title
stringlengths 12
112
| published
stringlengths 19
23
| url
stringlengths 28
28
| video_id
stringlengths 11
11
| channel_id
stringclasses 5
values | id
stringlengths 16
31
| text
stringlengths 0
596
| start
float64 0
37.8k
| end
float64 2.18
37.8k
|
|---|---|---|---|---|---|---|---|---|
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t456.0
|
and then here
| 456 | 460 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t458.0
|
we extract our sentences like
| 458 | 462 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t460.0
|
we did with the bag before so
| 460 | 464 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t462.0
|
we go sentences equals
| 462 | 468 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t466.0
|
and here we
| 466 | 470 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t468.0
|
want to write sentence
| 468 | 472 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t470.0
|
for sentence
| 470 | 474 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t472.0
|
in paragraph
| 472 | 476 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t474.0
|
dot split
| 474 | 480 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t478.0
|
and remember we have those
| 478 | 482 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t480.0
|
random empty sentences we don't want to
| 480 | 484 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t482.0
|
include those so we
| 482 | 486 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t484.0
|
write if sentence
| 484 | 488 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t486.0
|
is not equal
| 486 | 490 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t488.0
|
to that
| 488 | 492 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t490.0
|
empty sentence.
| 490 | 494 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t492.0
|
So we
| 492 | 496 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t494.0
|
have now so we're now looping
| 494 | 498 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t496.0
|
through each paragraph and we've
| 496 | 500 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t498.0
|
split each paragraph into sentences.
| 498 | 502 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t500.0
|
Now what we want to do
| 500 | 504 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t502.0
|
is check
| 502 | 506 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t504.0
|
if that paragraph e.g.
| 504 | 508 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t506.0
|
our new sentences variable has more
| 506 | 510 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t508.0
|
than one sentence. So
| 508 | 512 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t510.0
|
we'll do
| 510 | 514 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t512.0
|
number of sentences
| 512 | 516 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t514.0
|
equals the length of
| 514 | 518 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t516.0
|
sentences and
| 516 | 520 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t518.0
|
then we say if
| 518 | 522 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t520.0
|
number of sentences
| 520 | 524 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t522.0
|
is greater than one
| 522 | 528 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t526.0
|
oops OK don't excuse it
| 526 | 530 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t528.0
|
right now and then
| 528 | 532 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t530.0
|
we apply our 50-50 logic
| 530 | 534 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t532.0
|
and append
| 532 | 536 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t534.0
|
that to our actual training data.
| 534 | 538 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t536.0
|
Otherwise if it's just a single sentence
| 536 | 540 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t538.0
|
we don't actually add it
| 538 | 542 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t540.0
|
to the training data. I mean ideally
| 540 | 544 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t542.0
|
we would want to do something like that but
| 542 | 546 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t544.0
|
for this use case
| 544 | 548 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t546.0
|
I don't want to get
| 546 | 550 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t548.0
|
make things too complicated.
| 548 | 552 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t550.0
|
So the reason I'm doing that is for example this
| 550 | 554 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t552.0
|
sentence is just a single sentence in that
| 552 | 556 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t554.0
|
paragraph and we can't guarantee
| 554 | 558 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t556.0
|
that each continuous paragraph
| 556 | 560 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t558.0
|
is talking about the same subject
| 558 | 562 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t560.0
|
you might switch. So
| 560 | 564 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t562.0
|
for the sake of simplicity
| 562 | 566 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t564.0
|
I'm just going to ignore
| 564 | 568 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t566.0
|
the single sentence paragraphs
| 566 | 570 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t568.0
|
although we do have them in
| 568 | 572 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t570.0
|
our bag so they
| 570 | 574 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t572.0
|
can be pulled in as potential sentence
| 572 | 576 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t574.0
|
bees when we randomise
| 574 | 578 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t576.0
|
the selection.
| 576 | 580 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t578.0
|
Now what we want to do is
| 578 | 582 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t580.0
|
set the sentence that we will
| 580 | 584 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t582.0
|
start from. So we write
| 582 | 586 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t584.0
|
start equals
| 584 | 588 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t586.0
|
random
| 586 | 590 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t588.0
|
randint
| 588 | 592 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t590.0
|
so this is only if we have more than one
| 590 | 594 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t592.0
|
sentence remember.
| 592 | 596 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t594.0
|
So our
| 594 | 598 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t596.0
|
random randint so this
| 596 | 600 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t598.0
|
is going to be the start sentence
| 598 | 602 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t600.0
|
in the case that we use
| 600 | 604 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t602.0
|
sentence A and B consecutively
| 602 | 606 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t604.0
|
so we don't randomise sentence B
| 604 | 608 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t606.0
|
we want to make sure
| 606 | 610 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t608.0
|
that we have enough space
| 608 | 612 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t610.0
|
at the end of our sentences.
| 610 | 616 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t612.0
|
So here
| 612 | 618 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t616.0
|
to take both sentence A and
| 616 | 620 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t618.0
|
sentence B. So let's say
| 618 | 622 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t620.0
|
for example we have
| 620 | 624 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t622.0
|
I'm going to use an example here so we have
| 622 | 626 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t624.0
|
0, 1, 2, 3, 4
| 624 | 628 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t626.0
|
let's say this is our paragraph
| 626 | 630 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t628.0
|
we have 5 sentences in here
| 628 | 632 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t630.0
|
we want the start
| 630 | 634 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t632.0
|
sentence to sentence A
| 632 | 636 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t634.0
|
if we select 4
| 634 | 638 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t636.0
|
then we don't
| 636 | 640 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t638.0
|
have a sentence B to select from.
| 638 | 642 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t640.0
|
So what we are going to do here
| 640 | 644 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t642.0
|
is say you choose a
| 642 | 646 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t644.0
|
random integer between 0
| 644 | 648 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t646.0
|
and we want 3 to be the maximum
| 646 | 650 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t648.0
|
there.
| 648 | 652 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t650.0
|
So how do we do that?
| 650 | 654 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t652.0
|
We've got the number of
| 652 | 656 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t654.0
|
sentences here so this value will be 5
| 654 | 658 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t656.0
|
in this case so
| 656 | 660 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t658.0
|
we would say number of sentences
| 658 | 662 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t660.0
|
is 5 minus
| 660 | 664 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t662.0
|
2 because we don't
| 662 | 666 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.