This was extracted (@ 2024-03-20 21:10) from a list of minutes
which have been approved by the Board.
Please Note
The Board typically approves the minutes of the previous meeting at the
beginning of every Board meeting; therefore, the list below does not
normally contain details from the minutes of the most recent Board meeting.
WARNING: these pages may omit some original contents of the minutes.
Meeting times vary, the exact schedule is available to ASF Members and Officers, search for "calendar" in the Foundation's private index page (svn:foundation/private-index.html).
WHEREAS, the Board of Directors deems it no longer in the best interest of the Foundation to continue the Apache MXNet project due to inactivity; NOW, THEREFORE, BE IT RESOLVED, that the Apache MXNet project is hereby terminated; and be it further RESOLVED, that the Attic PMC be and hereby is tasked with oversight over the software developed by the Apache MXNet Project; and be it further RESOLVED, that the office of "Vice President, Apache MXNet" is hereby terminated; and be it further RESOLVED, that the Apache MXNet PMC is hereby terminated. Special Order 7A, Terminate the Apache MXNet Project, was approved by Unanimous Vote of the directors present.
No report was submitted.
## Description: The mission of Apache MXNet is the creation and maintenance of software related to a flexible and efficient library for Deep Learning ## Project Status: Current project status: Considering moving to the Attic Issues for the board: The project's current positioning is not gaining traction in light of the development in the open source deep learning framework space. Code development has mostly halted and community engagement slowed. While roll call of the MXNet PMC received responses from 21 PMC members, the PMC still needs either to find a critical mass to continue drive maintenance, or to find alternative position that MXNet can pivot to so that the community can gain traction in attracting new joiners. The discussion is happening on dev@ ## Membership Data: Apache MXNet was founded 2022-09-20 (9 months ago) There are currently 87 committers and 51 PMC members in this project. The Committer-to-PMC ratio is roughly 8:5. Community changes, past quarter: - No new PMC members. Last addition was Anirudh Subramanian on 2022-09-20. - No new committers were added. ## Project Activity: Software development activity: - 4 pull requests created in the last 3 months. ## Community Health: Community engagement has slowed significantly compared to last year, despite that the GenAI/large language model boom sparked increased general interest in deep learning and scalable systems.
@Justin: Follow up with the PMC Chair
## Description: The mission of Apache MXNet is the creation and maintenance of software related to a flexible and efficient library for Deep Learning ## Issues: - The activity from the community has reduced significantly as the main contributing groups gradually disengaged. - Part of the issue is the value proposition. Given the changes in the competitive space of deep learning, the existing plan no longer appear sufficient to attract enough engagement. To mitigate the situation, the PMC will discuss on potential new positioning of MXNet and find value proposition that makes sense considering the current landscape. ## Membership Data: Apache MXNet was founded 2022-09-20 (6 months ago) There are currently 87 committers and 51 PMC members in this project. The Committer-to-PMC ratio is roughly 8:5. Community changes, past quarter: - No new PMC members. Last addition was Anirudh Subramanian on 2022-09-20. - No new committers were added. ## Project Activity: Maintenance changes for v1.9: - Fix binary building process - Fix a long standing issue that causes segfault at exit and dataloader memory leakage. ## Community Health: The code development has slowed and the community engagement has reduced. 30 commits in the past quarter (-76% change) 5 code contributors in the past quarter (-50% change) 14 PRs opened on GitHub, past quarter (55% increase) 27 PRs closed on GitHub, past quarter (440% increase) 13 issues opened on GitHub, past quarter (44% increase) 6 issues closed on GitHub, past quarter (-25% change)
@Justin: pursue a roll call for MXNet
No report was submitted.
No report was submitted.
No report was submitted.
## Description: A Flexible and Efficient Library for Deep Learning. ## Issues: There are still a few trademark issues that are reoccuring after PMC requested correction. We will attempt to reach out to address them again. ## Membership Data: Apache MXNet was founded 2022-09-20 (19 days ago) There are currently 87 committers and 51 PMC members in this project. The Committer-to-PMC ratio is roughly 8:5. Community changes, past quarter: - No new PMC members (project graduated recently). - No new committers were added. ## Project Activity: Apache MXNet is working on the remaining post-graduation items such as removing incubation reference, updating websites. During the move, we found the current website publishing to tend to timeout due to website structure and size, and the timeout caused down-time. With infra's help, we got the website back online. To address the timeout, infra paused our website refresh and MXNet will work to reduce the website size and improve publishing flow. For project development, Apache MXNet is focusing on the 2.0 release, and have worked on a few performance regression fixes from community feedbacks on the alpha version. We plan to make the official release after addressing the performance issues and bugs. ## Community Health: Apache MXNet community size and contribution activities have been stable. We are getting 193 commits from 15 code contributors in the past quarter focusing on finalizing the 2.0 release.
## Description: A Flexible and Efficient Library for Deep Learning. ## Issues: There are still a few trademark issues that are reoccuring after PMC requested correction. We will attempt to reach out to address them again. ## Membership Data: Apache MXNet was founded 2022-09-20 (19 days ago) There are currently 87 committers and 51 PMC members in this project. The Committer-to-PMC ratio is roughly 8:5. Community changes, past quarter: - No new PMC members (project graduated recently). - No new committers were added. ## Project Activity: As the newly minted TLP, Apache MXNet is working on the remaining post-graduation items such as removing incubation reference, updating websites. For project development, Apache MXNet is focusing on the 2.0 release, and have worked on a few performance regression fixes from community feedbacks on the alpha version. We plan to make the official release after addressing the performance issues and bugs. ## Community Health: Apache MXNet community size and contribution activities have been stable. We are getting 193 commits from 15 code contributors in the past quarter focusing on finalizing the 2.0 release.
WHEREAS, the Board of Directors deems it to be in the best interests of the Foundation and consistent with the Foundation's purpose to establish a Project Management Committee charged with the creation and maintenance of open-source software, for distribution at no charge to the public, related to a flexible and efficient library for Deep Learning. NOW, THEREFORE, BE IT RESOLVED, that a Project Management Committee (PMC), to be known as the "Apache MXNet Project", be and hereby is established pursuant to Bylaws of the Foundation; and be it further RESOLVED, that the Apache MXNet be and hereby is responsible for the creation and maintenance of software related to a flexible and efficient library for Deep Learning; and be it further RESOLVED, that the office of "Vice President, Apache MXNet" be and hereby is created, the person holding such office to serve at the direction of the Board of Directors as the chair of the Apache MXNet Project, and to have primary responsibility for management of the projects within the scope of responsibility of the Apache MXNet Project; and be it further RESOLVED, that the persons listed immediately below be and hereby are appointed to serve as the initial members of the Apache MXNet Project: * Anirudh Subramanian <anirudh2290@apache.org> * Bing Xu <bingxu@apache.org> * Bob Paulin <bob@apache.org> * Carin Meier <cmeier@apache.org> * Chiyuan Zhang <pluskid@apache.org> * Chris Olivier <cjolivier01@apache.org> * Dick Carter <dickjc123@apache.org> * Eric Xie <jxie@apache.org> * Furkan Kamaci <kamaci@apache.org> * Haibin Lin <haibin@apache.org> * Henri Yandell <bayard@apache.org> * Hongliang Liu <hliu@apache.org> * Indhu Bharathi <indhub@apache.org> * Jackie Wu <wkcn@apache.org> * Jason Dai <jasondai@apache.org> * Jian Zhang <jianzhangzju@apache.org> * Joe Evans <jevans@apache.org> * Joe Spisak <jspisak@apache.org> * Jun Wu <reminisce@apache.org> * Leonard Lausen <lausen@apache.org> * Liang Depeng <ldpe2g@apache.org> * Ly Nguyen <lxn2@apache.org> * Madan Jampani <madjam@apache.org> * Marco de Abreu <marcoabreu@apache.org> * Markus Weimer <weimer@apache.org> * Mu Li <muli@apache.org> * Nan Zhu <codingcat@apache.org> * Naveen Swamy <nswamy@apache.org> * Przemysław Trędak <ptrendx@apache.org> * Qiang Kou <qkou@apache.org> * Qing Lan <lanking@apache.org> * Sandeep Krishnamurthy <skm@apache.org> * Sergey Kolychev <sergeykolychev@apache.org> * Sheng Zha <zhasheng@apache.org> * Shiwen Hu <yajiedesign@apache.org> * Tao Lv <taolv@apache.org> * Terry Chen <terrychen@apache.org> * Thomas Delteil <thomasdelteil@apache.org> * Tianqi Chen <tqchen@apache.org> * Tong He <the@apache.org> * Tsuyoshi Ozawa <ozawa@apache.org> * Xingjian Shi <sxjscience@apache.org> * YiZhi Liu <liuyizhi@apache.org> * Yifeng Geng <gengyifeng@apache.org> * Yu Zhang <yzhang87@apache.org> * Yuan Tang <terrytangyuan@apache.org> * Yutian Li <hotpxl@apache.org> * Zhi Zhang <zhreshold@apache.org> * Zihao Zheng <zihaolucky@apache.org> * Ziheng Jiang <ziheng@apache.org> * Ziyue Huang <ziyue@apache.org> NOW, THEREFORE, BE IT FURTHER RESOLVED, that Sheng Zha be appointed to the office of Vice President, Apache MXNet, to serve in accordance with and subject to the direction of the Board of Directors and the Bylaws of the Foundation until death, resignation, retirement, removal or disqualification, or until a successor is appointed. Special Order 7F, Establish the Apache MXNet Project, was approved by Unanimous Vote of the directors present.
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Complete general@incubator discussion
### Are there any issues that the IPMC or ASF Board need to be aware of?
We are currently
[discussing](https://lists.apache.org/thread/3rxzjcmo2y457y6r8ohz1j4qv49joyo
6)
graduation in the general@incubator list after the community discussed
and voted.
### How has the community developed since the last report?
* The number of GitHub contributors is +1, currently 871
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2.1k followers (+0%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.58k
subscribers (+0.6% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 7.26k
subscribers (+0.4%)
### How has the project developed since the last report?
1. Apache MXNet (incubating) 1.9.1 released.
2. Fixed some issues brought up in the
[graduation discussion
thread](https://lists.apache.org/thread/3rxzjcmo2y457y6r8ohz1j4qv49joyo6)
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2022-05-27
### When were the last committers or PPMC members elected?
2022-04-25
### Have your mentors been helpful and responsive?
Yes.
### Is the PPMC managing the podling's brand / trademarks?
Yes, discussed some potential violations with trademarks@ but was decided
no action to take.
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [ ] (mxnet) Furkan Kamaci
Comments:
- [X] (mxnet) Zhenxu Ke
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Complete general@incubator discussion
### Are there any issues that the IPMC or ASF Board need to be aware of?
We are currently
[discussing](https://lists.apache.org/thread/3rxzjcmo2y457y6r8ohz1j4qv49joyo
6) graduation in the general@incubator list after the community discussed
and voted.
### How has the community developed since the last report?
* The number of GitHub contributors is the same, currently 870
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2.1k followers (+0%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.57k
subscribers (+0.6% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 7.23k
subscribers (+0.3%)
### How has the project developed since the last report?
1. Community
[discussed](https://lists.apache.org/thread/l9h2qgb2vqs2y0cm46wh83sgomr9w190
) and
[voted](https://lists.apache.org/thread/py1nw6whov78sch5kkm1gcg7cv3zb2sv)
on graduating to a Apache Top Level Project
2. Apache MXNet (incubating) 1.9.1 RC0 created, currently undergoing dev
community vote to release.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2022-03-21
### When were the last committers or PPMC members elected?
2022-04-25
### Have your mentors been helpful and responsive?
Yes.
### Is the PPMC managing the podling's brand / trademarks?
Yes. From past brand usage review for MXNet third-party distributions, we
found several listings on AWS marketplace that needed update. The PPMC
reached out to the publishers of these listings for correction and fixed
them.
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [X] (mxnet) Furkan Kamaci
Comments:
- [X] (mxnet) Zhenxu Ke
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Successfully and smoothly make releases without WIP disclaimer.
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
No blocking issue.
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 967
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2k followers (+1%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.49k
subscriber (+2% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.96k
subscribers (+3%)
* Highlights in MXNet ecosystem
* GluonCV v0.10.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.10.0)
* GluonNLP MX2 NumPy version
(https://github.com/dmlc/gluon-nlp/tree/master)
* GluonTS v0.8.1 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.8.1)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye 2.3.24 release (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.9.0 release is in progress.
1.9.0 release is going through rc8 after a thorough review on the licenses
from the community and PPMC.
https://github.com/apache/incubator-mxnet/releases/tag/1.9.0.rc8
with 100+ patches of new features, improvements, and fixes.
This will be the first recent release without DISCLAIMER-WIP.
2) 2.0.0 development is close to completion.
MXNet 2.0 features interoperable ML and DL programming with Array API
standard implementation. This project aims to make array libraries like
NumPy, MXNet, Pytorch more interoperable. Besides standardized operators,
new methods and mechanism will be introduced to improve the
interoperability. 1) in MXNet2.0, Context class and some related array
methods like as_in_ctx will be replaced by Device and to_device to reduce
the learning curve for new users. 2) The improved DLPack API in MXNet
leverages the syncobject introduced in Async GPU dependency engine and can
well handle the cuda stream to improve the array libraries
interoperability
on different devices.
3) Github statistics of last month:
* October 6, 2021 – November 6, 2021: Excluding merges, 17 authors
have
pushed 49 commits to master and 58 commits to all branches. On master, 466
files have changed and there have been 17,398 additions and 11,371
deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2021-03-24
### When were the last committers or PPMC members elected?
2021-11-03
### Have your mentors been helpful and responsive?
Yes.
### Is the PPMC managing the podling's brand / trademarks?
Yes. From past brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that needed update. The PPMC
reached out to the publishers of these listings for correction and fixed
them.
In recent review, we found the following three entries to have regression
in branding and have reached out to them again.
https://aws.amazon.com/marketplace/pp/prodview-7nxhayhlbxwam (pending
update)
https://aws.amazon.com/marketplace/pp/prodview-wktuc2vochjwe (pending
update)
https://aws.amazon.com/marketplace/pp/prodview-dpniaffpcxfhc (pending
update)
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [ ] (mxnet) Furkan Kamaci
Comments:
- [ ] (mxnet) Kezhen Xu
Comments:
- [ ] (mxnet) Atri Sharma
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Successfully and smoothly make releases without WIP disclaimer.
- ONGOING.
2. Improve brand management.
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
Since last report, we recruited two new mentors for MXNet and they have
been very helpful in MXNet activities.
We conduct regular brand management review as part of the regular
reporting.
From recent brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that need update. The PPMC
reached out to the publishers of these listings for correction with only
last three remaining.
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 963
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2k followers (+1%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.46k
subscriber (+5% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.75k
subscribers (+3%)
* Highlights in MXNet ecosystem
* GluonCV v0.10.4 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.10.4)
* GluonNLP MX2 NumPy version
(https://github.com/dmlc/gluon-nlp/tree/master)
* GluonTS v0.8.0 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.8.0)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye 2.3.17 release (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.9.0 release is in progress.
https://github.com/apache/incubator-mxnet/releases/tag/1.9.0.rc6
with 100+ patches of new features, improvements, and fixes.
This will be the first recent release without DISCLAIMER-WIP.
2) Github statistics of last month:
* July 4, 2021 – August 4, 2021: Excluding merges, 14 authors have
pushed 14 commits to master and 29 commits to all branches. On master,
66 files have changed and there have been
12,224 additions and 5,745 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2021-03-24
### When were the last committers or PPMC members elected?
2021-07-16
### Have your mentors been helpful and responsive?
Yes.
### Is the PPMC managing the podling's brand / trademarks?
Yes. From recent brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that need update. The PPMC
reached out to the publishers of these listings for correction.
Here we report the status of these items.
https://aws.amazon.com/marketplace/pp/B073SHB43M?qid=1609989286506&sr=0-21&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/pp/B079225XXC?qid=1609989161134&sr=0-14&r
ef_=srh_res_product_title (pending update)
https://aws.amazon.com/marketplace/pp/B08L8H9NWD?qid=1609989286506&sr=0-28&r
ef_=srh_res_product_title (pending update)
https://aws.amazon.com/marketplace/pp/B08G8VXC1Q?qid=1609989286506&sr=0-29&r
ef_=srh_res_product_title (pending update)
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [X] (mxnet) Furkan Kamaci
Comments:
- [ ] (mxnet) Kezhen Xu
Comments:
- [ ] (mxnet) Atri Sharma
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address release issues, improve automation for license checks, and
make it easier for auditing.
- DONE
2. Successfully and smoothly make releases without WIP disclaimer.
- ONGOING.
3. Improve brand management.
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. The community fixed the issues in INCUBATOR-253 and will continue to
perform regular trademark and branding review.
2. Following the completion of license and release issue fixes, MXNet had
difficulty finishing the
releases votes in Incubator. Here are the stats for the two recent
successful releases:
| | 1.8.0 | 2.0.0.alpha |
|----------------------------------------|---------|-------------|
| duration | 39 days | 27 days |
| issues uncovered (fixed after release) | 5 | 2 |
| mentor participation | 50% | 25% |
We have one mentor who indicated the desire to step down if we can find
more mentors.
We will request for a few more active mentors to help us smoothly finish
the last mile before graduation.
3. We conduct regular brand management review as part of the regular
reporting.
From recent brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that need update. The PPMC
reached out to the publishers of these listings for correction. See status
in brand management section.
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 955
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2k followers (+3%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.39k
subscriber (+4.5% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.55k
subscribers (+1.8%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
28k subscriber (+7.6% since last report)
* Highlights in MXNet ecosystem
* Dive into Deep Learning has 67K 30-day active users, and has
attracted 31.3K stars & 350 contributors on GitHub. It has been adopted as
a textbook or reference book by 180+ universities from 40 countries, such
as Stanford, MIT, UC Berkeley, CMU, UCambridge.
* GluonCV v0.10.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.10.0)
* GluonNLP MX2 NumPy version
(https://github.com/dmlc/gluon-nlp/tree/master)
* GluonTS v0.6.7 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.6.7)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye 2.3.14 release (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.8.0 is released:
https://github.com/apache/incubator-mxnet/releases/tag/1.8.0
with 100+ patches of new features, improvements, and fixes.
2) 2.0.0 alpha is released:
https://github.com/apache/incubator-mxnet/projects/18
https://github.com/apache/incubator-mxnet/releases/tag/2.0.0.alpha
3) Github statistics of last month:
* March 7, 2021 – April 7, 2021: Excluding merges, 18 authors have
pushed 33 commits to master and 67 commits to all branches. On master, 10
files have changed and there have been 206 additions and 71 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2021-03-24
### When were the last committers or PPMC members elected?
2021-02-03
### Have your mentors been helpful and responsive?
Yes. Furkan has helped with all our recent releases (thanks!).
Markus expressed desire to step down as a mentor if we find more
mentors. Because of the large scope of the project and the several issues
we faced so far, we definitely wish to have more mentors to help.
### Is the PPMC managing the podling's brand / trademarks?
Yes. From recent brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that need update. The PPMC
reached out to the publishers of these listings for correction.
Here we report the status of these items.
https://aws.amazon.com/marketplace/pp/B07MP6Y8XT?qid=1609989161134&sr=0-17&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/pp/B07C49CVC1?qid=1609989286506&sr=0-30&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/pp/B01JJ31R8C?qid=1609989286506&sr=0-22&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/pp/B07F3YBMT9?qid=1609989286506&sr=0-23&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/pp/B084FXK9XH?qid=1609989286506&sr=0-24&r
ef_=srh_res_product_title (done)
https://aws.amazon.com/marketplace/search/results?x=0&y=0&searchTerms=%22MX
(pending)
https://aws.amazon.com/marketplace/pp/B073SHB43M?qid=1609989286506&sr=0-21&r
ef_=srh_res_product_title (pending update)
https://aws.amazon.com/marketplace/pp/B079225XXC?qid=1609989161134&sr=0-14&r
ef_=srh_res_product_title (pending update)
https://aws.amazon.com/marketplace/pp/B08L8H9NWD?qid=1609989286506&sr=0-28&r
ef_=srh_res_product_title (pending update)
https://aws.amazon.com/marketplace/pp/B08G8VXC1Q?qid=1609989286506&sr=0-29&r
ef_=srh_res_product_title (pending update)
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [X] (mxnet) Furkan Kamaci
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address licensing and trademark issues for the binary releases in the
community.
- ONGOING. Close to completion. See update in the next section.
2. Address release issues, improve automation for license checks, and
make it easier for auditing. Successfully and smoothly make releases
without WIP disclaimer.
- ONGOING.
3. Improve brand management.
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. The community is fixing the remaining releases that have license
issues. See summary below.
2. MXNet 2.0 first public beta, interoperable with NumPy, is on the way:
https://github.com/apache/incubator-mxnet/projects/18
RFC: https://github.com/apache/incubator-mxnet/issues/16167
3. The community is including brand usage reviews as part of the
quarterly report process to proactively manage our brand going forward.
#### Status on license and branding issues
Most of the issues from last update hinged on the resolution of LEGAL-516.
We are very glad that LEGAL-516 concluded that ASF projects compiled using
NVIDIA’s CUDA compiler may be distributed under the Apache License 2.0,
which unblocks many of the distribution practices that are essential to
the
wide adoption of software that utilizes GPU. We would like to thank for
the
help, the patience, and the leniency and flexibility that the incubator
and
legal granted us, and the numerous helps we received along the way from
the
ASF members and our community members from NVIDIA. Especially, I would
like
to thank Michael O’Conner and Triston Cao from NVIDIA for pushing this
through within their organization.
Based on the conclusion of LEGAL-516, binary distribution of CUDA-compiled
Apache projects can be properly licensed as ALv2. In addition, the
branding
issues from distribution pages from Amazon, NVIDIA, and Intel have all
been
acted on. As a result, among the 9 pending issues from last update, 5 can
be resolved. We are following up on the remaining items to resolve all
issues in INCUBATOR-253.
As part of our improvement in brand management, we are including
third-party brand usage review in our quarterly reports and we will keep
it
as part of our regular practice beyond graduation.
Status on open issues since last update:
1. Source and convenance binary releases containing Category X licensed
code. (pending item 5) * Source code releases by the PPMC do not contain
Category X code.
Takedown and backfill of compliant binary releases by
the
PPMC on repository.apache.org is in progress, see item 5.
PyPI releases
are
made by third-party. See item 8.
5. Maven releases containing Category X licensed code. (pending) *
Takedown
of binary releases on repository.apache.org initiated (i.e. INFRA-20442).
Based on the resolution of LEGAL-516, the takedown (and backfill) includes
those that include GPL licensed components. We are working on this in the
infra issue.
6. PyPI releases containing Category X licensed code. (Resolved) * The
third-party releases are compliant with branding guidelines since the
releases are from official source code releases and are properly licensed
with ALv2.
7. Docker releases containing Category X licensed code. (Resolved) * The
third-party releases are compliant with branding guidelines since the
releases are from official source code releases and are properly licensed
with ALv2.
9. Trademark and branding issues with PyPI and Docker releases. (Resolved)
* These release are compliant with trademark and branding requirements
since they don’t contain Category X licensed code and are licensed with
ALv2
based on item 6 and 7.
10. Trademark and brand issues with naming of releases. (Resolved) * There
are no binary releases by the PPMC besides the repository.apache.org
releases in item 5.
14, 22, 23. Branding and release of 3rd parties containing unreleased
code.
Known pages with issues: *
+https://docs.nvidia.com/deeplearning/frameworks/mxnet-release-notes/rel_20-
03.html+ (item 14, Resolved)
NVIDIA switched to "NVIDIA Optimized Deep
Learning Framework, powered by Apache MXNet” for naming.
+https://aws.amazon.com/marketplace/pp/B07YW8HVLD?qid=1595741035764&sr=0-4&r
ef_=srh_res_product_title+ (item 22, pending) PPMC reached out to Bitnami
directly regarding this listing. Bitnami responded but the page still
needs update. We followed up again with Bitnami this week.
+https://aws.amazon.com/marketplace/search/results?x=0&y=0&searchTerms=%22MX
Net+ (item 23, pending) PPMC reached out to AWS through internal channel
to
fix branding issue. Amazon intends to use AWS MX powered by Apache MXNet
naming convention. The naming change is in progress.
After last update, for item 15, SourceForge further added a disclaimer
that
the page is an automatic mirror of MXNet’s GitHub tags:
+https://sourceforge.net/projects/apache-mxnet.mirror/+.
In addition, MXNet PPMC identified several more listings that require
correction according to the branding guideline on AWS marketplace that are
related to AWS Deep Learning Container/DLAMI, and they have all been
resolved.
From recent brand usage review for MXNet third-party distributions,
we found several listings on AWS marketplace that need update. The PPMC
is reaching out to the publishers of these listings for correction.
https://aws.amazon.com/marketplace/pp/B079225XXC?qid=1609989161134&sr=0-14&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B07MP6Y8XT?qid=1609989161134&sr=0-17&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B073SHB43M?qid=1609989286506&sr=0-21&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B01JJ31R8C?qid=1609989286506&sr=0-22&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B07F3YBMT9?qid=1609989286506&sr=0-23&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B084FXK9XH?qid=1609989286506&sr=0-24&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B08L8H9NWD?qid=1609989286506&sr=0-28&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B08G8VXC1Q?qid=1609989286506&sr=0-29&r
ef_=srh_res_product_title
https://aws.amazon.com/marketplace/pp/B07C49CVC1?qid=1609989286506&sr=0-30&r
ef_=srh_res_product_title
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 948
(2019-10-08; +13% since last report)
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 2k followers (+3%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.33k
subscriber (+4.7% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.43k
subscribers (+2.3%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
26k subscriber (+4% since last report)
* Highlights in MXNet ecosystem
* Dive into Deep Learning has 50K 28-day active users, and has
attracted 28.0K stars & 333 contributors on GitHub. It has been adopted as
a textbook or reference book by 170+ universities from 40 countries, such
as Stanford, MIT, UC Berkeley, CMU, UCambridge.
* GluonCV v0.9.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.9.0)
* GluonNLP MX2 NumPy version
(https://github.com/dmlc/gluon-nlp/tree/master)
* GluonTS v0.6.4 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.6.4)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye 2.3.2 release (https://github.com/awslabs/sockeye)
* MXNet community held the "Apache MXNet Day" community meet-up on 12/14
with NVIDIA and AWS as sponsors and Apache as community sponsor. At the
meet-up, community members shared the latest progress in MXNet, and
discussed topics such as array API standardization across frameworks,
and the history and future of MXNet.
### How has the project developed since the last report?
1) 1.8.0 release is in progress:
https://github.com/apache/incubator-mxnet/releases/tag/1.8.0.rc2
with 100+ patches of new features, improvements, and fixes.
2) 2.0 alpha release:
https://github.com/apache/incubator-mxnet/projects/18
3) Github statistics of last month:
* Dec 5, 2020 – Jan 5, 2021: Excluding merges, 28 authors have pushed
31 commits to master and 60 commits to all branches. On master, 175 files
have changed and there have been 7,192 additions and 1,628 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2020-08-24
### When were the last committers or PPMC members elected?
2020-11-30
### Have your mentors been helpful and responsive?
Yes. Michael stepped down as a mentor and we are grateful for his helps
for our project.
Markus also expressed desire to step down as a mentor if we find more
mentors. Because of the large scope of the project and the several issues
we faced so far, we definitely wish to have more mentors to help.
### Is the PPMC managing the podling's brand / trademarks?
Yes. The PPMC conducted more extensive branding and trademarks review,
identified some violations and acted on resolving them with offenders, as
described in "Status on license and branding issues" section. In addition,
The community is including brand usage reviews as part of the quarterly
report process to proactively manage our brand going forward.
### Signed-off-by:
- [ ] (mxnet) Markus Weimer
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
- [X] (mxnet) Furkan Kamaci
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address licensing and trademark issues for the binary releases in the
community.
- ONGOING. See update in the next section.
2. Address release issues. Successfully and smoothly make releases
without WIP disclaimer.
- ONGOING.
3. Improve development process and tooling to help reduce the overhead of
releases
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. The community continues to work towards resolving license and branding
issues. See summary below.
2. MXNet 2.0 first public beta, interoperable with NumPy, is on the way:
https://github.com/apache/incubator-mxnet/projects/18
RFC: https://github.com/apache/incubator-mxnet/issues/16167
3. We are adopting Github Discussions
https://github.com/apache/mxnet/discussions in favor of Discourse forum
for closer connection between users and developers.
https://discuss.mxnet.apache.org/ (previously discuss.mxnet.io) will serve
as an archive for past user discussions.
#### Status on license and branding issues
The PPMC continues to make progress in resolving license and branding
issues. During the a recent release vote on 1.7.0, with the help from
Justin, a few more source distribution and branding issues were found.
Here is the status on the issues, tracked in INCUBATOR-253. As of now,
the PPMC resolved 15 issues, with 9 still left.
Most of the outstanding issues hinge on one key outstanding question,
which is whether the components in MXNet that are produced by CUDA NVCC
constitutes Category X code. PPMC member employed by Nvidia helped connect
Nvidia's representatives including Michael O'Connor, Director of Deep
Learning, who have been supportive in the efforts of clarification.
Progress is tracked in LEGAL-516. At the moment, NVIDIA's legal is
actively exploring options on resolving this issue. Michael requested
extension with MXNet PPMC to not treat the related component as Category-X
and delay actions until Oct. board meeting.
Also, we discoveredd that there was confusion around MXNet's status around
integration with Intel products. In MXNet, there has never been public
distribution of MXNet where MKL was included. There was an initial inquiry
around whether MKL builds can be enabled but was rejected, so this never
came into practice. Also, to my knowledge Intel has not produced custom
MXNet builds with closed-source components.
MXNet 1.7.0 release has completed.
So far, PPMC members from Intel (Tao), Nvidia (Dick), and Amazon (Leonard,
Henri, Qing, Sheng) have acted to help resolve the issues.
Status on open issues since last update:
1. Source and convenance binary releases containing Category X licensed
code. (pending)
Source code releases by the PPMC do not contain Category X code, no issue.
Takedown of binary releases by the PPMC on repository.apache.org is on
hold, see item 5.
PyPI releases are made by third-party. See item 8.
5. Maven releases containing Category X licensed code. (pending)
Takedown of binary releases on repository.apache.org initiated [1]. The
scope depends on the resolution of LEGAL-516 [2].
6. PyPI releases containing Category X licensed code. (pending)
There are no official PyPI releases. Whether the third-party releases are
compliant with branding guidelines depend on the resolution of LEGAL-516.
7. Docker releases containing Category X licensed code. (pending)
There are no official Docker releases. Whether the third-party releases
are
compliant with branding guidelines depend on the resolution of LEGAL-516.
9. Trademark and branding issues with PyPI and Docker releases. (pending)
There are no official PyPI or Docker releases. In addition, as they all
contain binary from unmodified MXNet code, whether they are compliant in
branding now solely depends on whether they contain Category X licensed
code. Refer to item 6, 7.
10. Trademark and brand issues with naming of releases. (pending)
There are no binary releases by the PPMC besides the repository.apache.org
releases in item 5.
12. Releases and other nightly builds on https://repo.mxnet.io /
https://dist.mxnet.io containing Category X licensed code (resolved)
Neither of the two site contains releases. The binaries are only intended
for testing pipelines and are made available only to MXNet developers. As
part of the effort to resolve branding concern, public references to these
sites are deleted [4].
14. to 23. Branding and release of 3rd parties containing unreleased code.
(pending)
Known pages with issues:
https://docs.nvidia.com/deeplearning/frameworks/mxnet-release-notes/rel_20-0
3.html (item 14, pending)
PPMC reached out to Nvidia. Pending action from Nvidia on branding
compliance and replying on whether unreleased code was included.
https://sourceforge.net/projects/apache-mxnet.mirror/ (item 15, resolved)
PPMC reached out to SourceForge. SourceForge added (incubating) in name.
As 1.7.0 release has completed, we removed release candidate tag from GitHub
and the the mirror automatically deleted the it [3]. As such, the original
issue is resolved. The PPMC will follow up with SourceForge on the
functionality of filtering out non-release tags.
AWS Marketplace related links
https://aws.amazon.com/marketplace/pp/B07YW8HVLD?qid=1595741035764&sr=0-4&re
f_=srh_res_product_title (item 22, pending)
PPMC reached out to Bitnami directly regarding this listing. Bitnami
responded and is working on the branding issues.
https://aws.amazon.com/marketplace/search/results?x=0&y=0&searchTerms=%22MXN
et (item 23, pending)
PPMC reached out to AWS through internal channel to fix branding issue.
Will update in the next few days.
In addition, MXNet PPMC identified several more listings that require
correction according to the branding guideline on AWS marketplace that are
related to AWS Deep Learning Container/DLAMI [5], and are actively working
on addressing them.
Reference links
[1]: https://issues.apache.org/jira/browse/INFRA-20442
[2]: https://issues.apache.org/jira/browse/LEGAL-516
[3]: https://sourceforge.net/projects/apache-mxnet.mirror/
[4]:
https://issues.apache.org/jira/browse/LEGAL-515?focusedCommentId=17174133&pa
ge=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-
17174133
[5]:
https://lists.apache.org/thread.html/r3f95f5766e8f1ee8fbd8183720804afdc1678f
0149f56144da
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 836
(2019-10-08; +7.0% since last report)
* Active discussions on user forums
* https://discuss.gluon.ai/ (Chinese, 8.5K registered users (+2.2%)
and 22.4K posts(+1.0% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.94k followers (+7.8%)
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.7k followers
(+2.7%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.27k
subscriber (+7.6% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.28k
subscribers (+1.4%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
25k subscriber (+4.1% since last report)
* Highlights in MXNet ecosystem
* Dive into Deep Learning has 74K 28-day active users, and has
attracted 26.5K stars & 322 contributors on GitHub. It has been
adopted as a textbook or reference book by 100+ universities from 30
countries, such as Stanford, MIT, UC Berkeley, CMU, UCambridge.
* GluonCV v0.8.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.8.0)
* GluonNLP v0.10.0 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.10.0)
* GluonTS v0.5.2 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.5.2)
* Multi-Model Server v1.1.2 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.1.2)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye 2.2.0 release (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.7.0 was released:
https://github.com/apache/incubator-mxnet/releases/tag/1.7.0
with over 492+ patches of new features, improvements, and fixes.
2) 2.0 project: https://github.com/apache/incubator-mxnet/projects/18
3) Github statistics of last month:
* Sept 7, 2020 – Oct 7, 2020: Excluding merges, 56 authors have
pushed 66 commits to master and 118 commits to all branches.
On master, 136 files have changed and there have been 4,393
additions and 1,154 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2020-08-24
### When were the last committers or PPMC members elected?
2020-09-17
### Have your mentors been helpful and responsive?
Yes.
### Is the PPMC managing the podling's brand / trademarks?
Yes. The PPMC conducted more extensive branding and trademarks review,
identified some violations and acted on resolving them with offenders, as
described in "Status on license and branding issues" section.
### Signed-off-by:
- [X] (mxnet) Markus Weimer
Comments: Thanks for the detailed report. The licensing issues are a
tough nutt to crack, but the work will benefit other projects and OSS
in general.
- [X] (mxnet) Michael Wall
Comments: Good report, keep pushing on the licensing issues.
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address licensing and trademark issues for the binary releases in the
community.
- ONGOING. See update in the next section.
2. Address release issues. Successfully and smoothly make releases
without WIP disclaimer.
- ONGOING.
3. Improve development process and tooling to help reduce the overhead of
releases
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. Binary Distribution Licensing Issue
2. MXNet 2.0 first public beta, interoperable with NumPy, is on the way:
https://github.com/apache/incubator-mxnet/projects/18
RFC: https://github.com/apache/incubator-mxnet/issues/16167
#### Issues with releases and distributions
##### Background
In May 2020 The MXNet PPMC has proactively initiated a ASF policy
compliance
review [1] and a license review [2] with the Apache Legal team.
The license review uncovered that
- Building unmodified MXNet release source code with the optional NVidia
GPU
support enabled results in a binary subject to restrictions of NVidia
EULA.
- PPMC members and committers uploaded convenience releases to
repository.apache.org which contain Category-X components. Both GPL and
NVidia EULA components were found.
The policy review uncovered that:
- Prior ASF guidance to the PPMC (December 2018 legal review [3]) was
incomplete
and did not include a reference to the "unwritten" rule that convenience
binary distributions created by third-parties using ASF Trademarks must
not
include Category-X components. Based on this discovery, the Draft
Downstream
Distribution Branding Policy was updated in June 2020 to include the
"unwritten" requirement. Based on the updated guidance, PPMC discovered
various third-party trademark infringements.
The policy review did not yet conclude on the questions if
- The PPMC may create nightly development builds (audience restricted to
dev
list subscribers as per Release policy [4]) for the purpose of testing and
developing MXNet;
##### List of issues and their status
Justin classified the issues into 14 items.
1) Source and convenance binary releases containing Category X licensed
code.
See summary from license review in Background section. Source code
releases do
not contain Category X code; Takedown of binary releases on
repository.apache.org is pending on Apache Infra. (Trademark infringements
of 3rd-parties such as on pypi are discussed separately)
2. Website giving access to downloads of non released/unapproved code.
Website contained links to nightly development builds which have been
removed [5];
Going forward the PPMC intends to begin periodical voting on Alpha and
Beta Releases which will then be linked from the website.
3. Website giving access to releases containing Category X licensed code.
Website contained links to third-party distributions incorporating
Category-X
components (see summary from license review above). Disclaimers were added
to the website clarifying the third-party status of the releases and their
licenses. [5]
4. Web site doesn't given enough warning to users of the issues with non
(P)PMC releases or making it clear that these are not ASF releases.
Website contained links to third-party distributions incorporating
Category-X
components (see summary from license review above). Disclaimers were added
to the website clarifying the third-party status of the releases and their
licenses. [5]
5. Maven releases containing Category X licensed code.
See summary from license review in Background section. Source code
releases do
not contain Category X code; Takedown of binary releases on
repository.apache.org is pending on Apache Infra. [6] (Trademark
infringements of 3rd-parties are discussed separately)
6. PyPI releases containing Category X licensed code.
There are no PiPy releases by the PPMC. Please refer to the trademark
infringement section of the report.
7. Docker releases containing Category X licensed code.
There are no Docker releases by the PPMC. Please refer to the trademark
infringement section of the report.
8. Docker releases containing unreleased/unapproved code.
There are no Docker releases by the PPMC. The existence of third-party
releases
containing unreleased code was approved in [3] and is also in line with
the current Downstream Distribution Branding Draft Policy. ("using any
particular revision from the development branch is OK" [3])
9. Trademark and branding issues with PiPy and Docker releases.
There are no PiPy releases by the PPMC. Please refer to the trademark
infringement section of the report.
10. Trademark and brand issues with naming of releases.
There are no binary releases by the PPMC besides the repository.apache.org
releases discussed above, which are being removed.
Please refer to the trademark infringement section of
the report.
11. Developer releases available to users and public searchable
https://repo.mxnet.io / https://dist.mxnet.io
Links to the nightly development builds were removed from the MXNet
website and
a robot.txt file was added to prevent indexing of the sites. These
websites are removed from Google search index.
12. Releases and other nightly builds on
https://repo.mxnet.io / https://dist.mxnet.io containing category X
licensed code.
Neither of the two site contains Releases. It is an open question of the
policy
review (see Background section above) if nightly development builds may or
may not contain Category X components.
13. Lack of clarity on all platforms for what is an ASF release and what
is not.
https://github.com/apache/incubator-mxnet/releases?after=1.2.0 previously
did not distinguish MXNet releases prior to MXNet joining the Incubator.
Disclaimers
were added. Other PPMC platforms do not contain references to non-ASF
releases
(MXNet releases made prior to MXNet joining the ASF). The PPMC is aware of
old third-party releases created prior to MXNet joining the ASF which are
still
available, but can be clearly separated from the ASF MXNet releases due to
the lack of reference to the Apache foundation. PPMC was able to find an
exemplar
such release at [7]. If there are concerns from the Incubator, PPMC can
request
the third-parties to take down these releases, as editing their
Description to include
references to events (MXNet joining Apache) is not
supported due to immutability constraints. [8]
14. Branding and release of 3rd parties containing unreleased code.
(e.g.
https://docs.nvidia.com/deeplearning/frameworks/mxnet-release-notes/rel_20-0
3.html)
Please refer to the trademark infringement section of the report.
[1]: https://issues.apache.org/jira/browse/LEGAL-515
[2]: https://issues.apache.org/jira/browse/LEGAL-516
[3]: https://s.apache.org/flvug
[4]: http://www.apache.org/legal/release-policy.html#publication
[5]:
https://github.com/apache/incubator-mxnet/commit/b6b40878f0aba2ba5509f3f3a4c
d517a654847ce#diff-19bc831c1dab6d92d2efc3b87ec5c740
[6]: https://issues.apache.org/jira/browse/INFRA-20442
[7]: https://pypi.org/project/mxnet/0.9.5/
[8]:
https://mail.python.org/pipermail/distutils-sig/2017-December/031826.html
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 781
(2019-03-28; +2.4% since last report)
* Active discussions on user forums
* https://discuss.mxnet.io/ (English, 2.8K registered users (+10.5%)
and 8.1K posts (+5.2%))
* https://discuss.gluon.ai/ (Chinese, 8.3K registered users (+2.5%)
and 22.2K posts(+1.3% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.8k followers
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.6k followers
(+8.3%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.18k
subscriber (+13.5% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 6.05k
subscribers (+6.0%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
24k subscriber (+9.0% since last report)
* Highlights in MXNet ecosystem
* Dive into Deep Learning has 73K 28-day active users, and has
attracted 23.9K stars & 295 contributors on GitHub. It has been
adopted as a textbook or reference book by 100+ universities from 27
countries, such as Stanford, MIT, UC Berkeley, CMU.
* GluonCV v0.7.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.7.0)
* GluonNLP v0.9.1 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.9.1)
* GluonTS v0.5.0 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.5.0)
* Multi-Model Server v1.1.1 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.1.1)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.6.0 was released:
https://github.com/apache/incubator-mxnet/releases/tag/1.6.0
with over 830+ patches of new features, improvements, and fixes.
2) 2.0 project: https://github.com/apache/incubator-mxnet/projects/18
3) Github statistics of last month:
* May 28, 2020 – June 28, 2020: Excluding merges, 59 authors have
pushed 79 commits to master and 107 commits to all branches.
On master, 2,010 files have changed and there have been 10,897
additions and 274,406 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2020-02-20
### When were the last committers or PPMC members elected?
2020-03-02
### Have your mentors been helpful and responsive?
Yes. In particular, Bob has provided guidance on the license and
trademark issue.
### Is the PPMC managing the podling's brand / trademarks?
PPMC notes that there are multiple trademark infringements based on both
the redistribution of MXNet with addition of unreleased code and the
redistribution of MXNet with Category-X GPL and Category-X NVidia
components. PPMC intends to handle both issues separately.
#### Unauthorized redistribution of unreleased code by third-parties
PPMC members have reached out to the offending third parties (Nvidia
Corporation
and Amazon Web Services) via inofficial channels and notified them of the
problem. If the problem is not resolved by the end of July 2020, PPMC will
request guidance from the Brand Management Team on how to formally notify
the offenders of their trademark infringement.
#### Unauthorized redistribution of Category-X GPL and NVidia CUDA EULA
components by third-parties
PPMC members note that the issue of "NVidia CUDA EULA infecting any
application
built with CUDA support" is an industry-wide problem. PPMC is not aware of
any individual or corporation correctly labeling their binary
distributions
subject
to the NVidia CUDA EULA. Instead, PPMC found that for example Facebook
claims distribution of PyTorch under BSD License (BSD-3) and Google claims
distribution of Tensorflow under Apache 2.0 License, despite both being
subject to the CUDA EULA. Thus, PPMC has contacted NVidia Corporation and
requested
NVidia Corporation to add clarifying language that applications based on
the CUDA SDK with material additional functionality may be licensed under
a license
of the application owner's choice, consistent with existing industry
"practice".
The issue was also discussed with NVidia and other Deep Learning Framework
implementers during the Nvidia Deep Learning Framework Developer Council
meeting,
during which NVidia promised to conclude their internal review and
follow-up with the PPMC.
PPMC thus recommends to give NVidia the chance to clarify and improve
their
license. As NVidia employs a team for working on MXNet, the PPMC is
optimistic
about receiving a detailed clarification and resolution from NVidia.
If NVidia fails to clarify their license or the resolution is
unsatisfactory
within Q3 2020, the PPMC will notify any third-parties about their license
infringement and ask them to take down or rename their redistributions
containing Category-X pieces.
Due the substantial overhead of trademark-infringement takedown notices
for any
involved party, PPMC is further awaiting NVidia's clarification prior to
contacting third-parties about trademark infringement due to inclusion of
GPL
components. This is to avoid sending two separate takedown notices in case
of an unsatisfactory response by NVidia.
The following downstream software distributors are known to the PPMC to be
using the name MXNet while redistributing Category-X components
- pypi.org
- hub.docker.com
- ngc.nvidia.com
- aws.amazon.com
### Signed-off-by:
- [X] (mxnet) Henri Yandell
Comments: Kudos to the project on the licensing review; respect to Bob
and Justin for their work as well.
- [ ] (mxnet) Markus Weimer
Comments:
- [X] (mxnet) Michael Wall
Comments: Good progress on the issues. They are complex but important.
- [X] (mxnet) Bob Paulin
Comments: Team is making thoughtful process on the issues. Lots of
good support and experiance coming from VPs of Brand and Incubator PMC.
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
Justin Mclean: Good to see progress on these issues. But I have two concerns
a) you may not have an active PPMC b) branding all release as "3rd party" may
not be the best way to solve the issue. Hopefully the license discussion will
be fruitful.
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address past release issues. Successfully and smoothly make releases
- ONGOING.
2. Improve development process and tooling to help reduce the overhead of
releases
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. MXNet 2.0 has started (10 projects completed):
https://github.com/apache/incubator-mxnet/projects/18
RFC: https://github.com/apache/incubator-mxnet/issues/16167
2. There have been stability issues and high cost associated with the CI
system due to complexity and high load.
* We are sufferring from high failure rate in CI due to technical
debts such as flaky tests.
* Efforts on reducing waste in time and cost on CI:
https://github.com/apache/incubator-mxnet/projects/19
* We intend to update our toolchain and improve engineering practices
to address technical debts and improve the development experiences.
* We want to improve automation and address licensing issues to reduce
release overhead.
* We seek experiences from other projects with extensive C/C++
development on validating for a diverse set of platforms,
and experiences on reducing the test complexity and runtime.
We appreciate any lessons other projects can share in these aspects.
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 781
(2019-03-28; +2.4% since last report)
* Active discussions on user forums
* https://discuss.mxnet.io/ (English, 2.5K registered users (+8.7%)
and 7.7K posts (+4.0%))
* https://discuss.gluon.ai/ (Chinese, 8.1K registered users (+6.9%)
and 21.9K posts(+1.3% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.8k followers (+5.9%)
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.4k followers
(+5.4%)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 1.04k
subscriber (+9.6% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 5.71k
subscribers (+5.7%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
22k subscriber (+4.8% since last report)
* Highlights in MXNet ecosystem
* Dive into Deep Learning has 50.5K 28-day active users, and has
attracted 20.7K stars & 230+ contributors on GitHub. It has been
adopted as a textbook or reference book by 70+ universities from 21
countries, such as MIT, UC Berkeley, CMU, UPenn, IIT Bombay, and NUS.
* GluonCV v0.6.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.6.0)
* GluonNLP v0.9.1 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.9.1)
* GluonTS v0.5.0 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.5.0)
* MXNet Model Server v1.1.0 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.1.0)
* DeepInsight (https://github.com/deepinsight/insightface)
* Sockeye (https://github.com/awslabs/sockeye)
### How has the project developed since the last report?
1) 1.6.0 was released:
https://github.com/apache/incubator-mxnet/releases/tag/1.6.0
with over 830+ patches of new features, improvements, and fixes.
2) 2.0 project: https://github.com/apache/incubator-mxnet/projects/18
3) Github statistics of last month:
* Feb 28, 2019 – March 28, 2019: Excluding merges, 40 authors have pushed
75 commits to master and 78 commits to all branches. On master, 404
files have changed and there have been 17,944 additions and 6,344
deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [X] Nearing graduation
- [ ] Other:
### Date of last release:
2020-02-20
### When were the last committers or PPMC members elected?
2020-03-02
### Have your mentors been helpful and responsive?
Yes, mentors have been responsive and helpful as usual
### Signed-off-by:
- [X] (mxnet) Henri Yandell
Comments:
- [X] (mxnet) Markus Weimer
Comments:
- [X] (mxnet) Michael Wall
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Address past release issues. Successfully and smoothly make releases
- ONGOING.
2. Improve development process and tooling to help reduce the overhead of
releases
- ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. MXNet 2.0 has started:
https://github.com/apache/incubator-mxnet/projects/18
RFC: https://github.com/apache/incubator-mxnet/issues/16167
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 763
(2019-12-31; +3.8% since last report)
* Active discussions on user forums
* https://discuss.mxnet.io/ (English, 2.3K registered users (+9.5%)
and 7.4K posts (+7.2%))
* https://discuss.gluon.ai/ (Chinese, 8.5K registered users (+10.4%)
and 21.6K posts(+2.3% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.7k followers (+6.2%)
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.4k followers
(+4.3%)
* Meetup group
(https://www.meetup.com/pro/deep-learning-with-apache-mxnet/) w/ 10
groups in 8 countries, 2207 members
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 949
subscriber (+12.8% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 5.4k
subscribers (+8.2%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
21k subscriber (+15.4% since last report)
* Highlights in MXNet ecosystem
* MXNet Gluon book published (https://zh.d2l.ai/) first 34k hard copies
* Dive into Deep Learning has 59K 28-day active users, and has
attracted 15K stars & 210+ contributors on GitHub. It has been adopted
as a textbook or reference book by 30+ universities in U.S., China,
Spain, Brazil, India, and Australia, such as MIT, UC Berkeley, CMU,
IIT Bombay, PKU, and SJTU.
* GluonCV v0.6.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.6.0)
* GluonNLP v0.8.2 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.8.2)
* GluonTS v0.4.2 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.4.2)
* MXNet Model Server v1.0.9 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.0.9)
* GluonFace (https://github.com/THUFutureLab/gluon-face)
* DeepInsight (https://github.com/deepinsight/insightface)
### How has the project developed since the last report?
1) 1.5.1 patch release and 1.6.0 in progress:
https://github.com/apache/incubator-mxnet/releases/tag/1.5.1
https://github.com/apache/incubator-mxnet/releases/tag/1.6.0
with over 830+ patches of new features, improvements, and fixes.
2) 2.0 project: https://github.com/apache/incubator-mxnet/projects/18
3) Many ongoing projects:
* numpy-compatbile deep learning:
https://github.com/apache/incubator-mxnet/projects/14
* CPU performance and quantization:
https://github.com/apache/incubator-mxnet/projects/15
* MKLDNN 1.0 upgrade:
https://github.com/apache/incubator-mxnet/projects/16
* New Website Launch:
https://github.com/apache/incubator-mxnet/projects/17
4) Github statistics of last month:
* Nov 30, 2019 – Dec 30, 2019: Excluding merges, 45 authors have pushed
109 commits to master and 131 commits to all branches. On master, 314
files have changed and there have been 20,474 additions and 3,842
deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [x] Nearing graduation
- [ ] Other:
### Date of last release:
2019-10-01
### When were the last committers or PPMC members elected?
2019-11-24
### Have your mentors been helpful and responsive?
Yes, mentors have been responsive and helpful as usual
### Signed-off-by:
- [ ] (mxnet) Henri Yandell
Comments:
- [ ] (mxnet) Markus Weimer
Comments:
- [x] (mxnet) Michael Wall
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Revisit Apache Maturity Model Assessment — ONGOING.
2. Address past release issues. Successfully and smoothly make releases -
ONGOING.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. Our community is planning for the long-term development of MXNet.
RFC: https://github.com/apache/incubator-mxnet/issues/16167
Discussion: https://github.com/apache/incubator-mxnet/issues/9686
2. Our community established the process for conflict resolution and
violation reporting.
https://github.com/apache/incubator-mxnet/blob/master/CODE_OF_CONDUCT.md
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 735
(2019-06-24; +4.0% since last report)
* Active discussions on user forums
* https://discuss.mxnet.io/ (English, 2.1K registered users (+17%)
and 9.4K posts (+23%))
* https://discuss.gluon.ai/ (Chinese, 7.7K registered users (+18%)
and 21.1K posts(+10% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.6k followers (+14%)
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.3k followers
(+8.7%)
* Meetup group
(https://www.meetup.com/pro/deep-learning-with-apache-mxnet/) w/ 10
groups in 8 countries, 2182 members
* (in China) Zhihu w/ 7.9k followers (+4.0%), WeChat official account
w/ 4.6k
followers. (+9.5% since last report)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 841
subscriber (+18.6% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 5.0k
subscribers (+13.6%)
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
18k subscriber (+20% since last report)
* Highlights in MXNet ecosystem
* MXNet Gluon book published (https://zh.d2l.ai/) first 34k hard copies
* Dive into Deep Learning has 59K 28-day active users, and has
attracted 15K stars & 210+ contributors on GitHub. It has been adopted
as a textbook or reference book by 30+ universities in U.S., China,
Spain, Brazil, India, and Australia, such as MIT, UC Berkeley, CMU,
IIT Bombay, PKU, and SJTU.
* GluonCV v0.5.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.5.0)
* GluonNLP v0.8.1 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.8.1)
* GluonTS v0.3.3 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.3.3)
* MXNet Model Server v1.0.7 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.0.7)
* GluonFace (https://github.com/THUFutureLab/gluon-face)
* DeepInsight (https://github.com/deepinsight/insightface)
### How has the project developed since the last report?
1) 1.5.0 release and 1.5.1 patch release:
https://github.com/apache/incubator-mxnet/releases/tag/1.5.0
https://github.com/apache/incubator-mxnet/releases/tag/1.5.1
with over 830+ patches of new features, improvements, and fixes.
2) 2.0 Roadmap RFC published:
https://lists.apache.org/thread.html/9d344832757860c0ec897cb79b84f4d552c6c37
e52ae33f2c92b50f7@%3Cdev.mxnet.apache.org%3E
3) Many ongoing projects:
* numpy-compatbile deep learning:
https://github.com/apache/incubator-mxnet/projects/14;
* New Website Launch:
https://mxnet.apache.org/
https://github.com/apache/incubator-mxnet/projects/17
4) Github statistics of last month:
* Aug 30, 2019 – Sept 30, 2019: Excluding merges, 54 authors have pushed
128 commits to master and 164 commits to all branches. On master, 1038
files have changed and there have been 55,926 additions and 32,705
deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [x] Nearing graduation
- [ ] Other:
### Date of last release:
2019-10-01
### When were the last committers or PPMC members elected?
2019-10-01
### Have your mentors been helpful and responsive?
Yes, mentors have been responsive and helpful as usual
### Signed-off-by:
- [ ] (mxnet) Henri Yandell
Comments:
- [ ] (mxnet) Markus Weimer
Comments:
- [X] (mxnet) Michael Wall
Comments: PPMC is doing a good job of working through an issue with
community interactions.
- [X] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
### Three most important unfinished issues to address before graduating:
1. Increase diversity in contributors, committers, and PMC members — NEAR
COMPLETION.
2. Identify remaining ICLAs or SGAs that need signing – NEAR COMPLETION.
3. Revisit Apache Maturity Model Assessment — TODO.
### Are there any issues that the IPMC or ASF Board need to be aware of?
1. Based on usability feedback from users, community has started a
redesign of the website.
### How has the community developed since the last report?
* The number of GitHub contributors increased to currently 707
(2019-06-24; +3.7% since last report)
* Active discussions on user forums
* https://discuss.mxnet.io/ (English, 1.8K registered users and 9.4K
posts (+68% since last report))
* https://discuss.gluon.ai/ (Chinese, 6.5K registered users and 37K
posts(+93% since last report))
* Active blogs and social media presence
* Medium (https://medium.com/apache-mxnet) w/ 1.4k followers
* Twitter account (https://twitter.com/ApacheMXNet) w/ 2.1k followers
* Meetup group
(https://www.meetup.com/pro/deep-learning-with-apache-mxnet/) w/ 10
groups in 8 countries, 2141 members
* (in China) Zhihu w/ 7.6k followers, WeChat official account w/ 4.2k
followers. (+40% since last report)
* Active video channels
* YouTube channel (https://www.youtube.com/apachemxnet) w/ 709
subscriber (+13% since last report)
* Chinese YouTube channel
(https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 4.4k
subscribers
* (in China) bilibili space (https://space.bilibili.com/209599371) w/
15k subscriber (+15% since last report)
* Highlights in MXNet ecosystem
* MXNet Gluon book published (https://zh.d2l.ai/) first 21.5k hard
copies
* MXNet Gluon book (www.d2l.ai, Dive into Deep Learning/D2L) released
* v1.0.0-rc0
(https://github.com/d2l-ai/d2l-zh/releases/tag/v1.0.0-rc0)
in Chinese
* v0.6.0 (https://github.com/d2l-ai/d2l-en/releases/tag/v0.6.0) in
English
* Dive into Deep Learning has 47.5K 28-day active users, and has
attracted 11.5K stars & 200+ contributors on GitHub. It has been adopted
as a textbook or reference book by 15+ universities in U.S., China,
Spain, and Australia, such as UC Berkeley.
* GluonCV v0.4.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.4.0)
* GluonNLP v0.6.0 release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.6.0)
* GluonTS v0.1.4 release
(https://github.com/awslabs/gluon-ts/releases/tag/v0.1.4)
This is a new toolkit for deep-learning based time-series modeling.
* MXNet Model Server v1.0.4 release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.0.4)
* GluonFace (https://github.com/THUFutureLab/gluon-face)
* DeepInsight (https://github.com/deepinsight/insightface)
### How has the project developed since the last report?
1) 1.4.1 patch release:
https://github.com/apache/incubator-mxnet/releases/tag/1.4.1
2) 1.5.0 release in progress (pre-release v1.5.0.rc1):
https://github.com/apache/incubator-mxnet/releases/tag/1.5.0.rc1
with over 750 patches of new features, improvements, and fixes.
3) Roadmap discussion on 2.0 in progress
https://github.com/apache/incubator-mxnet/issues/9686
4) Code donation from dmlc/mshadow in progress.
5) Many ongoing projects:
* numpy-compatbile deep learning:
https://github.com/apache/incubator-mxnet/projects/14;
* CPU performance and quantization:
https://github.com/apache/incubator-mxnet/projects/15;
* Mixed precision GPU training (AMP):
https://github.com/apache/incubator-mxnet/pull/14173, etc.
6) Github statistics of last month:
* May 24, 2019 – June 24, 2019: Excluding merges, 16 authors have pushed
88 commits to master and 140 commits to all branches. On master, 250 files
have changed and there have been 12,939 additions and 9,919 deletions.
### How would you assess the podling's maturity?
Please feel free to add your own commentary.
- [ ] Initial setup
- [ ] Working towards first release
- [ ] Community building
- [x] Nearing graduation
- [ ] Other:
### Date of last release:
2019-04-29
### When were the last committers or PPMC members elected?
2019-05-20
### Have your mentors been helpful and responsive?
Mentors have been providing helps per requests from community.
### Signed-off-by:
- [ ] (mxnet) Henri Yandell
Comments:
- [ ] (mxnet) Markus Weimer
Comments:
- [x] (mxnet) Michael Wall
Comments:
- [ ] (mxnet) Bob Paulin
Comments:
- [ ] (mxnet) Jason Dai
Comments:
### IPMC/Shepherd notes:
- Drew Farris (shepherd): Very active project. One mentor observed on the
mailing list, otherwise not a significant amount of mentor activity
apparent - possibly due to the proximity toward graduation.
- Justin Mclean: I'm not sure why the website redesign is an issue for the
IPMC or board. Can you please give some more detail.
A Flexible and Efficient Library for Deep Learning MXNet has been incubating since 2017-01-23. Three most important unfinished issues to address before graduating: 1. Increase diversity in contributors, committers, and PMC members — NEAR COMPLETION. 2. Identify remaining ICLAs or SGAs that need signing – NEAR COMPLETION. 3. Revisit Apache Maturity Model Assessment — TODO Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be aware of? 1. Community reached consensus on Gluon branding. In the context of the Gluon API in MXNet, it would be mentioned as MXNet Gluon. https://lists.apache.org/thread.html/af7dcb430e2cedf23d1531e79877e8bf2b40ec392e40853a2d7015da@%3Cdev.mxnet.apache.org%3E 2. Community addressed the IP/licensing issues discovered during previous releases. Updated status can be found here: https://cwiki.apache.org/confluence/display/MXNET/MXNet+Source+Licenses 3. MXNet has significant user presence in China that are worth optimizing for in terms of infrastructure. Setting up CDN in China requires ICP filing. How has the community developed since the last report? * The number of GitHub contributors increased to currently 682 (2019-04-22; +4.4% since last report) * Active discussions on user forums https://discuss.mxnet.io/ (English, 1.6K registered users and 5.6K posts) https://discuss.gluon.ai/ (Chinese, 6K registered users and 19.2K posts) * Active blogs and social media presence Medium (https://medium.com/apache-mxnet) w/ 1.3k followers (+30% since last report). Twitter account (https://twitter.com/ApacheMXNet) w/ 2k followers on (+16% since last report) Meetup group (https://www.meetup.com/pro/deep-learning-with-apache-mxnet/) w/ 10 groups in 8 countries, 1998 members (+65% since last report) (in China) Zhihu w/ 7.5k followers, WeChat official account w/ 3k followers. (reported for the first time) * Active video channels YouTube channel (https://www.youtube.com/apachemxnet) w/ 627 subscriber (+29% since last report) Chinese YouTube channel (https://www.youtube.com/channel/UCjeLwTKPMlDt2segkZzw2ZQ) w/ 4.1k subscribers (reported for the first time) (in China) bilibili space (https://space.bilibili.com/209599371) w/ 13k subscriber (reported for the first time) * Highlights in MXNet ecosystem MXNet Gluon book (www.d2l.ai, Dive into Deep Learning/D2L) released v1.0.0-rc0 (https://github.com/d2l-ai/d2l-zh/releases/tag/v1.0.0-rc0) in Chinese and v0.6.0 (https://github.com/d2l-ai/d2l-en/releases/tag/v0.6.0) in English. Dive into Deep Learning has 47.5K 28-day active users, and has attracted 9.3K stars & 200+ contributors on GitHub. It has been adopted as a textbook or reference book by 15+ universities in U.S., China, Spain, and Australia, such as UC Berkeley. GluonCV v0.4.0 release (https://github.com/dmlc/gluon-cv/releases/tag/v0.4.0) GluonNLP v0.6.0 release (https://github.com/dmlc/gluon-nlp/releases/tag/v0.6.0) MXNet Model Server v1.0.2 release (https://github.com/awslabs/mxnet-model-server/releases/tag/v1.0.2) GluonFace (https://github.com/THUFutureLab/gluon-face) DeepInsight (https://github.com/deepinsight/insightface) How has the project developed since the last report? 1) Released v1.4.0: https://github.com/apache/incubator-mxnet/releases/tag/1.4.0; https://blogs.apache.org/mxnet/entry/apache-mxnet-1-4-is1 2) Started v1.4.1 patch release https://lists.apache.org/thread.html/3bb49a1016fafd0840d14f099ce47c7a1822da45f7ca2187c0f03c64@%3Cdev.mxnet.apache.org%3E; Started roadmap discussion on short-term 1.5.0 release https://github.com/apache/incubator-mxnet/issues/14619; Started roadmap discussion on long-term 2.0 plan https://github.com/apache/incubator-mxnet/issues/9686; 3) Code donation from dmlc/mshadow. Community expressed the desire to assimilate dmlc/mshadow code base into mxnet. https://lists.apache.org/thread.html/c1ba34330b0eb52ef3a3a30da6d34964a35a01c 320e93067e94ed306@%3Cdev.mxnet.apache.org%3E. After discussion, DMLC reached agreement to donate dmlc/mshadow code to mxnet, which is its sole consumer. 4) Many ongoing projects: numpy-compatbile deep learning: https://github.com/apache/incubator-mxnet/projects/14; CPU performance and quantization: https://github.com/apache/incubator-mxnet/projects/15; Mixed precision GPU training (AMP): https://github.com/apache/incubator-mxnet/pull/14173, etc. 5) Github statistics of last month: * March 24, 2019 – April 24, 2019: Excluding merges, 53 authors have pushed 102 commits to master and 137 commits to all branches. On master, 495 files have changed and there have been 14,977 additions and 5,391 deletions. How would you assess the podling's maturity? Please feel free to add your own commentary. [x] Initial setup [x] Working towards first release [x] Community building [x] Nearing graduation Date of last release: 2019-03-04 MXNet 1.4.0 When were the last committers or PPMC members elected? As recent as 2019-04-12. New committer since last report (+12): Iblis Lin, Da Zheng, Steffen Rochel, Lin Yuan, Nicolas Modrzyk, Jackie Wu, Aston Zhang, Ding Kuo, Patric Zhao, Kevin Qin, Jiajun Wu, Jeremie Desgagne-Bouchard New PPMC member (+1): Qing Lan Have your mentors been helpful and responsive or are things falling through the cracks? In the latter case, please list any open issues that need to be addressed. Mentors continue providing guidance and support. Signed-off-by: [x](mxnet) Henri Yandell Comments: Noting that Gluon is a trademark of Microsoft + Amazon's. [ ](mxnet) Markus Weimer Comments: [x](mxnet) Michael Wall Comments: [ ](mxnet) Bob Paulin Comments: [ ](mxnet) Jason Dai Comments: IPMC/Shepherd notes: Justin Mclean: You might want to talk to eCharts as they are also setting up a CDN with infra's help.
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way --
ESTABLISHED.
2. Grow the community -- ONGOING.
3. Identify remaining ICLAs or SGAs that need signing – NEAR COMPLETION.
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?
None
How has the community developed since the last report?
* The community voted on updated process becoming project committer and
PPMC member, vote results:
https://lists.apache.org/thread.html/458f234120ed8fbf98ddd57aa42eafe398cdd7e
aeb89ac2be9d214c7@%3Cdev.mxnet.apache.org%3E
* The community voted on separating PMC and Committership -
https://lists.apache.org/thread.html/5bbd3b6daf5e89d17f92aed3aa8acd192841f11
822dcd21627ad5389@%3Cdev.mxnet.apache.org%3E
* We have active contributors from companies like Nvidia, Intel, Wolfram
Design and more. The number of github contributors increased from 506
(3/31/2018) to currently 653 (2019-01-02; +10% since last report)
* Active discussions on https://discuss.mxnet.io/ (English) and
https://discuss.gluon.ai/ (Chinese)
* We developed an active blog on Medium
(https://medium.com/apache-mxnet) and have now 1k followers (+12.2% since
last report).
* We established a YouTube channel (https://www.youtube.com/apachemxnet)
and have 485 subscriber (+31% since last report).
* We are tweeting to promote new content and community events such as
meetups and conferences. Our Twitter account has 1768 followers on
https://twitter.com/ApacheMXNet (17% since last report).
* We established a global MXNet meetup group with 8 groups (+4) in 7 (+4)
countries and 1209 member (+45% since last report).
https://www.meetup.com/pro/deep-learning-with-apache-mxnet/
* Highlights in MXNet ecosystem: GluonCV v0.3.0 release
(https://github.com/dmlc/gluon-cv/releases/tag/v0.3.0), GluonNLP v0.5.0
release
(https://github.com/dmlc/gluon-nlp/releases/tag/v0.5.0), MXNet Model
Server v1.0.1
release
(https://github.com/awslabs/mxnet-model-server/releases/tag/v1.0.1),
Gluon-zh book v0.7 release
(https://github.com/d2l-ai/d2l-zh/releases/tag/v0.7),
and active developments in Gluon-en (https://github.com/d2l-ai/d2l-en),
GluonFace (https://github.com/THUFutureLab/gluon-face) and DeepInsight
(https://github.com/deepinsight/insightface).
How has the project developed since the last report?
1) Released v1.3.1:
https://github.com/apache/incubator-mxnet/releases/tag/1.3.1
2) Completed vote for 1.4.0 release on dev@:
https://github.com/apache/incubator-mxnet/releases/tag/1.4.0;
https://lists.apache.org/thread.html/236554041a412415df783d652830272beb21d36
660e8e0d5e0237f58@%3Cdev.mxnet.apache.org%3E
Started vote on general@ on December 28, 2018:
https://lists.apache.org/thread.html/a645bdb72bc55e05ec57f5b07a95c2579971f0a
32ed257f31199e218@%3Cgeneral.incubator.apache.org%3E
3) Github statistics:
* https://github.com/apache/incubator-mxnet/pulse/monthly - December 2,
2018 – January 2, 2019
Excluding merges, 56 authors have pushed 114 commits to master and 126
commits to all branches.
On master, 359 files have changed and there have been 15,054 additions
and 3,531 deletions.
65 issues got closed and 61 new issues created during the reporting
period.
* https://github.com/apache/incubator-mxnet/pulse/monthly - November 2,
2018 - December 1, 2018
Excluding merges, 67 authors have pushed 202 commits to master and 283
commits to all branches.
On master, 521 files have changed and there have been 21,810 additions
and 8,161 deletions.
145 issues got closed and 98 new issue created during the reporting
period.
* https://github.com/apache/incubator-mxnet/pulse/monthly - October 2,
2018 - November 1, 2018
Excluding merges, 55 authors have pushed 125 commits to master and 135
commits to all branches.
On master, 617 files have changed and there have been 37,194 additions
and 7,424 deletions.
119 issues got closed and 131 new issue created during the reporting
period.
How would you assess the podling's maturity?
Please feel free to add your own commentary.
[ ] Initial setup
[ ] Working towards first release
[ ] Community building
[X ] Nearing graduation
[ ] Other:
Date of last release:
2018-11-29 MXNet 1.3.1
When were the last committers or PPMC members elected?
New committer:
Qin Lang (November 21, 2018;
https://lists.apache.org/thread.html/27a2943240180666ce9e9219a049322966c851a
259d908e241103d7a@%3Cdev.mxnet.apache.org%3E)
Thomas Delteil (November 20, 2018;
https://lists.apache.org/thread.html/d4ad366e81852547a732b4c31b7f3f020132e20
ba9c4cae7fd908f31@%3Cdev.mxnet.apache.org%3E)
Kellen Sunderland (November 21, 2018;
https://lists.apache.org/thread.html/4f18f85e627c7c0db5ad9ea46b1c277ef304f9d
445be6b7e1e3ae3af@%3Cdev.mxnet.apache.org%3E)
Tao Lv (Novmember 21, 2018;
https://lists.apache.org/thread.html/3b14dadf32c84b0249758f25db52602527311ec
0efc7bf94c93aaeee@%3Cdev.mxnet.apache.org%3E)
Rahul Huilgol (December 4, 2018:
https://lists.apache.org/thread.html/1b30b460d4bbc2ddb7a0bc7113e8d064ba75241
4d7d97a6e54d442f3@%3Cdev.mxnet.apache.org%3E)
Aaron Markham (December 4, 2018;
https://lists.apache.org/thread.html/f807eaa4e6477c6133e4949b7b38428adff5726
8620343d4c7c1e7ce@%3Cdev.mxnet.apache.org%3E)
Da Zhang (December 17, 2018;
https://lists.apache.org/thread.html/ffaa9425c9f5da77176e0ad92d136e70179720e
4f485456daf6d7231@%3Cdev.mxnet.apache.org%3E)
No new PPMC members got elected in the reporting period.
Have your mentors been helpful and responsive or are things falling
through the cracks? In the latter case, please list any open issues
that need to be addressed.
Mentors continue providing guidance and support.
Signed-off-by:
[ ](mxnet) Sebastian Schelter
Comments:
[ ](mxnet) Henri Yandell
Comments:
[X](mxnet) Markus Weimer
Comments:
[X](mxnet) Jim Jagielski
[X](mxnet) Michael Wall
Comments: Discussion on unapproved releases started, but needs
resolution. Good effort on community building.
[ ](mxnet) Bob Paulin
Comments:
[X](mxnet) Jason Dai
Comments:
IPMC/Shepherd notes:
Justin Mclean: The issue with unapproved releases and issues around that
should have been mentioned in this report.
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way --
ESTABLISHED.
2. Grow the community -- ONGOING.
3. Identify remaining ICLAs or SGAs that need signing – NEAR COMPLETION.
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?
None.
How has the community developed since the last report?
* We have active contributors from companies like Intel, Nvidia, Wolfram
Design and more. The number of github contributors increased from 506
(3/31/2018) to currently 592 (9/28/2018, 2.4% MoM),
* Active discussions on https://discuss.mxnet.io/ (English) and
https://discuss.gluon.ai/ (Chinese)
* We developed an active blog on Medium
(https://medium.com/apache-mxnet). We have posted 28 blogs YTD
(12.0% MoM), and have 891 (16.3% MoM) followers.
* We established a YouTube channel (https://www.youtube.com/apachemxnet)
and have 370 subscriber (+32.1% MoM) and 32 videos published.
* We are tweeting to promote new content and community events such as
meetups and conferences. Our Twitter account has 1,510 followers on
https://twitter.com/ApacheMXNet (3.6% MoM).
* 62 new public tutorials about MXNet have been published YTD in
addition to the Medium blogs. Details available at
https://lists.apache.org/thread.html/52f88e9dc7a6a2a1dfa5ad41c469fe2cdd1209a
0be2eb345bc2f9a96@%3Cuser.mxnet.apache.org%3E
https://lists.apache.org/thread.html/dea9184350f2fe87ce450722ead28072f763196
045f39859190f83f8@%3Cuser.mxnet.apache.org%3E
https://discuss.mxnet.io/t/apache-mxnet-digest-august-2018
* We established a global MXNet meetup group with 4 groups in 3
countries and 833 member.
https://www.meetup.com/pro/deep-learning-with-apache-mxnet/
* Community members participated at conferences and other venues,
staffing booths and giving a total of nine presentations about MXNet:
ACNA18 (Montréal, total 4 presentations about MXNet from community
members), Linux Foundation Open Source Summit (Vancouver), AI Meetup
(Toronto), MXNet Meetup (Seattle, San Francisco), Open Data Science
Conference (London), Big Data Summit (Boston), AI and Neural Networks
on Arm Summit at Linaro conference (Vancouver),O'Reilly AI conference
https://www.oreilly.com/ideas/machine-learning-in-the-cloudm
How has the project developed since the last report?
1) v1.3 release notes:
https://github.com/apache/incubator-mxnet/releases/tag/1.3.0
2) Community updated MXNet backend for Keras to v2.2.2
3) Github statistics
Statistics are captured at the end of each month from
https://github.com/apache/incubator-mxnet/pulse/monthly
July 2018
* Excluding merges, 62 authors have pushed 180 commits to master and 188
commits to all branches. On master, 614 files have changed and there
have been 29,908 additions and 26,648 deletions.
* 273 issues got closed and 154 new issue created during the reporting
period.
August 2018:
* Excluding merges, 68 authors have pushed 217 commits to master and 234
commits to all branches. On master, 463 files have changed and there
have been 20,882 additions and 6,404 deletions.
* 200 issues got closed and 98 new issue created during the reporting
period.
September 2018:
* Excluding merges, 47 authors have pushed 132 commits to master and 141
commits to all branches. On master, 439 files have changed and there
have been 13,797 additions and 5,871 deletions.
* 124 issues got closed and 87 new issue created during the reporting
period.
How would you assess the podling's maturity?
Please feel free to add your own commentary.
Podling is still having difficulties to grow committer community.
Maturity == Medium.
[ ] Initial setup
[ ] Working towards first release
[X] Community building
[ ] Nearing graduation
[ ] Other:
Date of last release:
2018-09-19 (v1.3) Apache MXNet-incubating 1.3.0 (major release)
was published on September 19, 2018
https://github.com/apache/incubator-mxnet/releases/tag/1.3.0
When were the last committers or PPMC members elected?
2018-08-10 Carin Meier
2018-09-27 additional mentors joined to help with diversity
among MXNet's mentors:
Jason Dai; Jim Jagielski; Bob Paulin; and Michael Wall.
2018-09-27 Suneel Marthi retired as mentor
Have your mentors been helpful and responsive or are things falling
through the cracks? In the latter case, please list any open issues
that need to be addressed.
We thank the mentors for ongoing support and guidance.
Signed-off-by:
[ ](mxnet) Sebastian Schelter
Comments:
[X](mxnet) Jason Dai
Comments:
[X](mxnet) Henri Yandell
Comments: Needs to work on converting contributors to committers.
[ ](mxnet) Markus Weimer
Comments:
[X](mxnet) Jim Jagielski
Comments:
[ ](mxnet) Bob Paulin
Comments:
[X](mxnet) Michael Wall
Comments:
IPMC/Shepherd notes:
Apache MXNet is a lean, flexible, and ultra-scalable deep learning framework
that supports state of the art in deep learning models, including
convolutional neural networks (CNNs) and long short-term memory networks
(LSTMs). The framework has its roots in academia and came about through the
collaboration and contributions of researchers at several top universities.
It has been designed to excel at computer vision, speech, language
processing and understanding, generative models, concurrent neural networks,
and recurrent neural networks. MXNet allows you to define, train, and deploy
networks across a wide array of use cases from massive cloud infrastructure
to mobile and connected devices. It provides a very flexible environment
with support for many common languages and the ability to utilize both
imperative and symbolic programming constructs. MXNet also very lightweight.
This allows it to scale across multiple GPUs and multiple machines very
efficiently, which is beneficial when conducting training on large datasets
in the cloud.
Apache MXNet has been incubating since 23-Jan, 2017.
Four most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way
-- ESTABLISHED.
2. Grow the community -- ONGOING.
3. Bring website up to Apache standard – COMPLETED.
4. Identify remaining ICLAs or SGAs that need signing – NEAR COMPLETION.
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be aware
of?
None
How has the community developed since the last report?
a) Various Slack channels, dev@ mailing lists, and user discussion
forums (http://discuss.mxnet.io) are being used actively. The contributors
have been working on having all discussions on the public dev@ mailing list
as much as possible. At least one of the discussions infringed on Apache
code of conduct and mentors had to step in and provide guidance to the
community.
b) After lengthy discussion the community agreed to establish
user@mxnet.incubator.apache.org with the goal to increase users as well as
grow the contributor community.
c) Community events:
- In April the community organized the first MXNet meetup in Seattle. We had
over 100 attendees, 15% from outside Amazon.
- In May the community organized a virtual hangout with two sessions to
collect feedback and ideas from the community.
- Korean MXNet community is growing well and most of contributors are
outside of Amazon:
https://www.meetup.com/ko-KR/MXNet-Korea-User-Group/events/248531181/ . The
Facebook group has 530 members in https://www.facebook.com/groups/mxnetkr
- Community members presented at
https://www.meetup.com/Artificial-Intelligence-in-Practice/events/250882931/?
_xtd=gqFyqDI5NDE5MTUyoXCkaXBhZA&from=ref
- Community members presented at meetups in Vancouver (link see under
Webinars)
- Community members presented Intro to Amazon SageMaker with a sentiment
analysis demo using MXNet and Gluon at AWS Loft ML Day in SF on 6-19 to ~150
attendees
- Community members presented Model serving with Model Server for MXNet at
AWS Loft ML Day in SF on 6-19 to ~150 attendees
- A committer presented Distributed inference with Spark and MXNet at Apache
Roadshow in Berlin, 13-14 June (talk abstract -
https://foss-backstage.de/session/distributed-inference-using-apache-mxnet-an
d-apache-spark)
- A committer presented Distributed inference with Spark and MXNet at
Spark+AI in SF, ~100 attendees (https://databricks.com/speaker/naveen-swamy)
- Community members presented Model serving with MXNet at dotAI conference
in Paris, France, ~300 attendees, May 30
(https://www.dotconferences.com/2018/05/hagay-lupesko-model-serving-for-deep-
learning)
- Community members presented ONNX with MXNet demo at prepareAI conference
in St Louis, May 8, ~100 attendees
(http://prepare.ai/conference/conference-agenda-details/)
- Community members presented Model serving with MXNet at prepareAI
conference in St Louis, May 8, ~100 attendees
(http://prepare.ai/conference/conference-agenda-details/)
- Community members presented Introduction to deep learning with MXNet and
Gluon, distinguished lecture, at BGU university in Israel, April 10, ~100
attendees (lhttps://www.cs.bgu.ac.il/~frankel/Lupesko/Lupesko.pdf)
- Community members presented Model serving with MXNet at AI IL meetup in
Tel Aviv, Israel, ~50 attendees
(https://www.meetup.com/artificial-intelligence-il/events/249312879/)
d) Blogs about MXNet: We established MXNet blog:
https://medium.com/apache-mxnet. As of today, the blog has 634 followers and
16 blogs published. Additionally, blogs have been publish related or about
MXNet at:
-
https://aws.amazon.com/blogs/machine-learning/category/artificial-intelligenc
e/apache-mxnet-on-aws/
- https://blogs.technet.microsoft.com/machinelearning/tag/mxnet
- https://zh.mxnet.io/blog/
- https://blogs.apache.org/mxnet
- https://github.com/chinakook/Awesome-MXNet
-
https://medium.com/mlreview/10-deep-learning-projects-based-on-apache-mxnet-8
231109f3f64
-
https://medium.com/datreeio/training-with-keras-mxnet-on-amazon-sagemaker-43a
34bd668ca
-
https://medium.com/@julsimon/a-quick-look-at-the-swish-activation-function-in
-apache-mxnet-1-2-79d9ff9d1673
- http://gigasquidsoftware.com/blog/2018/06/03/meet-clojure-mxnet-ndarray/
-
https://dzone.com/articles/ingesting-apache-mxnet-gluon-deep-learning-results
-
https://becominghuman.ai/an-introduction-to-the-mxnet-api-part-1-848febdcf8ab
- https://cosminsanda.com/posts/counting-object-with-mxnet-and-sagemaker/
- CVPR 2018 paper – Relation Networks for Object Detection -
https://arxiv.org/abs/1711.11575 and
https://github.com/msracver/Relation-Networks-for-Object-Detection
- Gluon-CV related blogpost -
https://dzone.com/articles/using-apache-mxnet-gluoncv-with-apache-nifi
-
https://aws.amazon.com/blogs/machine-learning/maximize-training-performance-w
ith-gluon-data-loader-workers/
-
https://medium.com/apache-mxnet/accelerating-deep-learning-on-cpu-with-intel-
mkl-dnn-a9b294fb0b9
-
https://medium.com/apache-mxnet/mxnet-for-pytorch-users-in-10-minutes-a735386
3406a
-
https://medium.com/apache-mxnet/gluoncv-deep-learning-toolkit-for-computer-vi
sion-9218a907e8da
-
https://medium.com/apache-mxnet/mxboard-mxnet-data-visualization-2eed6ae31d2c
- https://medium.com/apache-mxnet/mxnet-gluon-in-60-minutes-3d49eccaf266
- https://medium.com/apache-mxnet/announcing-apache-mxnet-1-2-0-d94f56da154b
-
https://medium.com/apache-mxnet/mxnet-1-2-adds-built-in-support-for-onnx-e2c7
450ffc28
-
https://aws.amazon.com/blogs/machine-learning/the-importance-of-hyperparamete
r-tuning-for-scaling-deep-learning-training-to-multiple-gpus/
-
https://medium.com/apache-mxnet/image-classification-with-mxnet-scala-inferen
ce-api-8ab6ce1bbccf
-
https://medium.com/apache-mxnet/object-detection-with-mxnet-scala-inference-a
pi-9049230c77fd
-
https://medium.com/apache-mxnet/scala-api-for-deep-learning-inference-now-ava
ilable-with-mxnet-v1-2-bcb13235db95
-
https://medium.com/apache-mxnet/train-using-keras-mxnet-and-inference-using-m
xnet-scala-api-49476a16a46a
- https://medium.com/apache-mxnet/page-segmentation-with-gluon-dcb4e5955e2
- https://medium.com/apache-mxnet/announcing-keras-mxnet-v2-2-4b8404568e75
- https://devblogs.nvidia.com/tensor-core-ai-performance-milestones/
-
https://aws.amazon.com/blogs/machine-learning/use-pre-trained-models-with-apa
che-mxnet/
e) Webinars, Technical talks and lectures about MXNet:
- Thomas Delteil presented at meetups in Vancouver:
https://www.youtube.com/watch?v=RgIa3_BjGyk&t=163s ,
https://www.youtube.com/watch?v=mN15vKIyfoA and
https://www.youtube.com/watch?v=K120xBnY6OA
- Gluon debugging: https://www.youtube.com/watch?v=6-dOoJVw9_0
- TVM stack: https://www.youtube.com/watch?v=DaCPJrTwT00
- https://mxnet.incubator.apache.org/tutorials/vision/cnn_visualization.html
- visual search: https://www.youtube.com/watch?v=9a8MAtfFVwI
- Cifar 10 super convergence: https://www.youtube.com/watch?v=O0XTkQPkUio
- Sparse Tensors: https://www.youtube.com/watch?v=smZfsYhDFkY
- Mixed precission training with MXNet:
https://www.youtube.com/watch?v=pR4KMh1lGC0
- See all videos on https://www.youtube.com/apachemxnet -
- Thomas Delteil presented a lecture on MXNet Gluon and Deep Learning at the
Machine Learning Summer School in Buenos Aires in front of 170 students.
http://mlss2018.net.ar/
How has the project developed since the last report?
a) The community released MXNet 1.2 with significant feature
enhancements:
1. Scala Inference API
2. ONNX model import
3. Model Quantization with calibration
4. MKL-DNN Integration
5. Improved exception handling for operators
6. Enhanced FP16 support
7. Profiling enhancements
Detailed release notes are provided on Apache Wiki and Github.
See
https://cwiki.apache.org/confluence/display/MXNET/Apache+MXNet+%28incubating%
29+1.2.0+Release+Notes
b) Github: In April the project had 523 contributors. Excluding merges,
62 authors have pushed 199 commits to master and 258 commits to all
branches. In May the project had 531 contributors. Excluding merges, 55
authors have pushed 141 commits to master and 177 commits to all branches.
In June the project had 550 contributors. Excluding merges, 66 authors have
pushed 171 commits to master and 221 commits to all branches. We are working
on encourage more contributors to the project.
b) The community voted to adopted Jira for issue management. The change
is methodology is partially adopted. On June 8th a committer started a vote
to stop using Jira which created a passionate debate and is still ongoing.
c) List of design proposals published on MXNet Apache Wiki : All design
proposals have been or are being discussed on dev@mxnet.apache.org. Four
design proposals originated from external contributors. See
https://cwiki.apache.org/confluence/display/MXNET/Design+Proposals.`
d) Ecosystem development:
1. MXNet Model Serving:
- Repo: https://github.com/awslabs/mxnet-model-server
-
https://medium.com/apache-mxnet/model-server-for-apache-mxnet-adds-support-fo
r-gluon-models-dd398a710f94
2. ONNX support
- Repo: https://github.com/onnx/model-zoo
- Repo: https://github.com/onnx/models
-
https://medium.com/apache-mxnet/mxnet-1-2-adds-built-in-support-for-onnx-e2c7
450ffc28
3. Keras: The community release MXNet backend for Keras, a high level
API for deep learning.
- Repo - https://github.com/awslabs/keras-apache-mxnet/
-
https://medium.com/@julsimon/apache-mxnet-as-a-backend-for-keras-2-9993f97843
e7
- https://medium.com/apache-mxnet/announcing-keras-mxnet-v2-2-4b8404568e75
-
https://aws.amazon.com/blogs/machine-learning/apache-mxnet-incubating-adds-su
pport-for-keras-2/
-
https://medium.com/apache-mxnet/keras-gets-a-speedy-new-backend-with-keras-mx
net-3a853efc1d75
e) Documentation: We continue to improve the documentation on
Architecture guides, How To’s, Tutorials, and APIs continue to be improved.
43 new public tutorials related and about MXNet have been published YTD.
- https://thomasdelteil.github.io/CNN_NLP_MXNet/
- https://gluon-crash-course.mxnet.io/
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/python/t
ypes_of_data_augmentation.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/python/d
ata_augmentation_with_masks.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/gluon/da
tasets.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/onnx/inf
erence_on_onnx_model.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/onnx/fin
e_tuning_gluon.md
-
https://aws.amazon.com/blogs/machine-learning/speeding-up-apache-mxnet-using-
the-nnpack-library/
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/python/d
ata_augmentation.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/gluon/da
ta_augmentation.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/gluon/na
ming.md
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/gluon/cu
stom_layer.md
- https://github.com/ThomasDelteil/VisualSearch_MXNet
-
https://github.com/apache/incubator-mxnet/blob/master/docs/tutorials/gluon/pr
etrained_models.md
- https://mxnet.incubator.apache.org/tutorials/gluon/save_load_params.html
- http://mxnet.incubator.apache.org/tutorials/python/profiler.html
-
How would you assess the podling's maturity?
Podling is still having difficulties to grow the contributor and
committer community. Maturity == Low to Medium.
Please feel free to add your own commentary.
[ ] Initial setup
[ ] Working towards first release
[X] Community building
[ ] Nearing graduation
[ ] Other
Date of last release: (latest to oldest releases)
1. Apache MXNet-incubating 1.2.0 (major release) was published on May
21 2018 - https://github.com/apache/incubator-mxnet/releases/tag/1.2.0
When were the last committers or PPMC members elected?
During the reporting period 1 contributor was elected as committer. The PPMC
elected Jim Jagielski as mentor.
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments:
[ ](mxnet) Suneel Marthi
Comments:
[ ](mxnet) Markus Weimer
Comments:
[ ](mxnet) Henri Yandell
Comments:
[X](mxnet) Jim Jagielski
Comments:
IPMC/Shepherd notes:
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way --
ONGOING.
2. Grow the community -- ONGOING.
3. Bring website up to Apache standard -- ONGOING
4. Identify remaining ICLAs or SGAs that need signing -- ONGOING
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be aware
of?
None
How has the community developed since the last report?
a) Various Slack channels, dev@ mailing lists, and user discussion forums
(http://discuss.mxnet.io) are being used actively. The contributors have been
working on having all discussions on the public dev@ mailing list as much as
possible. This is an ongoing improvement process where the focus will be to
reduce the scope of private discussions to only a few individuals before it is
put on the public dev@ mailing list so that the Apache MXNet community gets a
fair chance in influencing the final outcome/decision of the discussion.
b) O'Reilly published a series of blogs about MXNet:
https://www.oreilly.com/ideas/anomaly-detection-with-apache-mxnet
https://www.oreilly.com/ideas/logo-detection-using-apache-mxnet
https://www.oreilly.com/ideas/generative-model-using-apache-mxnet
https://www.oreilly.com/ideas/build-a-recurrent-neural-network-using-apache-mxnet
c) A new blog post published on 5-Feb about MXNet Model Server for Apache
MXNet introduces ONNX support and Amazon CloudWatch integration
https://aws.amazon.com/blogs/machine-learning/model-server-for-apache-mxnet-introduces-onnx-support-and-amazon-cloudwatch-integration/
d) A new blog post published on 13-Mar about MXNet - Deploy Gluon Models to
AWS DeepLens using a simple Python API
https://aws.amazon.com/blogs/machine-learning/deploy-gluon-models-to-aws-deeplens-using-a-simple-python-api/
e) Borealis AI (AI lab of Royal Bank of Canada) published a new blog post on
Feb about the comparison between PyTorch and MXNet – Standardizing a machine
learning framework for applied research – PyTorch vs MXNet
http://www.borealisai.com/2018/02/16/standardizing-a-machine-learning-framework-for-applied-research/
f) Microsoft AI research published a new blog post on Mar about the
performance comparison between multiple frameworks including MXNet – Comparing
deep learning frameworks: A Rosetta stone approach
https://blogs.technet.microsoft.com/machinelearning/2018/03/14/comparing-deep-learning-frameworks-a-rosetta-stone-approach/
g) Beeva Labs published a new blog post on Feb about MXNet – Accelerating the
training of deep neural networks with MXNet on AWS P3 instances
https://www.bbva.com/en/accelerate-training-deep-neural-networks-mxnet-aws-p3-instances/
h) Webinars and talks on Apache MXNet on the social media platform:
https://www.youtube.com/watch?v=DSNvm29kIAo
https://www.youtube.com/watch?v=kINQpQiee7g
https://www.youtube.com/watch?v=kWgCbjtsAAM
https://www.youtube.com/watch?v=YjDTL_Kjfww
i) Thomas Delteil spoke at Meetup in Vancouver -
https://www.meetup.com/LearnDataScience/events/248473200/?rv=ea1&_xtd=gatlbWFpbF9jbGlja9oAJDA0ODlhZGJjLWQ0YzYtNDJmOC04MDFiLTZiMWU1OWEwOWEwYw
- and talked about Convolutional Neural Networks for NLP
j) A public meeting is planned for April 24th in Seattle -
https://cwiki.apache.org/confluence/display/MXNET/Seattle
k) MXNet 1.2 release is in preparation and release notes are being reviewed at
https://cwiki.apache.org/confluence/display/MXNET/%5BWIP%5D+Apache+MXNet+%28incubating%29+1.2.0+Release+Notes
l) Community voted and adopted Jira for issue management -
https://issues.apache.org/jira/projects/MXNET/issues/MXNET-267?filter=allopenissues
m) Established MXNet youtube channel -
https://www.youtube.com/channel/UCQua2ZAkbr_Shsgfk1LCy6A. Please subscribe!
n) MXNet 1.1 was released on 19-Feb, 2018 with extensive support and help from
various community members and timely guidance from the Apache MXNet Mentors.
How has the project developed since the last report?
a. The community released MXNet 1.1 that makes MXNet more stable and
production ready: https://blogs.apache.org/mxnet/entry/1-1-0-release-makes
b. From a statistics perspective, based on the Github insights, found here:
https://github.com/apache/incubator-mxnet/pulse/monthly, in Mar 2018, 60
authors pushed 164 commits to master, with updates to 624 files including
23.1K additions and 5.6K deletions. Historically, in Dec 2017, 51 authors
pushed 115 commits to master, with updates to 489 files including 13K
additions and 9K deletions. Historically, in Sep 2017, 62 authors pushed 171
commits to master, with updates to 467 files including 26K additions and 7K
deletions. We are working on finding more contributors to the project.
c. Documentation- Architecture guides, How To's, Tutorials, and APIs continue
to be improved.
d. More advanced features (e.g. ONNX support, mixed precission modeling, Scala
inference API, MKL-DNN integratino etc) and bug-fixes requested by the user
community continue to be added.
How would you assess the podling's maturity?
Podling is still being established in Apache - hence maturity == Low.
[ ] Initial setup
[ ] Working towards first release
[X] Community building
[ ] Nearing graduation
[ ] Other
Date of last release: (latest to oldest releases) A major release Apache
MXNet-incubating 1.1.0 was released on 19-Feb, 2018.
https://github.com/apache/incubator-mxnet/releases/tag/1.1.0
When were the last committers or PPMC members elected?
Jan 18 2018 - Jun Wu and Marco de Abreu elected as committer and PPMC member
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments:
[ ](mxnet) Suneel Marthi
Comments:
[X](mxnet) Henri Yandell
Comments: I have noted on list that the comment on
reducing private conversation needs to go much further. The sentiment in
item (a) does not reach Apache's minimum bar and 'fair chance' is
insulting.
[ ](mxnet) Markus Weimer
Comments:
Apache MXNet is an open-source, scalable, distributed and high-performance
deep learning framework that allows you to define, train, and deploy deep
neural networks on a wide array of devices, from cloud infrastructure to
mobile devices. It is highly scalable, allowing for fast model training, and
supports a flexible programming model and multiple languages. Apache MXNet
allows you to mix symbolic and imperative programming flavors to maximize both
efficiency and productivity. Apache MXNet is built on a dynamic dependency
scheduler that automatically parallelizes both symbolic and imperative
operations on the fly. A graph optimization layer on top of that makes
symbolic execution fast and memory efficient. The Apache MXNet library is
portable and lightweight, and it scales to multiple GPUs and multiple
machines.
Apache MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way --
ONGOING. 2. Grow the community -- ONGOING. 3. Bring website up to Apache
standard -- ONGOING 4. Identify remaining ICLAs or SGAs that need signing --
ONGOING
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be aware
of?
Request the Incubator PMC to add more mentors to the project, preferably
mentors employed with Open Source companies like Hortonworks that have prior
experience in handling multiple open source projects that were part of an
ecosystem.
How has the community developed since the last report?
a) Various Slack channels, dev@ mailing lists, and user discussion forums
(http://discuss.mxnet.io) are being used actively. The contributors have been
working on having all discussions on the public dev@ mailing list as much as
possible. If some discussions happen in private, they are eventually brought
out on dev@ with all perspectives well-represented. This is an ongoing
improvement process where the discussion will be put on the public dev@
mailing list so that the Apache MXNet community gets a fair chance in
influencing the final outcome/decision of the discussion.
b) O’Reilly published a series of blogs about MXNet, including ones with deep
matrix factorization using Apache MXNet:
https://www.oreilly.com/ideas/sentiment-analysis-with-apache-mxnet
https://www.oreilly.com/ideas/deep-matrix-factorization-using-apache-mxnet
https://www.oreilly.com/ideas/apache-mxnet-in-the-wolfram-language
c) A blog post published on 25-Oct about MXNet – an open source binary neural
network implementation based on MXNet:
https://aws.amazon.com/blogs/ai/research-spotlig ht-bmxnet-an-open-source-binary-neural-network-implementation-based-on-mxnet/
d) A blog post published on 01-Nov about the availability of Nvidia Volta GPU
support and Sparse Tensor support:
https://aws.amazon.com/blogs/ai/a pache-mxnet-release-adds-support-for-new-nvidia-volta-gpus-and-sparse-tensor/
e) A new blog post published on 08-Nov showing MXNet 0.12 extends Gluon
Functionality:
https://aws.amazon.com/blogs/ai/apache-mxn et-version-0-12-extends-gluon-functionality-to-support-cutting-edge-research/
f) A blog post published on 08-Nov introducing Model Server for MXNet:
https://aws.amazon.com/blogs/ai/introducing-model-server-for-apache-mxnet/
g) A blog post published on 7-Nov demonstrating performance and scalability
of MXNet:
h ttps://techburst.io/mxnet-the-real-world-deep-learning-framework-2690e56ef81f
h) Members of the community have conducted open meetups to share information
on Apache MXNet: https://www.meetup.com/Apache-MXNet-learning-group/
i) Talks on Apache MXNet have been held in various universities and
conferences across the world including US, China, etc.:
https://www.youtube.com/watch?v=me1qOzSg8MU
https://www.youtube.com/watch?v=9IrvDHRQaaA
https://www.youtube.com/watch?v=4PbSZRYXa3o
https://www.youtube.com/watch?v=RRy-3VXA0nw
j) MXNet 1.0 was released on 04-Dec, 2017 with extensive support and help
from various community members and timely guidance from the Apache MXNet
Mentors.
How has the project developed since the last report?
a. The community released MXNet 1.0 that is production ready, simplifies deep
learning experience, and significantly improves performance with cutting-edge
features described here:
https://blogs.apache.org/mxnet/entry/milestone-v1-0-release-for
b. Documentation- Architecture guides, How To’s, Tutorials, and APIs continue
to be improved.
c. Support for Perl language bindings - contributed by Sergey Kolychev.
d. More advanced features (e.g. sparse tensor, advanced indexing, gradient
compression) and bug-fixes requested by the user community continue to be
added.
e. Community took complete end-to-end ownership of the continuous integration
process in order to enable reliable testing on a wide set of back ends (IoT
devices to GPU clusters).
How would you assess the podling's maturity?
Podling is still being established in Apache and no efforts being made in
increasing community - hence maturity == Low.
Please feel free to add your own commentary.
[ ] Initial setup
[ ] Working towards first release
[X] Community building
[ ] Nearing graduation
[ ] Other:
Date of last release:
2017-12-04
When were the last committers or PPMC members elected?
Sergey Kolychev was elected as a committer and PPMC member in October 2017 for
contributing the Perl language bindings. There is a plan to convert more
contributors into committers in early 2018.
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments:
[X](mxnet) Suneel Marthi
Comments: 1. All design decisions and project
roadmap seem to be done internally at team huddles and rarely the community
is ever involved in making any decisions. 2. There is no effort being made
to discuss roadmap or project issues on public mail lists. 3. All roadmap
planning seems to happen internally. 4. No effort is being made to build a
diverse community around the project. 5. Most committers (all of whom are
employed by a single vendor) appear to be resistant in moving to using
apache tools such as JIRA or adopting the Apache Way of growing diverse
community.
[X](mxnet) Henri Yandell
Comments:
[X](mxnet) Markus Weimer
Comments:
Apache MXNet is an open-source deep learning framework that allows you to define, train, and deploy deep neural networks on a wide array of devices, from cloud infrastructure to mobile devices. It is highly scalable, allowing for fast model training, and supports a flexible programming model and multiple languages. MXNet allows you to mix symbolic and imperative programming flavors to maximize both efficiency and productivity. MXNet is built on a dynamic dependency scheduler that automatically parallelizes both symbolic and imperative operations on the fly. A graph optimization layer on top of that makes symbolic execution fast and memory efficient. The MXNet library is portable and lightweight, and it scales to multiple GPUs and multiple machines.
Apache MXNet has been incubating since 2017-01-23.
The most important issues to address in the move towards graduation:
1. Establish a predictable release process consistent with Apache Way -- Ongoing.
2. Grow the Community - Ongoing
3. Update project website to Apache standards - In Progress
4. Identify remaining ICLAs and SGAs that need to be addressed - in progress
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?
None
How has the community developed since the last report?
a) Various Slack channels and dev@ mailing lists are being used actively. New user support and discussion forum created with the guidance of Apache members at: http://discuss.mxnet.io
b) O’Reilly published a series of blogs about MXNet, including ones with step-by-step instructions to implement a convolutional neural network to classify traffic signs with Apache MXNet:
https://www.oreilly.com/ideas/classifying-traffic-signs-with-mxnet-an-introduction-to-computer-vision-with-neural-networks
https://www.oreilly.com/ideas/apache-mxnetthe-fruit-of-cross-institutional-collaboration
https://www.oreilly.com/ideas/self-driving-trucks-enter-the-fast-lane-using-deep-learning
c) A new blog post published on 28-July showing users how to exploit the unique features of Apache MXNet with a cheat sheet: https://aws.amazon.com/blogs/ai/exploiting-the-unique-features-of-the-apache-mxnet-deep-learning-framework-with-a-cheat-sheet/
d) New Blog series by Viacheslav Kovalevskyi on Apache MxNet in Depth published on Medium https://blog.kovalevskyi.com/mxnet-distributed-training-explained-in-depth-part-1-b90c84bda725
e) Members of the community have conducted open meetups to share information on Apache MXNet: https://www.meetup.com/Apache-MXNet-learning-group/
f) Talks on Apache MXNet have been held in various universities and conferences across the world including US, China, etc.:
https://www.youtube.com/watch?v=GBkOMtc9BIk
https://www.youtube.com/watch?v=kGktiYF5upk
g) Presently working towards an upcoming 0.12.0 release targeted for October or November 2017.
How has the project developed since the last report?
a) The code base was migrated from http://github.com/dmlc/mxnet to https://github.com/apache/incubator-mxnet on 28-July, 2017. The website has also been migrated to this repository.
b) From a statistics perspective, in July 2017, 54 authors pushed 140 commits to master, with updates to 358 files including 22K additions and 3K deletions. In Sep 2017, 62 authors pushed 171 commits to master, with updates to 467 files including 26K additions and 7K deletions.
c) Documentation- Architecture guides, How To’s, Tutorials, and APIs continue to be improved.
d) More features (e.g. operators, algorithms) and bug-fixes requested by the user community continue to be added.
How would you assess the podling's maturity?
Podling is still being established in Apache - hence maturity == Low, but the project has a very diverse set of contributors.
Please feel free to add your own commentary.
[ ] Initial setup
[ ] Working towards first release
[X] Community building
[ ] Nearing graduation
[ ] Other:
Date of last release:
A maintenance release Apache MXNet-incubating 0.11.0 with few bug-fixes was released on 05-Sep, 2017.
https://github.com/apache/incubator-mxnet/releases/tag/0.11.0
A maintenance release Apache MXNet-incubating 0.10.0 Post 2 with few bug-fixes was released on 17-July, 2017.
https://github.com/apache/incubator-mxnet/releases/tag/0.10.0.post2
When were the last committers or PPMC members elected?
Ly Nguyen, Haibin Lin and Madan Jampani were added as committers and PPMC members in June 2017.
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments: Second Suneel's comment that the project should work towards more communication on its mailinglists.
[X](mxnet) Suneel Marthi
Comments: Plenty of activity on the project, would love to see more discussions happening on mail lists or Slack.
[ ](mxnet) Markus Weimer
Comments:
[X](mxnet) Henri Yandell
Comments: Noting acknowledgement that Infra JIRA requests will be going via mentors due to both the reported
volume of requests and the requests coming from contributors and not committers. Also noting that
I've been active on legal-discuss@ regarding MXNet ICLA/source headers, and in discussion offline
with an employer of contributors who has concerns with ICLA signing.
IPMC/Shepherd notes:
johndament: The podling has come under fire recently for what is perceived to be an over abundant amount of requests coming from non-committers. If nothing else, it's a sign that they are anxious to get going here at Apache and perhaps they need to look at their communication model, moving more discussion on list, as well as potentially voting in new committers.
MXNet is an open-source deep learning framework that allows you to define,
train, and deploy deep neural networks on a wide array of devices, from
cloud infrastructure to mobile devices. It is highly scalable, allowing for
fast model training, and supports a flexible programming model and multiple
languages. MXNet allows you to mix symbolic and imperative programming
flavors to maximize both efficiency and productivity. MXNet is built on a
dynamic dependency scheduler that automatically parallelizes both symbolic
and imperative operations on the fly. A graph optimization layer on top of
that makes symbolic execution fast and memory efficient. The MXNet library
is portable and lightweight, and it scales to multiple GPUs and multiple
machines.
MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Migrate code (GitHub) and website to Apache Infra.
2. Grow the community
3. Establish a reliable Release process consistent with Apache Way.
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?
None
How has the community developed since the last report?
a) On 5/27 MXNet published a comprehensive edit and makeover of the
documentation including tutorials, how-to’s, APIs and architecture guides.
This was a broad effort that involved over 40 contributors.
b) The PMC voted in a new committer who has been helping with the code
migration and setup of the test infrastructure. We are making slow but
steady progress towards getting the GitHub code migrated. The target date
for migration is 7/17. Website migration will happen after that.
c) Slack and dev@ are being used more actively.
d) Two presentations/workshops on Apache MXNet at the O’Reilly AI Conf on 6/27 and 6/28
e) A new blog post published on 6/23 showing users how to Build a
Real-time Object Classification System with ApacheMXNet on Raspberry Pi.
https://aws.amazon.com/blogs/ai/build-a-real-time-object-classification-system-with-apache-mxnet-on-raspberry-pi/
How has the project developed since the last report?
a) Since the last report 42 authors have pushed 326 commits to master.
b) Documentation- Architecture guides, How To’s, Tutorials, and APIs have been improved.
c) More features (e.g. operators) requested by the user community have been added.
d) A new Perl language binding for MXNet was added.
How would you assess the podling's maturity?
Podling's still getting established in Apache - so maturity == Low.
Please feel free to add your own commentary.
[X] Initial setup
[ ] Working towards first release
[ ] Community building
[ ] Nearing graduation
[ ] Other:
Date of last release:
No Release yet, project is still getting established in Apache.
When were the last committers or PPMC members elected?
Ly Nguyen added as a committer and PPMC member in June 2017.
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments:
[X](mxnet) Suneel Marthi
Comments:
[X](mxnet) Markus Weimer
Comments:
[X](mxnet) Henri Yandell
Comments:
MXNet is an open-source deep learning framework that allows you to define,
train, and deploy deep neural networks on a wide array of devices, from
cloud infrastructure to mobile devices. It is highly scalable, allowing for
fast model training, and supports a flexible programming model and multiple
languages. MXNet allows you to mix symbolic and imperative programming
flavors to maximize both efficiency and productivity. MXNet is built on a
dynamic dependency scheduler that automatically parallelizes both symbolic
and imperative operations on the fly. A graph optimization layer on top of
that makes symbolic execution fast and memory efficient. The MXNet library
is portable and lightweight, and it scales to multiple GPUs and multiple
machines.
* A list of the three most important issues to address in the move
towards graduation.
1. Migrate code(GitHub) and website to Apache.
2. Grow the community:
2.1 Improving documentation including APIs & Tutorials.
2.2 Improving user-experience, for example improved error messages.
3 Establish a dependable, Apache-way consistent release process.
3.1 Features: New operators requested by user community.
Accelerate performance on CPUs and GPUs.
* Any issues that the Incubator PMC or ASF Board might wish/need to be
aware of:
Code and website have taken time to get moving on. The plan is to discuss
these with Apache Infra at ApacheCon and then get the code migrated in May.
* How has the community developed since the last report.
1. The community was engaged for contributions to API documentation and
tutorials.
2. Slack channels have been created for the community to contribute
discussions to (though they need to move to the ASF Slack channel that
was recently created).
3. In the last month, excluding merges, 51 authors have pushed 165 commits
to master and 180 commits to all branches. On master, 502 files have
changed and there have been 26,246 additions and 12,188 deletions. Count of
Closed Issues = 62, Count of New Issues = 146, Count of Merged Pull
Requests = 161, Count of Proposed Pull Requests = 27.
* How has the project developed since the last report.
On GitHub:
1. The API Documentation has improved.
2. More features (e.g. operators) requested by the user community has been added.
3. Hardware acceleration like cuDDN6 integration and MKL ML package
integration was completed.
4. A new Perl language binding for MXNet was added.
5. Apache MxNet talk at Apache BigData North America on May 18, 2017
* How does the podling rate their own maturity.
Maturity = Low. Initial setup and adjusting community to Apache-style interactions.
Date of last release:
No @Apache release yet.
When were the last committers or PPMC members elected?
No new committers added yet.
Signed-off-by:
[ ](mxnet) Sebastian Schelter
Comments:
[X](mxnet) Suneel Marthi
Comments:
[ ](mxnet) Markus Weimer
Comments:
[X](mxnet) Henri Yandell
Comments: Starting to see movement on moving things to Apache.
The dev@ list is beginning to see traffic.
Signed-off-by:
[ ](mxnet) Sebastian Schelter
Comments:
[ ](mxnet) Suneel Marthi
Comments:
Its been 4 months since this project has been proposed for Apache
Incubator and so far nothing's been done to move the project to Apache.
All activity is still happening on the original project github -
http://github.com/dmlc/mxnet and all conversations still happen on the
project's Gitter channel. Since the last report from March 2017 (which
btw was drafted and filed by me) there's been ZERO traction on moving
the project to Apache. There are upcoming talks in various confs like -
https://conferences.oreilly.com/artificial-intelligence/ai-ny/public/schedule/detail/58186
that make no reference to the fact that the project is now 'Apache'.
The reason me and Hen have not filed a report yet for April 2017 is due
to the fact we would rather one of the committers on the project took
the initiative to do it as opposed to a mentor covering for the
project.
[ ](mxnet) Markus Weimer
Comments:
[ ](mxnet) Henri Yandell
Comments:
IPMC/Shepherd notes:
Justin Mclean:
No report yet. Slow to start up but at least one Mentor active.
A Flexible and Efficient Library for Deep Learning
MXNet has been incubating since 2017-01-23.
Three most important issues to address in the move towards graduation:
1. Move the code and website to Apache Infrastructure.
2. Establish a formal release process and schedule, allowing for dependable
release cycles in line with Apache development process.
3.
Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be
aware of?
PPMC to discuss adding in some community members who asked to join while
the Incubator vote was in progress.
How has the community developed since the last report?
Project is still getting set up in Incubator.
How has the project developed since the last report?
Project is still getting set up, all proposed committers have submitted
their ICLAs.
How would you assess the podling's maturity?
Please feel free to add your own commentary.
[X] Initial setup
[ ] Working towards first release
[ ] Community building
[ ] Nearing graduation
[ ] Other:
Date of last release:
No Release yet
When were the last committers or PPMC members elected?
Project is being set up with the initial set of committers.
Signed-off-by:
[X](mxnet) Sebastian Schelter
Comments:
[X](mxnet) Suneel Marthi
Comments:
[X](mxnet) Markus Weimer
Comments:
[X](mxnet) Henri Yandell
Comments:
As project already exists in the public, this incubation is about
moving the development over without stopping the momentum of the
project, and then learning about the Apache development processes.
Previous conversations were Slack/GitHub-issue based, so making
decisions on the email list will be the first likely adaptation. A
dependency on ZeroMQ will be the primary discussion point for a first
release at Apache. PPMC list is still low on subscriptions, with
less than half of the PPMC members subscribed.
MXNet is an open-source deep learning framework that allows you to define, train, and deploy deep neural networks on a wide array of devices, from cloud infrastructure to mobile devices. It is highly scalable, allowing for fast model training, and supports a flexible programming model and multiple languages. MXNet has been incubating since 2017-01-23. Three most important issues to address in the move towards graduation: 1. Getting ICLAs signed/private@ populated. 2. Sharing documentation/mentor thoughts on being an Apache committer. 3. Migrating the code/issues over to Apache. Any issues that the Incubator PMC (IPMC) or ASF Board wish/need to be aware of? None How has the community developed since the last report? 1. This is our first report. 2. The mailing lists are created; with the dev@ list being well populated. 3. ICLAs are being signed. How has the project developed since the last report? 1. Trademark review seems complete. 2. A vendor, AWS, worked with Sally on a blog posting[1]. Date of last release: No Apache release yet; still being setup. When were the last committers or PPMC members elected? At Proposal time. We have a few folk who asked to join the incubation proposal after the vote started and will be discussed once private@ is populated. [1] - https://aws.amazon.com/blogs/aws/excited-about-mxnet-joining-apache/ Signed-off-by: [X](mxnet) Henri Yandell [X](mxnet) Markus Weimer [X](mxnet) Sebastian Schelter [X](mxnet) Suneel Marthi Shepherd/Mentor notes: John D. Ament: I find it a bit odd that the linked article uses "MXNet" repeatedly but only refers to it as "Apache MXNet" once near the end of the page.