[jira] [Created] (SINGA-487) Distributed Module Development and Improvement
YEUNG SAI HO created SINGA-487: -- Summary: Distributed Module Development and Improvement Key: SINGA-487 URL: https://issues.apache.org/jira/browse/SINGA-487 Project: Singa Issue Type: New Feature Components: Module Reporter: YEUNG SAI HO Here I open a new Jira ticket for development and improvement of distributed learning module. E.g., Firstly, I will make the distributed module to be compatible with python multiprocessing. -- This message was sent by Atlassian Jira (v8.3.2#803003)
[jira] [Updated] (SINGA-487) Distributed Module Development and Improvement
[ https://issues.apache.org/jira/browse/SINGA-487?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] YEUNG SAI HO updated SINGA-487: --- Description: Here I open a new Jira ticket for long term development and improvement of distributed learning module. E.g., Firstly, I will make the distributed module to be compatible with python multiprocessing. was: Here I open a new Jira ticket for development and improvement of distributed learning module. E.g., Firstly, I will make the distributed module to be compatible with python multiprocessing. > Distributed Module Development and Improvement > -- > > Key: SINGA-487 > URL: https://issues.apache.org/jira/browse/SINGA-487 > Project: Singa > Issue Type: New Feature > Components: Module >Reporter: YEUNG SAI HO >Priority: Major > > Here I open a new Jira ticket for long term development and improvement of > distributed learning module. > E.g., Firstly, I will make the distributed module to be compatible with > python multiprocessing. > -- This message was sent by Atlassian Jira (v8.3.2#803003)
[jira] [Created] (SINGA-490) Optimize performance of stochastic gradient descent (SGD)
YEUNG SAI HO created SINGA-490: -- Summary: Optimize performance of stochastic gradient descent (SGD) Key: SINGA-490 URL: https://issues.apache.org/jira/browse/SINGA-490 Project: Singa Issue Type: Improvement Components: Core Reporter: YEUNG SAI HO I create a new Jira ticket to optimize the performance of stochastic gradient descent (SGD). The first step is to fuse the small operations so as to increase GPU computation efficiency and decrease latency. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Created] (SINGA-478) Python 3 uses __itruediv__ instead of __idiv__
YEUNG SAI HO created SINGA-478:
--
Summary: Python 3 uses __itruediv__ instead of __idiv__
Key: SINGA-478
URL: https://issues.apache.org/jira/browse/SINGA-478
Project: Singa
Issue Type: Improvement
Components: Core
Reporter: YEUNG SAI HO
We need to add __itruediv__ in tensor.py because the original __idiv__ is not
supported by python 3 anymore.
To understand the problem, let's study the following code first:
{code:java}
from singa import tensor
from singa import device
import numpy as np
Y = np.ones(shape=[10],dtype=np.float32) * 10.0
y = tensor.from_numpy(Y)
y.to_device(device.get_default_device())
def divide(y):
y /= 10
divide(y)
print(tensor.to_numpy(y))
{code}
Without adding the {color:#33}__itruediv__{color} function, the result is
as follows, which means that the /= operation is not in place:
{code:java}
[10. 10. 10. 10. 10. 10. 10. 10. 10. 10.]
{code}
After adding the __itruediv__ function, the result is as follows, which means
that the /= operation is in place:
{code:java}
[1. 1. 1. 1. 1. 1. 1. 1. 1. 1.]
{code}
This is because the {color:#33}__idiv__ operation is for python 2, while
__itruediv__ is for python 3. Therefore, if we do not add the __itruediv__
operator in tensor.py, it just uses a default operation which is not in
place.{color}
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
[jira] [Created] (SINGA-473) Implement ONNX Operators in Autograd: Trigonometry functions
YEUNG SAI HO created SINGA-473: -- Summary: Implement ONNX Operators in Autograd: Trigonometry functions Key: SINGA-473 URL: https://issues.apache.org/jira/browse/SINGA-473 Project: Singa Issue Type: New Feature Components: Module Reporter: YEUNG SAI HO The Open Neural Network Exchange (ONNX) is a format for interchanging neural network models between AI systems. There is a list of ONNX operators defined in https://github.com/onnx/onnx/blob/master/docs/Operators.md We are going to implement more ONNX operators in our Autograd module (incubator-singa/python/singa/autograd.py). First of all, this ticket implements 11 unary Trigonometry functions in the Autograd module, which are: 1. cos 2. cosh 3. sin 4. sinh 5. tan 6. acos 7. acosh 8. asin 9. asinh 10. atan 11. atanh -- This message was sent by Atlassian JIRA (v7.6.14#76016)
[jira] [Created] (SINGA-491) Code Cleaning with the Reference of LGTM Analysis Result
YEUNG SAI HO created SINGA-491: -- Summary: Code Cleaning with the Reference of LGTM Analysis Result Key: SINGA-491 URL: https://issues.apache.org/jira/browse/SINGA-491 Project: Singa Issue Type: Improvement Components: Core Reporter: YEUNG SAI HO Code Cleaning with the Reference of LGTM Analysis Result Since LGTM has been applied for our code analysis (see SINGA-484), I added this JIRA ticket so I can also help cleaning the code. At my first glance, the code cleaning according to many issues alerted by LGTM seems to be not difficult for me. -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (SINGA-493) Update news and website
[ https://issues.apache.org/jira/browse/SINGA-493?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] YEUNG SAI HO updated SINGA-493: --- Description: I add a JIRA ticket to update news of SINGA. There are two steps: 1. PR to update the doc/en/index.rst (may modify or add related files) 2. After the above mentioned PR in is merged, follow the steps in http://singa.apache.org/en/develop/contribute-docs.html to submit PR for updating the website was: I add a JIRA ticket to update news of SINGA. There are two steps: 1. PR to update the doc/en/index.rst (may modify or add related files) 2. Follow the steps in http://singa.apache.org/en/develop/contribute-docs.html to submit PR for updating the website > Update news and website > --- > > Key: SINGA-493 > URL: https://issues.apache.org/jira/browse/SINGA-493 > Project: Singa > Issue Type: Improvement > Components: Documentation >Reporter: YEUNG SAI HO >Priority: Major > > I add a JIRA ticket to update news of SINGA. There are two steps: > 1. PR to update the doc/en/index.rst (may modify or add related files) > 2. After the above mentioned PR in is merged, follow the steps in > http://singa.apache.org/en/develop/contribute-docs.html to submit PR for > updating the website -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Updated] (SINGA-493) Update news and website
[ https://issues.apache.org/jira/browse/SINGA-493?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] YEUNG SAI HO updated SINGA-493: --- Summary: Update news and website (was: Updata news and website) > Update news and website > --- > > Key: SINGA-493 > URL: https://issues.apache.org/jira/browse/SINGA-493 > Project: Singa > Issue Type: Improvement > Components: Documentation >Reporter: YEUNG SAI HO >Priority: Major > > I add a JIRA ticket to update news of SINGA. There are two steps: > 1. PR to update the doc/en/index.rst (may modify or add related files) > 2. Follow the steps in > http://singa.apache.org/en/develop/contribute-docs.html to submit PR for > updating the website -- This message was sent by Atlassian Jira (v8.3.4#803005)
[jira] [Created] (SINGA-493) Updata news and website
YEUNG SAI HO created SINGA-493: -- Summary: Updata news and website Key: SINGA-493 URL: https://issues.apache.org/jira/browse/SINGA-493 Project: Singa Issue Type: Improvement Components: Documentation Reporter: YEUNG SAI HO I add a JIRA ticket to update news of SINGA. There are two steps: 1. PR to update the doc/en/index.rst (may modify or add related files) 2. Follow the steps in http://singa.apache.org/en/develop/contribute-docs.html to submit PR for updating the website -- This message was sent by Atlassian Jira (v8.3.4#803005)
