系統設計者的話：學英文根本不用背，你講中文，你有在背中文嗎？學英文也是一樣的道理，那是自然而然就會的東西，用背的方式最後肯定都是不行的。 更多...

最好的學習就是有英文的父母，不斷的跟你說話、耐心的教導你，時間長了，你也就慢慢就融入其中了。

但是我們沒有英文父母怎麼辦？看影片學英文就學的起來嗎？學不起來。

我們從小到大，也看過了不少英文電影，至少也看過1、200部也跑不掉，但是為什麼看了這麼多的英文電影，英文還是很爛呢？

主要的原因是我們沒有從基礎性的強化學習。舉個例子，西方的幼兒出生時，他至少聽了各種大人講英文至少聽了2年，才慢慢開始說英文、與大人互動。

而且都是一字一句慢慢的說、慢慢的教、漸進式的、慢慢的、越來越習慣、越來越聽的懂、越來越會說的過程。

那麼我們看的影片、電影過程中，都是大人成熟型的講話速度、方式、語調，我們怎麼可能跟的上？

一旦跟不上，思想中就會產生覺的好難哦、我不行種種念頭，那麼產生了這種潛在式的排斥心理作用，那麼你就更難學的會了。


所以我們必須打好基礎、深厚的基礎，萬丈高樓平地起，唯有打下深厚的基礎，後面你走的路才會越來越順。

程式界裏有句話，程式債欠的越多，最後就只能放棄。學英文也是一樣，英文單字債欠的越多，你就直接上來文法、句型、語調，最後也只能放棄。

基於上述的思維，我打造了這套系統，從最基礎的單字學習起，不斷的強化、加深你對這個單字的聽力，透過各式各樣的影片、各式各樣的人、各種不同的場景。

就只是為了訓練一個單字的聽力，無數的人跟你講同一個單字，講個20遍、50篇、100遍(且不重覆)，我就不信學不起來。


這套系統做出來後，我自己在使用的過程中，確實真的能聽出來他們講的這個單字，而且他們講的速度再快也能聽的出來。

能聽的出來，就知道他講的大概意思；能夠知道他講的大概意思，就能夠用幾個簡單的單字回應；能夠回應，也就能夠對話；當能夠對話，就能夠再修正文法、句型、語調。


我的目地就是希望自己至少能聽的懂人家在說什麼，用英文跟人家對話，大家共同努力、學習。若有其它想法或對這套系統設計的建議，也可以與我聯絡：ichich2013@gmail.com

Tensor2Tensor (TensorFlow @ O’Reilly AI Conference, San Francisco '18)
觀看次數：395
tensor
two
tensor
which
is
a
library
we've
built
on
top
of
tensor
flow
to
organize
the
world's
models
and
data
script
that
pre-processed
the
data
in
a
and
the
pre-processing
so
you
really
can
regression
tests
and
actually
know
that
not
break
with
tensorflow
110
and
that
team
on
launching
neural
networks
for
alignments
of
phrases
then
you
translate
the
phrases
and
then
you
try
to
realign
human
translators
get
is
about
30
and
about
twenty
twenty-one
and
it's
been
amazing
so
then
the
neural
networks
came
and
this
newell
translator
translates
advantage
of
neural
networks
is
between
translators
do
at
least
on
like
the
high
blue
scores
we
reduced
the
asynchronously
so
there
were
delayed
gradients
it's
a
regularizer
so
you
pre-processing
is
also
partially
Google
and
neat
so
the
tensor
2
tensor
library
attentional
layers
and
we
designed
this
not
just
reproducible
in
principle
reproducible
open
and
test
them
on
different
configurations
and
different
layer
is
tensor
flow
and
tensor
flow
in
in
the
newer
releases
it
helps
to
build
input
pipelines
that
are
performant
layers
and
caris
which
are
higher-level
everything
that's
that's
what
tensor
to
tensor
does
so
it's
a
library
it's
standard
ones
images
imagenet
c4m
nist
image
captionings
koko
translations
for
modeling
datasets
speech
to
text
music
net
that's
a
version
of
wavenet
ours
tms
then
like
algorithmic
models
like
neural
for
that
so
tensile
-
tensor
are
data
tensorflow
means
pr
from
github
we
import
internally
and
ams
grad
adaptive
learning
rate
methods
latest
optimizer
the
latest
optimizer
now
it's
a
very
good
optimizer
at
a
it's
it's
a
very
good
optimizer
saves
a
brain
deepmind
czech
Republic
on
github
and
germany
so
yeah
so
as
I
told
you
it's
unified
data
set
what
hyperparameters
go
train
pre-process
how
to
resize
images
and
so
a
summarization
data
set
that's
been
used
in
academia
it's
from
cnn
and
Daily
does
well
on
summarization
you
want
regression
tests
that
start
this
change
in
tensorflow
often
comes
out
in
the
regression
tests
versions
of
tensorflow
because
some
reversed
every
data
set
in
tensor
two
underscore
ev
it
reverses
inputs
and
targets
and
generative
models
they
can
just
do
underscore
F
it
reverses
the
well
but
that's
the
point
currently
it's
generative
model
one
gpu
might
not
be
multi-gpu
you
just
say
worker
gpu
equals
just
make
batches
8
times
larger
run
the
you
get
state-of-the-art
performance
so
optimize
so
accuracy
perplexity
these
to
03
logarithmically
or
uniformly
sophisticated
non-differentiable
optimize
that's
how
you
do
it
and
like
something
we
call
mesh
tensor
flow
that
tensor
-
tensor
which
allows
to
do
model
cores
you
have
or
splitted
a
twice
on
our
models
are
subclass
of
this
d2t
outputs
and
the
losses
if
you
need
so
don't
regression
test
we
allow
them
to
ramping
up
on
regression
testing
we're
side
we
are
on
to
simplifying
the
gpus
and
multi
gpus
moving
to
thing
is
the
trainer
line
and
the
mesh
tensorflow
allows
to
corpus
of
German
but
no
matching
and
you
well
hyper
hyper
parameter
tuning
is
an
integrated
now
when
we're
doing
more
and
reinforcement
learning
ganz
well
as
I
on
one
of
the
things
mesh
tensorflow
for
terabytes
of
ram
that's
how
many
even
at
versailles
applies
to
it
or
a
matrix
multiplication
modern
devices
they
have
like
2d
they're
state-of-the-art
on
translation
language
modeling
it's
the
next
step
in
in
in
the
mesh
tensorflow
pods
but
it
also
runs
on
clusters
of
hardware
and
with
the
tensor
2
tensor

相關影片

加Line好友 每日8AM 推播1單字

加入好友

其他人正在學習的單字

1. this
2024-05-06 06:58
2. split
2024-05-06 06:58
3. cleanse
2024-05-06 06:58
4. ends
2024-05-06 06:58
5. uploading
2024-05-06 06:58
6. allergy
2024-05-06 06:58
7. bein
2024-05-06 06:57
8. flank
2024-05-06 06:57
9. aman
2024-05-06 06:57
10. politician
2024-05-06 06:57
11. I
2024-05-06 06:57
12. mobile
2024-05-06 06:57
13. unable
2024-05-06 06:57
14. eyesight
2024-05-06 06:57
15. plunged
2024-05-06 06:57

本月觀看影片單字排行榜

本月查閱單字排行榜

本月聆聽單字排行榜