chrishkchris opened a new pull request #549: SINGA-493 Update news: SINGA
graduation as a TLP
URL: https://github.com/apache/incubator-singa/pull/549
The added message is :
"SINGA was graduated as a Top-Level Project (TLP) of the Apache Software
Foundation in October 2019."
The b
chrishkchris commented on issue #549: SINGA-493 Update news: SINGA graduation
as a TLP
URL: https://github.com/apache/incubator-singa/pull/549#issuecomment-543056342
Has changed the message "SINGA has graduated as a Top-Level Project (TLP) of
the Apache Software Foundation on 16 Oct 2019."
chrishkchris opened a new pull request #2: SINGA-493 Update news: SINGA
graduation as a TLP
URL: https://github.com/apache/incubator-singa-site/pull/2
-
chrishkchris commented on issue #549: SINGA-493 Update news: SINGA graduation
as a TLP
URL: https://github.com/apache/incubator-singa/pull/549#issuecomment-543065170
--
nudles merged pull request #549: SINGA-493 Update news: SINGA graduation as a
TLP
URL: https://github.com/apache/incubator-singa/pull/549
This is an automated message from the Apache Git Service.
To respond to the message, p
nudles merged pull request #2: SINGA-493 Update news: SINGA graduation as a TLP
URL: https://github.com/apache/incubator-singa-site/pull/2
This is an automated message from the Apache Git Service.
To respond to the message, p
chrishkchris opened a new pull request #550: SINGA-493 Update incubator-singa
to singa
URL: https://github.com/apache/incubator-singa/pull/550
Corrected many words fro
chrishkchris opened a new pull request #3: SINGA-493 Update incubator-singa to
singa
URL: https://github.com/apache/incubator-singa-site/pull/3
Corrected many word fro
nudles merged pull request #3: SINGA-493 Update incubator-singa to singa
URL: https://github.com/apache/incubator-singa-site/pull/3
This is an automated message from the Apache Git Service.
To respond to the message, please l
chrishkchris opened a new pull request #4: SINGA-493 Update some text
URL: https://github.com/apache/incubator-singa-site/pull/4
Updated some text
This is an automated message from the Apache Git Service.
To respond to the mes
nudles merged pull request #4: SINGA-493 Update some text
URL: https://github.com/apache/incubator-singa-site/pull/4
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub
nudles merged pull request #550: SINGA-493 Update incubator-singa to singa
URL: https://github.com/apache/incubator-singa/pull/550
This is an automated message from the Apache Git Service.
To respond to the message, please lo
nudles merged pull request #540: SINGA-476 added softmax with axis
URL: https://github.com/apache/incubator-singa/pull/540
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
Wentong-DST opened a new pull request #551: [SINGA-495] Improve Java tensor
test case format
URL: https://github.com/apache/incubator-singa/pull/551
This is an automated message from the Apache Git Service.
To respond to the
Wentong-DST opened a new pull request #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/incubator-singa/pull/552
- Softsign function:
https://www.tensorflow.org/versions/r1.15/api_docs/python/tf/nn/softsign
- Softplus function:
moazreyad opened a new pull request #553: SINGA-493 Remove incubator from the
website
URL: https://github.com/apache/singa/pull/553
This is an automated message from the Apache Git Service.
To respond to the message, please
nudles merged pull request #553: SINGA-493 Remove incubator from the website
URL: https://github.com/apache/singa/pull/553
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
nudles commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-545962476
Thanks for the PR!
Could you add the two functions in
https://github.com/apache/singa/blob/master/src/core/tensor/tens
chrishkchris opened a new pull request #5: SINGA-493 Remove incubator from the
website
URL: https://github.com/apache/singa-site/pull/5
SINGA-493 Remove incubator from the website
This is an automated message from the Apache
nudles merged pull request #547: SINGA-493 Add an additional build path
URL: https://github.com/apache/singa/pull/547
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHu
nudles merged pull request #5: SINGA-493 Remove incubator from the website
URL: https://github.com/apache/singa-site/pull/5
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
Wentong-DST commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-546371709
> Thanks for the PR!
> Could you add the two functions in
https://github.com/apache/singa/blob/master/src/core/te
dcslin opened a new pull request #554: [bug]batchnorm running mean var not
updated
URL: https://github.com/apache/singa/pull/554
Hi team,
Shouldn't `autograd` `batchnorm` class update its running_mean and
running_var from c++ backend? I could not find anywhere `running_mean/var` are
u
dcslin commented on issue #548: SINGA-476 DNNL(upgrade of MKLDNN) integration
and add softmax
URL: https://github.com/apache/singa/pull/548#issuecomment-546712187
this is ready for review. thank you.
This is an automated mess
chrishkchris commented on issue #554: [bug]batchnorm running mean var not
updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546714563
I think this is pass by reference (the running mean and variance singa
tensors' blocks are fed to the function cudnnBatchNormalizationForwa
chrishkchris edited a comment on issue #554: [bug]batchnorm running mean var
not updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546714563
I think this is pass by reference (the running mean and variance singa
tensors' blocks are fed to the function
cudnnBatchNormalizat
chrishkchris commented on issue #554: [bug]batchnorm running mean var not
updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546757182
Oh I see. The running mean and var are updated using NOT in place assignment.
running_mean = running_mean * bnh.factor + mean * (1 - bnh.
chrishkchris commented on issue #554: [bug]batchnorm running mean var not
updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546778720
Yes, I see why the running mean is not updated. This is because
running_mean = running_mean * bnh.factor + mean * (1 - bnh.factor);
It
chrishkchris removed a comment on issue #554: [bug]batchnorm running mean var
not updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546757182
Oh I see. The running mean and var are updated using NOT in place assignment.
running_mean = running_mean * bnh.factor + mean * (
chrishkchris removed a comment on issue #554: [bug]batchnorm running mean var
not updated
URL: https://github.com/apache/singa/pull/554#issuecomment-546778720
Yes, I see why the running mean is not updated. This is because
running_mean = running_mean * bnh.factor + mean * (1 - bnh.factor
dcslin closed pull request #554: [bug]batchnorm running mean var not updated
URL: https://github.com/apache/singa/pull/554
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
dcslin commented on issue #554: [bug]batchnorm running mean var not updated
URL: https://github.com/apache/singa/pull/554#issuecomment-547283131
Hi @chrishkchris thank you for the help. This is false alarm. the code is
working as @chrishkchris described.
---
joddiy commented on issue #541: SINGA-476 added 4d test on batchnorm
URL: https://github.com/apache/singa/pull/541#issuecomment-547385195
ready to merge
This is an automated message from the Apache Git Service.
To respond to t
chrishkchris opened a new pull request #555: SINGA-490 Optimization on GPU
Malloc and Cudastream
URL: https://github.com/apache/singa/pull/555
The performance is improved after some optimization on GPU Malloc and
Cudastream:
See the performance (1) after, and (2) before the modifica
dcslin commented on issue #514: SINGA-482 tc comprehension integration
URL: https://github.com/apache/singa/pull/514#issuecomment-548703862
Ready for review.
This is an automated message from the Apache Git Service.
To respond
moazreyad closed pull request #323: SINGA-290 Upgrade to Python 3
URL: https://github.com/apache/singa/pull/323
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and
moazreyad closed pull request #340: SINGA-274 Create Debian packages
URL: https://github.com/apache/singa/pull/340
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub a
moazreyad opened a new pull request #556: SINGA-493 Update team page
URL: https://github.com/apache/singa/pull/556
Change PPMC to PMC
Remove "Mentors" since SINGA is graduated now from the incubator. All the
mentors are now part of the PMC.
Add Xiangrui Cai as committer
Correct She
dcslin opened a new pull request #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub an
joddiy commented on issue #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557#issuecomment-549822479
Hi, shicong, I have add some test cases in autograd. It seems this implement
works fine for __add__, __sub__, but return all zero values for __mul__. can
you ch
chrishkchris opened a new pull request #558: SINGA-487 Parallelize Computation
and Communication using CUDA stream concurrency
URL: https://github.com/apache/singa/pull/558
This PR deals with the Parallelization of Computation and Communication
using CUDA stream concurrency, which can redu
nudles merged pull request #558: SINGA-487 Parallelize Comp. and Comm. using
CUDA stream concurrency
URL: https://github.com/apache/singa/pull/558
This is an automated message from the Apache Git Service.
To respond to the m
chrishkchris opened a new pull request #559: SINGA-493 Update downloads.md
URL: https://github.com/apache/singa/pull/559
Updated the download page to add "incubating" after the names of the previous
releases in the downloads page.
chrishkchris commented on issue #555: SINGA-490 Optimization on GPU Malloc and
Cudastream
URL: https://github.com/apache/singa/pull/555#issuecomment-550299437
Before I do further checking of the code, I submit the passed unit test
results first: (1) CPP level operation unit test, (2) Pytho
chrishkchris commented on issue #559: SINGA-493 Update downloads.md
URL: https://github.com/apache/singa/pull/559#issuecomment-550325105
In additional to downloads.md,
I have also added the change of doc/en/develop/how-to-release.rst in this PR,
However, for the following lines
chrishkchris commented on issue #555: SINGA-490 Optimization on GPU Malloc and
Cudastream
URL: https://github.com/apache/singa/pull/555#issuecomment-550738649
I checked the source code of unit test which covers the value check for
functions like sumrow and sumcolumn, so the value should be
chrishkchris edited a comment on issue #555: SINGA-490 Optimization on GPU
Malloc and Cudastream
URL: https://github.com/apache/singa/pull/555#issuecomment-550738649
I checked the source code of unit test which covers the value check for
functions like sumrow and sumcolumn, so the value sh
joddiy commented on issue #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557#issuecomment-550752560
Hi, shicong, please use this to test __mul__
```
def test_tensor_add_api(self):
def _run_test(s1, s2):
x_0 = np.random.random(s1).asty
dcslin commented on issue #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557#issuecomment-551142223
Hi @joddiy thank you for the info. this is tested and test case are updated
This is an automated
chrishkchris opened a new pull request #560: SINGA-487 Accumulate gradients to
reduce network latency
URL: https://github.com/apache/singa/pull/560
This PR reduces the network latency by accumulate gradients in a memory
buffer before sending out with NCCL.
This can reduce much of the TC
FeynmanDNA opened a new pull request #561: update gitignore for Sphinx-Doxygen
doc built files
URL: https://github.com/apache/singa/pull/561
current doc-site generated files are not git checked, thus proposed this
gitignore update
--
nudles merged pull request #561: update gitignore for Sphinx-Doxygen doc built
files
URL: https://github.com/apache/singa/pull/561
This is an automated message from the Apache Git Service.
To respond to the message, please l
nudles merged pull request #551: SINGA-495 Improve Java tensor test case format
URL: https://github.com/apache/singa/pull/551
This is an automated message from the Apache Git Service.
To respond to the message, please log on
nudles merged pull request #556: SINGA-493 Update team page
URL: https://github.com/apache/singa/pull/556
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use th
nudles merged pull request #560: SINGA-487 Fusing gradients to reduce network
latency
URL: https://github.com/apache/singa/pull/560
This is an automated message from the Apache Git Service.
To respond to the message, please
Wentong-DST commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553225239
@chrishkchris , can you please help review? Thanks.
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553226326
ok, I will do the peer review. Thanks!
This is an
chrishkchris commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r345563360
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tens
dcslin commented on issue #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557#issuecomment-553279138
This is ready for merging
This is an automated message from the Apache Git Service.
To respond t
chrishkchris commented on issue #555: SINGA-490 Optimization on GPU Malloc and
Cudastream
URL: https://github.com/apache/singa/pull/555#issuecomment-553281211
Double checking of cuda codes (I removed the cuda memset, so I need to check
everything related to cuda)
I have to check thr
Wentong-DST commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r345883207
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tenso
Wentong-DST commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553873183
Hi @chrishkchris , I have added test cases for these two cuda functions.
Please help review! Thanks.
--
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553887595
This is the build log. We can see the missing codes from the log.
Since you put cuda::softsign(num, inPtr, outPtr
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553887595
This is the build log. We can see the missing codes from the log.
Since you put cuda::softsign(num, inPtr,
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553887595
This is the build log. We can see the missing codes from the log.
Since you put cuda::softsign(num, inPtr,
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553900186
For example, it is something like this (the following is for reference which
is not tested):
1. In math_kern
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553900186
For example, it is something like this (the following is for reference which
is not tested):
1. In ma
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553900186
For example, it is something like this (the following is for reference which
is not tested):
1. In ma
Wentong-DST commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-553999136
Hi @chrishkchris , I have added code as suggested, however, I still observe
some errors while running tests.
So
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554126176
In the test case, you used the tensor function which output a tensor p with
a input tensor cc, but the function can
Wentong-DST commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554618286
Hi @chrishkchris , I have added two functions to header file of tensor,
built and passed the test.
Please help ve
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554625840
This is the running result of test case:
```
ubuntu@ip-172-31-16-93:~/incubator-singa/build/bin$ ./test_si
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554627773
For example, you can change the logic as follows:
aa=a is {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f}
so cc=a is {-2.
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554627773
For example, you can change the logic as follows:
aa=a is {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f}
so cc=a
chrishkchris edited a comment on issue #552: SINGA-496 Implement softplus and
softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-554627773
For example, you can consider the logic as follows:
aa=a is {1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f}
so cc=
chrishkchris commented on issue #559: SINGA-493 Update downloads.md
URL: https://github.com/apache/singa/pull/559#issuecomment-555047048
Concerning the above comments, the following is amended in
how-to-release.rst :
1. Line89*: svn add to “dist/release/singa”
2. Line113*:
chrishkchris commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r347767285
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tens
chrishkchris opened a new pull request #562: SINGA-487 Add support of gradient
compression to half precision
URL: https://github.com/apache/singa/pull/562
In this PR, I add an API in opt.py for using half precision in gradient
transfer.
Here is the training accurate test using 16 bi
joddiy commented on issue #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557#issuecomment-43615
ready to merge
This is an automated message from the Apache Git Service.
To respond to the messa
nudles merged pull request #555: SINGA-490 Optimization on GPU Malloc and
Cudastream
URL: https://github.com/apache/singa/pull/555
This is an automated message from the Apache Git Service.
To respond to the message, please l
nudles merged pull request #559: SINGA-493 Update downloads.md
URL: https://github.com/apache/singa/pull/559
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and us
nudles merged pull request #562: SINGA-487 Add support of gradient compression
to half precision
URL: https://github.com/apache/singa/pull/562
This is an automated message from the Apache Git Service.
To respond to the messa
nudles merged pull request #557: Broadcast cudnn test passing
URL: https://github.com/apache/singa/pull/557
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use
chrishkchris opened a new pull request #6: SINGA-493 Update download page and
how to release
URL: https://github.com/apache/singa-site/pull/6
The following web pages are updated:
1. Download SINGA
2. How to prepare a release
--
chrishkchris commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r348291980
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tens
chrishkchris commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r348291980
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tens
dcslin commented on issue #541: SINGA-476 added 4d test on batchnorm
URL: https://github.com/apache/singa/pull/541#issuecomment-555872515
Hi @nudles , @joddiy , conflicts are resolved.
This is an automated message from the Apa
chrishkchris opened a new pull request #563: SINGA-490 Improve Efficiency of
ReLU, Add, Sub, EltwiseMult
URL: https://github.com/apache/singa/pull/563
This PR deals with two issues: 1. Improve ReLU, 2. Optimize Add, Sub,
EltwiseMult after the last update with broadcasting:
1. The Re
Wentong-DST commented on a change in pull request #552: SINGA-496 Implement
softplus and softsign functions for tensor math
URL: https://github.com/apache/singa/pull/552#discussion_r350325886
##
File path: test/singa/test_tensor_math.cc
##
@@ -120,6 +120,32 @@ TEST_F(Tenso
chrishkchris commented on issue #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552#issuecomment-558432861
Thanks! I have double checked by building and running the test cases, where
no problem is found. The code looks goo
chrishkchris opened a new pull request #564: SINGA-487 Asynchronous training
algorithm with partial parameters synchronization
URL: https://github.com/apache/singa/pull/564
In this PR, there are two things added:
1. An experimental feature for distributed training is added in opt.py,
joddiy commented on issue #545: SINGA-494 Singa autograd improvement
URL: https://github.com/apache/singa/pull/545#issuecomment-558742160
@dcslin Hi, shicong, please check the singa op of Pow(both cpu and gpu) and
Div(only gpu). The pow doesn't have the correct result. And for the div op of
dcslin commented on issue #548: SINGA-476 DNNL(upgrade of MKLDNN) integration
and add softmax
URL: https://github.com/apache/singa/pull/548#issuecomment-558919970
HI @nudles , conflicts are resolved can ready for merge.
This
nudles merged pull request #564: SINGA-487 Asynchronous Training with Partial
Parameters Synchronization
URL: https://github.com/apache/singa/pull/564
This is an automated message from the Apache Git Service.
To respond to t
nudles merged pull request #563: SINGA-490 Improve Efficiency of ReLU, Add,
Sub, EltwiseMult
URL: https://github.com/apache/singa/pull/563
This is an automated message from the Apache Git Service.
To respond to the message,
nudles merged pull request #6: SINGA-493 Update download page and how to release
URL: https://github.com/apache/singa-site/pull/6
This is an automated message from the Apache Git Service.
To respond to the message, please log
nudles merged pull request #552: SINGA-496 Implement softplus and softsign
functions for tensor math
URL: https://github.com/apache/singa/pull/552
This is an automated message from the Apache Git Service.
To respond to the m
nudles merged pull request #541: SINGA-476 added 4d test on batchnorm
URL: https://github.com/apache/singa/pull/541
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub
dcslin commented on issue #545: SINGA-494 Singa autograd improvement
URL: https://github.com/apache/singa/pull/545#issuecomment-559392671
Hi @joddiy kindly help to review https://github.com/apache/singa/pull/565
much appreciated.
--
dcslin opened a new pull request #565: fixed broadcast div pow
URL: https://github.com/apache/singa/pull/565
fixed div and pow operator when broadcasting
broadcast tests covered add, sub, mul, pow, div on transpose cases
ref. https://github.com/apache/singa/pull/545
--
1 - 100 of 3428 matches
Mail list logo
