►
From YouTube: IncEng MLOps Update - 2022-08-12
Description
Kickstarting ML Experiment Tracking
ML Experiment Tracking Epic: https://gitlab.com/groups/gitlab-org/-/epics/8560
This update: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/60
All Updates: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/16
A
Hello,
everyone
and
welcome
to
another
update
for
incubation
engineer
mlaps
at
gitlab.
My
name
is
eduardo
and
today
is
august
8th
august
12th,
and
this
is
the
first
time
I'm
not
going
to
talk
about
jupiter
notebooks,
because
now
we're
moving
on
with
machine
learning.
A
Experiment
tracking
we've
been
talking
about
this
for
a
while,
under
the
guise
of
let's
create
a
let's
deploy
more
flow
along
gitlab,
but
I've
been
we've
been
chatting
around
the
technical
side
and
how
to
learn
this
and
we've
decided
to
implement
machine
learning,
experiment
tracking
within
gitlab
itself,
reasons
being
it's
better
to
integrate
across
the
ecosystem.
A
The
unique
value
proposition
here
is
that
this
is
not
just
a
tool
that
acts
on
its
own.
This
is
supposed
to
be
a
tool
that
integrates
across
the
entire
gitlab
so,
for
example,
machine
learning
the
experiment
tracking
can
integrate
to
merge
requests.
When
you
create
a
merge
request
that
contains
a
model,
it
already
creates
the
experiment
tracking.
For
that
specs.
From
tracking
I
mean
you
train
a
lot
of
different
models
which
we'll
call
candidates
and
with
different
parameters,
and
then
you
can
compare
the
results
for
each
of
them.
A
So
we
already
started
working
on
this.
The
first,
mr,
is
already
being
reviewed,
and
we
expect
to
have
the
mvp
by
15.5,
the
15.5.
There
we
defined
the
mvp
as
having
users
being
able
to
log
the
experiments,
log,
the
candidates,
log,
the
parameters
and
metrics
into
gitlab,
but
furthermore,.