropy
new 3415099 SINGA-317 Extend ImageBatchIter to read labels in general
format
new 5aecb47 SINGA-318 - Extend ImageBatchIter to read image list file
without meta info
new b6dba21 Update Chinese doc page
new 336ef55 SINGA-319 Replace Singa to SINGA in all files in the code and
documentation
new fefcab9 SINGA-319 Fix link errors on the index page
new 34d6102 Fix link bugs in Chinese page
new 4e7db05 SINGA-315 - Reduce memory footprint by Python generator for
parameter gradient
new 6c39e45 Merge branch 'SINGA-315'
new fd06722 Merge PR #328
new 62f5e18 Add vertical flip function and method in image_tool.py
new cd485a7 Fix the problem when batchnorm layer connect to dense layer
new e0d2738 random crop and resize
new 07ef01f Average gradients otherwise loss overflows
new f0ebf1e fix a bug in test_net.py due to the update of backward() API
new 12db1be Merge branch 'master' of
https://github.com/apache/incubator-singa
new 2ce7229 SINGA-324 Extend RNN layer to accept variant seq length
across batches
new 6a162ad SINGA-325 - Fix the bug of numpy version mis-match
new d0855e2 update the version of openblas to 0.2.14 which is the lowest
version on anaconda cloud
new 802e5e2 update the version of swig and numpy in meta.yaml
new 334c27d update depdent lib version according to their availability on
anaconda cloud
new 2cdc172 SINGA-326 - Add Inception V4 for ImageNet classification
new c628e90 update the openblas version to 0.2.19 which is available on
anaconda cloud
new 9eea5b5 Merge branch 'SINGA-324'
new a94fa10 Merge branch PR #332
new 913417a SINGA-327 Fix the bug of loading model saved in bin file
new 4fe5271 SINGA-328 - Add VGG models for ImageNet classification
new b6874d4 SINGA-329 - Support layer freezing during training
(fine-tuning)
new fc4d1cc add inception v3
new 59b0167 add inception v3
new 40124db udpate inception_v4 for the final and aux endpoints; update
the meta.yml as the protobuf binary file is not backward compatible. singa
compiled on protobuf3.0 cannot run with protobuf 3.2 runtime.
new 3f39cfa convert param for both v3 and v4
new e67bda9 fix a bug for random_crop_resize which would result in
negative crops
new 63c6ae1 fix a bug which calls randint with empty range
new a39ed5c SINGA-326 - Add Inception V4 for ImageNet classification
new e5c438c update openblas version to 0.2.19
new 14f0d8c SINGA-290 Upgrade to Python 3
new c94b3df SINGA-290 Upgrade to Python 3
new 8c9b594 SINGA-290 Upgrade to Python 3
new bfeb612 SINGA-290 Upgrade to Python 3
new bd5a8f8 SINGA-290 Upgrade to Python 3
new ff1806b SINGA-290 Upgrade to Python 3
new 01b7a74 SINGA-290 Upgrade to Python 3
new 57dbe42 update conda build files for py3
new 0f86cd5 SINGA-290 Upgrade to Python 3
new 4bd6827 update swig to 3.0.10 for python 3
new 0330af0 SINGA-290 Upgrade to Python 3
new 8d03bd8 SINGA-290 Upgrade to Python 3
new bf121fd Singa-330 fix openblas building on i7 7700k
new d30920f SINGA-290 Upgrade to Python 3
new aa9747b Merge branch 'SINGA-328'
new 074c551 update documentation for py3
new 4e60fa2 update the documentation about the schedule and community
new f5af1c3 SINGA-331 Fix the bug of tensor division operation
new 7827351 SINGA-332 Fix the bug of loading data in train.py of cifar10
new 41d06b0 Fix SWIG version for Python 2
new 7a641bf SINGA-223 Update the documentation
new b4ea650 Merge branch 'PR346' into jan22
new 95b535d SINGA-345 Update Jenkins and fix bugs in compliation
new 163452e SINGA-345 Update Jenkins and fix bugs in compliation
new e09dff4 SINGA-348 Support autograd MLP Example
new 755eba6 SINGA-348 Support autograd MLP Example
new f42d4d0 SINGA-348 Support autograd MLP Example
new 060e7df SINGA-348 Support autograd MLP Example
new c9f9471 SINGA-346 Update cudnn from V5 to V7
new a88efa0 Singa-341 Added stride functionality to tensors for CPP
new e27498d SINGA-347 Create a function that supports einsum provide the
tensordot function to do the tensor multiplication of two tensors along
specified axes
new 9a3ce58 SINGA-347 Create a function that supports einsum 1.test the
tensordot function and fix some error in the function 2.tweak the code to be
more readable and fix the some error in the comment
new f595f10 SINGA-347 Create a function that supports einsum 1. provide
functions to calculate einsum calculation but still needs one function to make
the tensor diagonalization along one axis 2. give a credit to numpy for
tensordot
new b09aed3 SINGA-347 Create a function that supports einsum 1.provide
the diag() function which could make the matrix diagonalized along the given
axis 2.provide some comments for some existing function 3. to do: provide
complete comments and fix the codes. It still needs test under Singa environment
new 2ec06ed SINGA-347 Create a function that supports einsum 1. fix some
calculation error in the tensor mult function 2. by doing the fixing, the test
using numpy could pass
new ef5fdaf SINGA-346 Update cudnn from V5 to V7
new 98e9d2f SINGA-346 Update cudnn from V5 to V7
new 4ccb72e SINGA-346 Update cudnn from V5 to V7
new 9090160 SINGA-346 Update cudnn from V5 to V7
new 2ec1364 SINGA-347 Create a function that supports einsum assuming
have the numpy.repeat, finish the einsum function (using elementwisemult, no
need to use tensordot)
new fa1b023 SINGA-347 Create a function that supports einsum 1. finish
the einsum function in python but still needs some funciton to do in cpp 2.
function TODO list to finish the function in cpp(just like numpy
function):a.sum(A,axis = None) b.repeat(A,repeats) c.transpose(A,axes = None)
new c037e2f SINGA-347 Create a function that supports einsum 1. revise to
make it similar to numpy.einsum 2. write an unit test for einsum
new 90015d6 SINGA-347 Create a function that supports einsum remove idea
new 4f5f902 SINGA-347 Create a function that supports einsum 1. delete
.idea
new f548fa3 SINGA-347 Create a function that supports einsum revise the
example
new 08675e3 Merge branch 'PR355'
new b2cbbb8 SINGA-223 Fix side navigation menu in the website
new fb7b0d3 Merge branch 'PR358'
new 72b1a69 SINGA-350 Error from python3 test
new 2d25561 SINGA-350 Error from python3 test
new 362c873 SINGA-352 Complete SINGA documentation in Chinese version
new bd774b5 SINGA-350 Error from python3 test
new 26db8d2 SINGA-353 Fix the bug of test_einsum Fix the bug of dimension
mismatch in test_einsum
new 6b1ed79 Merge branch 'pr361'
new 3a37398 Merge branch 'pr360'
new f73ae5a Merge branch 'pr358'
new faa22f7 SINGA-354 Update travis scripts to use conda-build for all
platforms
new 30ad60d SINGA-294 Add instructions for building on Windows
new ceea70c SINGA-294 Add USE_OPENCL
new 4b3a914 SINGA-354 Update travis scripts to use conda-build for all
platforms
new c7f751e use absolute python path
new 24963c0 SINGA-354 Update travis scripts to use conda-build for all
platforms
new 9696255 SINGA-354 Update travis scripts to use conda-build for all
platforms
new 43c33e9 SINGA-352 Some fixes in documentation
new 8f67ceb SINGA-294 Remove old instructions
new a6c1b8e SINGA-358: Consolidated RUN steps and cleaned caches in
Docker containers
new 9281cc4 SINGA-352 Hide past releases by default
new 00ba636 Merge branch 'pr363'
new 0407efb update the conda channel of recommonmark
new b5600d3 Merge branch 'SINGA-352'
new 26101ee Singa-351 Added stride support and cudnn codes to cuda
new 75f9a0e misc. changes and further abstraction of some cudnn codes
new a44d2e7 Merge branch 'master' of github.com:apache/incubator-singa
into SINGA-341-351
new c92ddb7 Merge branch 'pr366' into latest
new c61a0d8 SINGA-294 Add instructions to run unit tests on Windows
new 62073ee SINGA-359 Create alias for conda packages
new d580419 SINGA-359 Create alias for conda packages
new 35b9fed SINGA-359 Create alias for conda packages
new 00573e2 SINGA-359 Create alias for conda packages
new c5ca8e8 SINGA-337 Add test cases for code
new c52e2aa Streamlining of tensor.h file by moving respective member
functions to cpp or cuda file. Removal of shape_multipliers_ attribute in
tensor.h. Changed read-in tensors to be passed as reference instead of pointer
new 6d43cc9 SINGA-361 Add git instructions for contributors and committers
new 3e2b75c reformat the code
new 2195af2 SINGA-294 Add instructions for CUDA support on Windows
new 394d78d Merge branch 'pr373'
new 600f27e Merge branch 'pr367' into latest
new fa0ed01 fix the bug from using type Shape
new 1bee4d2 SINGA-356 Error using travis tool to build SINGA on mac os
new 92b892a Merge branch 'pr370'
new 9c0e0af update installation doc
new 18c346f SINGA-294 Add instructions to run CUDA unit tests on Windows
new 84189a2 SINGA-366 Add social networks links to SINGA website
new 30e704c Merge branch 'pr378'
new bfd8ce9 SINGA-368 Fix the bug in Cifar10 examples
new 6bcd5d0 update the cifar10 example
new dd58f49 SINGA-363 Fix some bugs in imagenet examples
new 01b29b3 SINGA-369 fix the errors in examples 1.fix the error of
"Tensor object has no attribute singa_tensor"
new d619e44 SINGA-349 Create layer operations for autograd
new a9d495a SINGA-349 Create layer operations for autograd
new 195b4d4 SINGA-349 Create layer operations for autograd
new 51c242b SINGA-349 Create layer operations for autograd
new e4610e0 SINGA-349 Create layer operations for autograd
new 5abcc6e SINGA-349 Create layer operations for autograd
new 6402a53 SINGA-349 Create layer operations for autograd
new b136ac0 SINGA-349 Create layer operations for autograd
new f70f120 SINGA-349 Create layer operations for autograd
new 8146852 SINGA-349 Create layer operations for autograd
new ed464ef SINGA-349 Create layer operations for autograd
new 40e609a SINGA-349 Create layer operations for autograd
new 403843d SINGA-349 Create layer operations for autograd
new 6d7d629 SINGA-349 Create layer operations for autograd
new 8a1d98a SINGA-363 Add DenseNet for Imagenet classification
new 32b00c0 SINGA-369 fix the errors in examples fix the error of
UnicodeDecodeError
new 0cf960c Merge branch 'pr382'
new 29b87b9 Merge branch 'pr381'
new 01ed877 SINGA-369 the errors of examples in testing
new c5b454d SINGA-369 the errors of examples in testing
new 760df2f SINGA-364 V1.2 Release Preparation
new c761dc8 SINGA-364 V1.2 Release Preparation
new c67c3b6 SINGA-364 V1.2 Release Preparation
new c343ff9 SINGA-352 Complete SINGA documentation in Chinese version
new 16c6111 SINGA-362 Add functions to support einsum function
new b235e9b SINGA-352 Add Chinese documentations for SINGA examples
new 7d25ed9 SINGA-362 Add funcitons to support einsum function 1. fix the
bug and support some test case 2. need to do some fix and use transpose
function developed by yisen
new 5e8f6a4 SINGA-362 Add functions to support einsum function 1.add the
transpose function in python and use the function in existing functions 2.one
problem need to fix: the tensor cannot be reshaped after transpose
new 8d9eb29 SINGA-362 Add functions to support einsum function 1.fix one
problem in device.cc 2.use add(t,0) to reset the stride and could use reshape
after transpose
new f9e7caa SINGA-370 Improvement to tensor reshape and various misc.
changes related to SINGA-341 and 351
new 4940fef SINGA-362 Add functions to support einsum function 1. change
the api of repeat,reshape,transpose to be similar to numpy api 2. have some
change in the reshape function to make it modified to 'Tensor Reshape' instead
of 'void Reshape'(it is same as Yisen's revise)
new 04f2c3c Update the webpages (links) for V1.2.0
new f7a6190 fix some typos in links and Chinese doc
new 59ef841 fix a typo in download page
new a515c38 Merge branch 'pr386'
new 810fed8 SINGA-376 Update the download page
new dcdd6ea SINGA-376 Update the download page
new 30ac41b SINGA-371 Implement functional operations in c++ for autograd
new fc181cd SINGA-371 Implement functional operations in c++ for autograd
new af95cc1 SINGA-371 Implement functional operations in c++ for autograd
new d48dea0 SINGA-371 Implement functional operations in c++ for autograd
new 2cac057 SINGA-371 Implement functional operations in c++ for autograd
new c57b87a SINGA-371 Implement functional operations in c++ for autograd
new 78e1fc2 SINGA-371 Implement functional operations in c++ for autograd
new 5c8504a SINGA-371 Implement functional operations in c++ for autograd
new b944795 SINGA-376 Update the download page
new dfe4478 SINGA-371 Implement functional operations in c++ for autograd
new 381d318 Apache V2 License header added to readme
new f03c188 format the indent of download page
new 24d1189 Apache V2 License header added to all readme files
new 9b06c29 Merge branch 'readme_update'
new 189958a SINGA-371 Implement functional operations in c++ for autograd
new e68ea2e SINGA-371 Implement functional operations in c++ for autograd
new f7ff0be SINGA-352 Synchronize Chinese docs with English ones
new c8df172 Merge branch 'pr388'
new e248e44 Merge branch 'pr394'
new 10f3aa1 SINGA-362 Add functions to support einsum function delete the
repetitive reshape and transform, which are the same as yisen
new 7a19e63 Merge branch 'pr385'
new aa9c52a SINGA-371 Implement functional operations in c++ for autograd
new 82ef417 SINGA-371 Implement functional operations in c++ for autograd
new 15c0230 SINGA-371 Implement functional operations in c++ for autograd
new 5340b65 SINGA-371 Implement functional operations in c++ for autograd
new 4a45ee6 SINGA-371 Implement functional operations in c++ for autograd
new ac5f4eb SINGA-371 Implement functional operations in c++ for autograd
new 56292f1 Merge branch 'pr387'
new e209203 SINGA-346 Update cudnn from V5 to V7
new e16cea1 SINGA-371 Implement functional operations in c++ for autograd
new 81908a8 SINGA-381 - Update the autograd API to yeild the gradients
new 58e6640 SINGA-380 Fix bugs from Reshape
new b30d7ea SINGA-380) Fix bugs from Reshape
new a105b24 SINGA-379 Implement batchnorm operation and its related
functions for autograd
new 10274f3 SINGA-379 Implement batchnorm operation and its related
functions for autograd
new 8654f89 SINGA-379 Implement batchnorm operation and its related
functions for autograd
new ce1a733 SINGA-379 Implement batchnorm operation and its related
functions for autograd
new f134a24 SINGA-379 Implement batchnorm operation and its related
functions for autograd
new 571818e SINGA-378 Implement maxpooling operation and its related
functions for autograd
new fb5cb9a SINGA-378 Implement maxpooling operation and its related
functions for autograd
new a362918 SINGA-378 Implement maxpooling operation and its related
functions for autograd
new 054f303 SINGA-382 Implement concat operation for autograd
new 76779be SINGA-382 Implement concat operation for autograd
new 2b5c3f7 SINGA-384 Implement ResNet using autograd API
new 117dfcf SINGA-385 Add new python module for optimizers
new e651c1a SINGA-384 Implement ResNet using autograd API
new 870c5df SINGA-382 Implement concat operation for autograd
new 2cd08c4 modify the command for conda-build
new a025d72 SINGA-223 Update team page
new 08a232f change the source field back to git url in conda meta file
new da1e2b1 Merge branch 'pr404'
new f2f4d1f Merge branch 'pr405'
new 6c28abd SINGA-387 Modified the design of autograd backward engine and
correct some mistakes in it
new 2fea345 SINGA-387 Modified the design of autograd backward engine and
correct some mistakes in it
new b55b046 SINGA-387 Modified the design of autograd backward engine and
correct some mistakes in it
new 770d6cd SINGA-387 Modified the design of autograd backward engine and
correct some mistakes in it
new 7df6a5d SINGA-388 Develop some RNN layers by calling tiny operations
like matmul, addbias
new 5dc17b9 SINGA-388 Develop some RNN layers by calling tiny operations
like matmul, addbias
new a44a01c SINGA-388 Develop some RNN layers by calling tiny operations
like matmul, addbias
new 65756e6 SINGA-390 Update installation.md
new 0cd4e30 SINGA-388 Develop some RNN layers by calling tiny operations
like matmul, addbias
new 2224d5f Merge branch 'pr407'
new ca70bdf SINGA-383 Add Separable Convolution for autograd
new d5422a4 SINGA-383 Add Separable Convolution for autograd
new b1610d7 SINGA-344 Add a GAN example
new 8aac80e SINGA-383 Add Separable Convolution for autograd
new f8cd7e3 Merge branch 'pr410'
new 1ecafde Add Mean Square Error loss operation and its unit test case
new eec0d52 Merge branch 'pr410'
new db92c75 SINGA-392 Update autograd API to Pytorch style
new 3769b4b SINGA-394 Add Rafiki as a submodule of Singa
new 84e210b update rafiki to the latest commit
new 9aebe2f SINGA-395 Add documentation for autograd APIs
new f720a8f SINGA-395 Add documentation for autograd APIs
new 4a1b1e2 SINGA-396 Upgrade `protobuf` to 3.6.1 to match Tensorflow's
new 8143cf4 SINGA-395 Add documentation for autograd APIs
new 90256fc SINGA-408 Unsupportive function definition in python3
new e08af42 Merge branch 'pr419'
new e29ef5f use tqdm for progress bar in cifar10 example
new 99bae02 SINGA-409 [Singa 1.2.0] Basic `singa-cpu` import throws error
new 3e10c03 Merge branch 'pr418' into latest
new 3d688be switch off the verbose mode for vgg example
new c67e8c0 SINGA-405 Add How to Release instructions
new 1283441 SINGA-410 Add two function, set_params() and get_params(),
for Autograd Layer class
new 43c6bd6 SINGA-410 Add two function, set_params() and get_params(),
for Autograd Layer class
new e1b3975 SINGA-411 Local variable 'params' referenced before assignment
new 01224bb SINGA-410 Add two function, set_params() and get_params(),
for Autograd Layer class
new 7470b20 SINGA-411 Local variable 'params' referenced before assignment
new f506ef2 SINGA-411 Local variable 'params' referenced before assignment
new 9846092 SINGA-415 Moving SINGA to Gitbox
new 73f9637 SINGA-411 Local variable 'params' referenced before assignment
new d7383c3 SINGA-410 Add two function, set_params() and get_params(),
for Autograd Layer class
new 12cfe83 Merge pull request #425 from moazreyad/SINGA-415
new ae8eb7c Merge pull request #422 from xuewanqi/set_get_params
new bd2df8f Merge pull request #421
new b125221 SINGA-417 #close adding a link to Apache Security page
new f712520 update recent news
new a98d293 Merge pull request #428 from moazreyad/SINGA-417
new 66cae9e Add Rafiki paper, fix broken link
new 0aea1d5 SINGA-418 Unsupportive type 'long' in python3
new a24f5fd Merge pull request #430 from xuewanqi/SINGA-418
new fbce911 add autograd to index and update year
new 98d25f6 SINGA-405 Fix CO50 in SINGA maturity model assessment
new b4b4329 Merge pull request #429 from moazreyad/singa-site-rafiki
new 23adf09 SINGA-422 ModuleNotFoundError: No module named '_singa_wrap'
new 20c74c9 Merge pull request #432 from nudles/SINGA-422
new 0d6b1ae SINGA-422 ModuleNotFoundError: No module named '_singa_wrap'
The 1082 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
