Skip to content

Commit 0316251

Browse files
authored
Merge pull request #314 from jaeyounkim/master
RFC: Creating SIG Models
2 parents d1b3f7a + 0c84bd8 commit 0316251

File tree

1 file changed

+91
-0
lines changed

1 file changed

+91
-0
lines changed

rfcs/20201023-sig-models.md

Lines changed: 91 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,91 @@
1+
# Creating SIG Models
2+
3+
| Status | Proposed |
4+
:-------------- |:---------------------------------------------------- |
5+
| **RFC #** | 314 |
6+
| **Author(s)** | Jaeyoun Kim (jaeyounkim@google.com), Jing Li (jingli@google.com), Mike Liang (mliang@google.com), Shuangfeng Li (shuangfeng@google.com) |
7+
| **Sponsor** | Thea Lamkin (thealamkin@google.com) |
8+
| **Updated** | 2020-10-30 |
9+
10+
## What is this group for?
11+
12+
This group is for discussions and collaborations on enabling community
13+
contributions to [TensorFlow Model Garden](https://github.com/tensorflow/models)
14+
and [Tensorflow Hub](https://github.com/tensorflow/hub).
15+
16+
SIG Models will focus on empowering the community to contribute state-of-the-art
17+
model implementation in TensorFlow 2. It will benefit the whole community by
18+
providing recommended implementations and models with reproducible results.
19+
20+
SIG Models will have several subgroups (e.g., SIG Models Vision and SIG Models
21+
NLP) covering different machine learning areas. There are several SIG leads for
22+
each group to coordinate the contributions, run community events like contests,
23+
and maintain the code quality through the review process. Each group has the
24+
flexibility to operate differently.
25+
26+
SIG Models is also a place for community discussions and sharing best practices of
27+
using TensorFlow 2 for state-of-the-art research. Furthermore, SIG Models provides
28+
product feedback to help TensorFlow to be improved.
29+
30+
## Who will be part of it?
31+
32+
SIG Models are open, membership will be entirely public, and anybody interested
33+
in model contributions or participating in the discussion can join the SIG.
34+
There can be different ways to participate:
35+
36+
* Everyone in the community can pick GitHub model tasks and contribute models.
37+
* They can also join the regular SIG meetings or email groups to participate in
38+
the discussions. To participate, request an invitation to join the SIG mailing
39+
list. Archives of the mailing list are publicly accessible.
40+
41+
We will identify SIG leads from the community to run the SIG Models groups.
42+
Initially, we plan to set up the SIG Models Vision group:
43+
44+
* SIG Models Vision leads
45+
* George Thiruvathukal, gkt@cs.luc.edu (Loyola University Chicago)
46+
* Yung-Hsiang Lu, yunglu@purdue.edu (Purdue University)
47+
* Co-leads and sponsors from TensorFlow
48+
* Jaeyoun Kim, jaeyounkim@google.com (TensorFlow Model Garden)
49+
* Jing Li, jingli@google.com (TensorFlow Model Garden)
50+
* Mike Liang, mliang@google.com (TensorFlow Hub)
51+
* Shuangfeng Li, shuangfeng@google.com (TensorFlow Lite)
52+
53+
## What initial problems will the group tackle?
54+
55+
We will start with the SIG Models Vision group. Initially, the group leads will
56+
run a contest to motivate the community to contribute state-of-the-art computer
57+
vision models using TensorFlow 2.
58+
59+
For example:
60+
61+
* Define/release model implementation tasks, rules, and evaluation criteria for
62+
the contest
63+
* Review initial submissions and provide more support for selected high-quality
64+
submissions
65+
* Review final submissions and announce results
66+
67+
## What modes of communication do you intend to use?
68+
69+
We plan to have a mailing list
70+
([models@tensorflow.org](https://groups.google.com/a/tensorflow.org/forum/#!forum/models)),
71+
arrange a regular (e.g., monthly) video conference call rotated by the SIG
72+
leads. We will also use a Gitter chat channel
73+
(https://gitter.im/tensorflow/sig-models) for discussions.
74+
75+
As we have multiple SIG subgroups, each subgroup may run meetings and events
76+
using different modes of communication.
77+
78+
## Launch plan
79+
80+
* Video conference calls with initial interested parties to finalize the charter
81+
* SIG set up with initial group members
82+
* SIG added to community pages on tensorflow.org
83+
* SIG leads and sponsors start discussions about initial work items
84+
* Write a blog post about the SIG Models with the initial achievements and
85+
welcome more members
86+
87+
## Code of Conduct
88+
89+
As with all forums and spaces related to TensorFlow, SIG Models is subject to
90+
the [TensorFlow Code of
91+
Conduct](https://github.com/tensorflow/tensorflow/blob/master/CODE_OF_CONDUCT.md).

0 commit comments

Comments
 (0)