►
From YouTube: Deep Learning Cloud with Filecoin
Description
Join us for Filecoin Liftoff Week, an action-packed series of talks, workshops, and panels curated by the web3 community to celebrate the Filecoin mainnet launch and chart the network’s future. https://liftoff.filecoin.io/
Events take place all week, October 19-23, 2020.
For more information on Filecoin
- visit the project website: https://filecoin.io/
- or follow Filecoin on Twitter: https://twitter.com/Filecoin
Get Filecoin community news and announcements in your inbox, monthly: http://eepurl.com/gbfn1n
A
A
This
is
charles
from
mbfs,
so
I'm
charles,
the
founder
of
mbs
canada
and
today
I
would
like
to
bring
one
the
deep
learning
club
with
firecoin.
So
first
of
all
about
us,
so
mfs
canada,
the
company
name,
is
nebula.
Aic
is
a
cloud-based
solution,
provider
of
cpu
container
technology
headquarters
in
montreal
canada.
A
So
we
build
a
platform
called
rncap
iphone,
which
is
a
international
age
computing
platform
which
is
using
different
technologies
to
manage
the
data
center
gpu
cloud
computing
restoration
system
currently
like
previously
we're
using
the
azure
and
aws
s3
for
saving
the
permanent
data.
But
during
the
space
with
one
and
spreadsheets
two,
we
decided
to
integrate
with
frye
coin,
to
provide
a.
A
Efficiency,
the
solution,
storage
solution
and
the
project
currently
is
under
canada.
Next
generation
network
support,
so
we
first
of
all
do
a
high-level
platform
design.
A
We
have
a
company
doing
design
investigation,
industrial
labor
applications,
we
previously
purely
some
decentralized
air
application
where
it
has
a
pass
platform
for
people
access,
different
apis
and
they
decided
working
with
the
block
freight
com
blockchain
to
get
the
storage
solutions
and
for
other
infrastructure.
It
includes
the
data
center
different
gpus
for
doing
cpu
calculation,
because
we
found
that
it's
a
great
fit
to
using
gpu,
both
in
through
in
storage
proof
in
this
paracord
under
the
gpu
for
deep
learning.
So.
B
A
A
purpose
design
we
had,
we
built
a
very
complicated
system,
it
can
be
for
applications,
it
can
be
trading
platform.
A
A
So,
when
we
integrated
this
platform,
we
already
supported
the
pi
gate,
which
is
developed
by
another
team
in
vancouver.
Canada-
sorry,
not
a
yes
package
and
we
support
a
container-based
database.
So
if
you
want
to
have
a
local
database
running
that
platform,
you
don't
need
to
install
by
ourself.
We
already
integrated.
You
just
need
to
enter
the
image
name
and
it
will
automatically
build
the
container
for
you,
and
it
also
give
you
the
external
access
of
the
dns.
We
solve
that
part
as
well.
A
We
provide
the
one
key
turn
on
gpu
accelerator
choice,
so
just
like
a
google
collab,
you
don't
need
to
install
the
nvidia
driver
by
yourself.
You
just
need
to
click
one
button
and
the
gpu
is
enabled
for
you
usage.
We
also
provide
a
multi-data
center
support.
So
when
you
have
not
enough
gpu
in
one
data
center,
you
can
immediately
transfer
to
another
data
center
without
your
conscience,
we're
embedded
as
a
power
get
engine.
A
A
A
Here
I
would
just
like
to
see
a
video
about
the.
This
is
our
interface.
You
can
see
here.
We
have
mbfs
here.
This
supports
the
upload
file,
task,
screen
history
and
the
public
access.
So
just
go
here
see
so
you
can
get
your
file
from
look
system
and
you
click
submit
a
few
moments.
You
will
see
that
it
says
successfully
uploaded
to
the
system
and
give
you
the
give
you
the
previous
say
id,
which
you
will
re
use
it
to
retrieve
your
data.
A
The
upload
maybe
take
a
couple
hours
due
to
the
selling
process.
Yeah.
After
that
you
can
see
here,
you
can
go
to
the
public
data
set,
which
we've
already
pre-loaded
lots
of
data
set
of
natural
and
assigns.
So
if
you
want
to
use
those
datasets,
it's
for
public
access,
everybody
can
use
it.
You
don't
need
to
upload
those
terabytes
label
data
by
yourself.
You
can
just
using
those
id
and
directly
download
it
here
yeah,
so
you
can
also
search
it,
and
this
is
a
system
for
the
access
when
using
target.
A
Now
you
can
see
that
you
have
a
gpu
acceleration
power
on,
so
you
don't
need
to
spend
a
lot
of
time
to
install
gpu
driver.
If
you
want
to
gpu,
you
can
use
it
directly
and
here
is
a
pie
gate
system
which
is
built
by
vancouver
team
with
just
you
use
it
after
you
download
the
data
from
here
retrieve
all
the
data
you
can
see.
All
the
data
types
here
can
direct
unzip.
A
A
So
you
don't
want
to
be
charged
by
gpu
anymore,
and
the
cool
thing
here
is
that
science,
fire
coin
is
also
using
the
gpu
for
ceiling.
So
in
some
cases,
if
the
server
is
not
accepted
enough
deals
during
the
selling
process,
we
can
reuse
those
gpus
to
do
the
deep
learning.
So
you
don't
need
to
worry
that
if
I
don't
do
selling,
I
cannot
provide
a
service
to
our
client.
A
And
this
is
the
given
to
us
that
we
were
using
theory
for
the
power
gate.
We
have
the
dlc
check
from
time
to
time.
We
check
all
the
power
gate
systems
running
and
we
check
how
the
configuration
goes.
We're
also
running
a
small
uploader
to
upload
a
bunch
of
scientific
data
to
the
system
more
efficiently.
A
Yeah-
and
here
is
the
different
wave
is
uploading.
We
are
in
local
system.
You
can
see
that
we
have
a
source
folder.
We
can
translate
to
small
folder
and
some
further
at
the
end,
it
will
be
uploaded
to
the
system
yeah.
This
is
our
game,
and
this
is,
we
are
working
in
the
data
center.
A
This
is
our
dev
team
in
montreal,
and
this
is
our
company
and
yes,
thanks
to
everyone
who
gave
us
the
opportunity
to
introduce
our
team
and
for
anybody
interesting
to
take
a
trail
of
our
platform,
I'm
one
willing
to
provide
100
hours
for
deep
learning.
Thank
you.
B
Thank
you
very
much,
charles.
Let
me
ask
you
some
questions
before
before
you
go.
What
are
the
kinds
of
applications
that
you
think
people
will
will
will
start
using
your
system
for
and
kind
of
from
what
roadmap.
A
Yes,
I
already
like,
because
I
have
two
type
of
application-
is
the
best
for
the
platform.
First
of
all
is
people
who
are
using
python
the
one
who's
training,
some
python.
They
want
to
use
at
the
moment
price
zone
and
they
want
to
do
some
deep
learning
algorithm
because
for
everybody
knows
that
the
diploma
is
very
expensive.
Amazon
charges
you
about
three
dollars
per
hour,
with
just
one
gpu
2080
ti,
but
with
this
platform
it's
much
cheaper.
B
Yeah
and
and
what
about
what
kinds
of
data
sets
do
you
think
would
attract
a
lot
of
users?
So,
for
example,
there's
a
lot
of
really
useful
machine
learning
data
sets
out
there,
maybe
having
those
two
filecoin
already.
Would
that
attract
some
of
the
applications?
Can
you
maybe
talk
about
what
some
of
them.
A
Yes,
I
already
like
during
spaces
too.
The
firecoin
is
already
provide
lots
of
curated
data,
set,
something
like
corporate
19
data
set
and
common
core
data
site
and
lots
of
image
detection
to
create
that
this
kind
of
deals
that
we
get
it
from
the
public
license
from
the
cargo,
the
monkey
detections.
So
all
those
data
sets
which
is
have
a
public
access,
but
especially
those
for
something
like
image,
detection,
which
is
a
very
hot
spot
in
our
current
ai
industry.
A
People
are
really
interested
because
we
were
working
with
concordia
university
and
the
mcgear
university.
The
trade
training,
those
cancer
data,
our
platform
people,
but
we
have
some
issues
before
because
we
don't
have
firepoint
previously
to
using
amazon.
It
charges
a
lot,
but
when
we
destroyed
our
local
data
search,
we
cannot
share
with
other
people.
So
fragile
gave
us
a
very
good
way
to
share
the
data
on
the
same
platform
to
allow
enable
everybody
access
it,
because
when
you
uploaded
a
terabyte
of
data,
it's
not
an
easy
job.
A
It
takes
a
few
days
even
months,
and
sometimes
when
you
have
a
data
correction,
you
when
we
want
to
find
a
second
deck
of
copy,
it's
not
so
easy
with
the
fragrance
current
ecosystem
algorithm,
it
enabled
people,
the
miners,
provide
very
stable
storage
with
very
high
service
level
agreement
because
they
needed
to
continue
to
provide
the
storage
instead
of
just
turn
off
the
machine.
At
any
time,
we
saw
some
other
blockchain
projects,
progress
through
which
they
are
good
or
even
like
an
api
base.
A
But
the
big
problem
is
that
they
cannot
guarantee
that
when
people
want
to
have
those
data
in
place,
they
can
always
access
it,
because
people
can
just
turn
up.
So
during
I'm,
I'm
personally,
I'm
running
through
others
space
one
and
two
from
my
personal
perspective,
yeah,
it's
really
hard,
but
I
think
during
the
process
it
gave
people
more
sense
what
kind
of
service
providers
they
want
to
be
in
the
future.
If
this
thing
can
they
are
not
comfortable
to
provide
a
service
to
public,
for
example,
those
natural
science
data.
A
I
think
they
should
really
think
about
student
continue
or
not,
because
this
is
a
very
serious
problem,
while
looking
forward
to
provide
a
service
can
be
better
than
amazon
and
google.
It's
not
just
something
works,
so
this
is
my
current
feeling
about
the
entire
system.
B
Yeah,
that's
that's
super
exciting
and
a
really
great
application
for
for
falcon
and
for
the
broader
storage
and
deep
learning
is,
is
a
massive
part
of
computation
and
it
is
fantastic
to
see
already
an
application
platform
that
can
do
that.
B
Would
you?
How
can
how
can
the
community
help
you
more?
Should
they
try
it
out?
Is
there
should
they
start
building
building
with
it?
What
is
the
next
step
for
for
people
out
there.
A
Yeah
for
us,
like
we're
already
talking
with
some
community
people
like
like
us,
the
pocket
team
and
also
texas
data
io.
We
have
some
private
con
messaging
together,
they're
very
interesting
to
try
to
use
it
but
develop
that,
because
for
now,
when
we
check
the
power
gate,
they
already
have
lots
of
things
built,
but
they
also
using
notebook,
which
is
great.
But
the
problem
is
that
when
they
use
into
the
wood
they
need
to
start
an
instance
locally,
but
now
they
don't
need
to
do
that.
A
We
can
work
it
together,
building
platform
together,
so
people
can
go
there
and
science.
We
already
have
two
data
data
centers
in
canada.
When
people
want
to
use
it
to
moving
forward
with
gpu,
we
can
provide
those
free
gpu
hours
for
community
access,
even
college
students
who
is
studying
in
new
york
and
canada.
They
can
see
that
okay,
I
want
a
gpu
and
they
can
submit
the
applications
to
frycoin
and
they
are
more.
A
We
are
more
willing
to
provide
those
usage
we
can
for
now
we
have
about
30
gpus
in
place,
but
another
of
them
currently
working
for
setting.
As
you
know,
I'm
also
minor,
but
now
I'm
more
willing
to
provide
the
deeper
learning
this
setting
process
together.
So
we
can
take
a
look
about
how
much
we
can
using
for
gpu
different
learning.
B
Yeah,
that's
super
exciting.
Thank
you
so
much
charles
for
this
great
talk.
Thank
you
for
building
this
great
application
and
looking
forward
to
how
it
develops.
Thank
you
so
much.