►
From YouTube: IncEng MLOps Update - 2022-11-07
Description
GitLab ML Experiment Tracking SSoT: https://gitlab.com/groups/gitlab-org/-/epics/8560
Feedback Issue: https://gitlab.com/gitlab-org/gitlab/-/issues/381660
All Updates: https://gitlab.com/gitlab-org/incubation-engineering/meta/-/issues/16
A
Today
we're
going
to
talk
a
little
bit
more
about
machine
learning,
experience
tracking
following
up
on
what
we're
doing
so
far
for
those
who
are
new
here,
machine
learning,
experience
tracking
is
a
way
to
organize
all
the
training
and
the
metadata
that
you
collect
for
your
models
and
we
are
trying
to
figure
out
if
it
makes
sense
to
have
this
at
gitlab
as
part
of
gitlab
platform.
So
the
current
update
is
that
now
the
biggest
one
is
that
we
can
also
hold
the
artifacts
for
a
for
a
candidate.
A
So
when
you
run
your
machine
learning
training
in
addition
to
collect
your
metadata
that
you
have
over
here,
like
all
the
parameters
and
the
metrics,
you
can
also
collect
the
artifacts
with
gitlab.
Again.
This
is
zero
set
up
by
the
by
the
data
scientist
or
by
platform
engineer.
It
already
uses
gitlab
itself
for
everything,
so
it
just
works.
There's
no
setup
needed
all
working
through
the
integration
with
ML
flow.
A
We're
also
now
adding
details
over
each
candidate,
so
just
the
screen
for
you
to
check
all
the
possibilities.
This
is
very
ugly,
but
I
will
be
iterating
over
this,
but
the
most
important
part
is
that
tomorrow
or
on
November
8th,
we
will
be
starting
internal
testing,
so
our
teams
are
gitlab
will
start
testing,
but
this
integration
we
already
had
a
conversation
with
customers
that
already
want
to
use
this
on
the
current
state,
so
on
15.6
I'll
be
working
on
on
fixing
a
little
bit
or
making
it
more
stable.
A
But
on
Fifth
with
the
functionality
shipped
at
15.6,
which
is
in
two
weeks,
we
will
start
close
tests
with
with
some
customers.
So
that's
very
exciting.
I'm
really
hopeful
about
the
feedback
that
we
will
get
from
this
and
how
the
tool
will
move,
and
this
feature
of
overtime
have
a
good
one.