►
From YouTube: IncEng MLops Update - 2022-04-22
Description
Usage, progress and Backlog update
All Updates: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/16
This Update: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/52
A
Hello,
everyone
and
welcome
to
another
update
for
incubation
engineer
for
emerald
ops
here
at
gitlab.
Another
quick
update
today.
So,
let's
start
with
usage
here
we
see
a
the
evolution
of
comments
on
python,
notebooks,
jupiter,
notebooks
versus
all
files.
We
clearly
see
a
huge
growth
by
the
by
the
beginning
of
this
year,
which
it's
not
really
it's
hard
to
draw
causal
relationships,
but
it's
heavily
correlated
with
the
release
of
14.5
that
happened
on
november
and
that's
when
we
released
the
cleaner,
jupiter
notebook
diffs.
A
So
that's
a
very
reassuring
data
point
that
you
have
that
users
is
actually
wanting
this.
This
is
actually
creating
a
new
behavior
on
users
and
allowing
for
code
reviews
and
discussions
on
jupyter
notebooks
next
up
progress.
So
on
the
last
few
weeks
we
were
busy
releasing
a
multitude
of
performance
improvements
of
fixing,
some
bugs
with
the
with
the
toggle
between
raw
and
semantic
tips
and
with
the
new
algorithm
that
creates
the
dips,
but
that
is
done.
A
A
So
it's
a
process
that
we
might
take
a
little
bit
more
time,
but
but
it's
moving
and
it's
it's
going
well
and
finally,
we
I
I'm
trying
a
new
way
of
setting
up
my
backlog
of
ideas,
so
we
use
rice
framework
for
identifying
opportunities
for
incubation
engineering,
and
I
thought
it
would
be
nice
as
well
to
have
this
for
mlabs
incubation
engineering.
Specifically.
So
I
I
listed
my
ideas.
I
gathered
feedback
from
the
community.
A
I
get
feedback,
I
gathered
feedback
internally
as
well,
and
I
came
up
with
this
rice
plus
kind
of
which
I
add,
collaboration
and
mlaps.
Branding
to
the
computation
and
the
clear
winner
so
far
is
an
integration
between
gitlab
and
overflow
and
the
flow
is
a
model
registry
open
source
from
databricks
maintained
by
databricks,
and
it's
a
good
way
to
explore
how
we
how
it
would
like
the
relationship
between
git
lab
and
not
only
ammo
flow,
but
modern
registries
in
general.
A
How
can
we
surface
that
information
in
which
steps
of
the
devops
life
cycle
does
that
information
make
sense
and
the
way
to
go
over
there?
Some
other
ideas
that
we
have
there?
It's
jupiter
lab
it's
a
gitlab
blogging
for
jupiter,
lab
runners
for
ml
use
cases,
the
analysis,
repository
and
improving
pipeline
usage
for
email
use
cases.
A
User
feedback
is
really
important.
So
if
you're
watching
this,
I
am
listening
to
you
and
yeah.
So
if
you
have
any
ideas
you
can
go
to
to
the
epic
and
just
leave
it
there
and
then
I
will
create
an
issue
for
that
and
explore
the
rice,
which
is
stands
for
rich
impact,
confidence
and
effort,
and
that's
what
I
had
for
today,
great
chatting
with
you.