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-t42.0
|
data I'm going to be using requests
| 42 | 46 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t44.0
|
for that so I'm going to import requests.
| 44 | 48 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t46.0
|
And
| 46 | 50 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t48.0
|
for our actual training of the model
| 48 | 52 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t50.0
|
we're going to be using both
| 50 | 54 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t52.0
|
HuginFaces Transformers and
| 52 | 56 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t54.0
|
PyTorch. So I need to
| 54 | 58 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t56.0
|
import Transformers
| 56 | 60 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t58.0
|
and I'm going
| 58 | 62 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t60.0
|
to import a BERT tokeniser
| 60 | 64 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t62.0
|
class
| 62 | 68 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t66.0
|
and also a BERT
| 66 | 70 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t68.0
|
for Net Sentence Prediction class.
| 68 | 72 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t70.0
|
So BERT
| 70 | 74 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t72.0
|
for Net Sentence
| 72 | 78 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t76.0
|
Prediction.
| 76 | 80 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t78.0
|
And as well as that we need to import
| 78 | 82 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t80.0
|
Torch.
| 80 | 84 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t82.0
|
So once we've imported all those
| 82 | 86 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t84.0
|
we can initialise our tokeniser
| 84 | 88 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t86.0
|
and model. So tokeniser
| 86 | 90 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t88.0
|
equals BERT tokeniser
| 88 | 94 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t92.0
|
from Pre-trained.
| 92 | 98 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t96.0
|
I'm going to be using BERT based on case
| 96 | 100 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t98.0
|
for this example.
| 98 | 102 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t100.0
|
Obviously you can use another BERT model
| 100 | 104 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t102.0
|
if you'd like.
| 102 | 106 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t104.0
|
So I'm going to copy that
| 104 | 108 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t106.0
|
and initialise our
| 106 | 110 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t108.0
|
model as well.
| 108 | 112 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t110.0
|
Okay.
| 110 | 116 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t114.0
|
And we can run that.
| 114 | 118 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t116.0
|
And now let's extract
| 116 | 120 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t118.0
|
this data. So this warning
| 118 | 122 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t120.0
|
here, don't need to worry about that. That's just saying
| 120 | 124 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t122.0
|
if you are using this model
| 122 | 126 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t124.0
|
for inference you shouldn't
| 124 | 128 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t126.0
|
because you need to train it
| 126 | 130 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t128.0
|
a little bit. So
| 128 | 132 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t130.0
|
we don't need to worry about that, that's fine.
| 130 | 134 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t132.0
|
Because we are going to be training.
| 132 | 136 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t134.0
|
Now what we do need to do
| 134 | 138 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t136.0
|
is get this data.
| 136 | 140 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t138.0
|
So new data equals
| 138 | 142 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t140.0
|
requests.get
| 140 | 144 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t142.0
|
and we just take
| 142 | 146 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t144.0
|
this link. I will keep
| 144 | 148 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t146.0
|
this link in the description
| 146 | 150 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t148.0
|
for you so you can just copy it across
| 148 | 152 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t150.0
|
if you want to follow along.
| 150 | 154 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t152.0
|
And we should see that we get a 200
| 152 | 156 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t154.0
|
response there, that's good.
| 154 | 158 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t156.0
|
So all we need to do is extract the
| 156 | 160 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t158.0
|
text from that and we're going to store it
| 158 | 162 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t160.0
|
in another variable
| 160 | 164 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t162.0
|
here, text variable.
| 162 | 166 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t164.0
|
And if we go back to our
| 164 | 168 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t166.0
|
text variable and if we
| 166 | 170 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t168.0
|
just have a quick look at what we have in there
| 168 | 172 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t170.0
|
we see that we have all of these paragraphs.
| 170 | 174 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t172.0
|
This is from the
| 172 | 176 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t174.0
|
Meditations by Marcus Aurelius.
| 174 | 178 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t176.0
|
You can get that book online, that's why
| 176 | 180 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t178.0
|
I'm using it. And the language
| 178 | 182 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t180.0
|
is a bit unique as well so that's
| 180 | 184 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t182.0
|
why I want to use it because
| 182 | 186 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t184.0
|
when we're using net sentence prediction
| 184 | 188 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t186.0
|
we're training the BERT model
| 186 | 190 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t188.0
|
to better comprehend
| 188 | 192 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t190.0
|
the style of language that
| 190 | 194 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t192.0
|
we train it on. So
| 192 | 196 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t194.0
|
in here we have our
| 194 | 198 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t196.0
|
paragraphs and they're all separated by a new
| 196 | 200 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t198.0
|
line character. So I'm just going to add
| 198 | 202 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t200.0
|
another little bit of code here
| 200 | 204 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t202.0
|
which is a split by new line
| 202 | 206 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t204.0
|
character. And we have a look
| 204 | 208 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t206.0
|
here, we now have a list
| 206 | 210 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t208.0
|
containing paragraphs
| 208 | 212 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t210.0
|
and that's our training data,
| 210 | 214 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t212.0
|
that's what we want to be using.
| 212 | 218 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t216.0
|
So when
| 216 | 220 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t218.0
|
we're using NSP we
| 218 | 222 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t220.0
|
want to create a 50-50
| 220 | 224 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t222.0
|
split of sentences
| 222 | 226 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t224.0
|
that are random and sentences
| 224 | 228 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t226.0
|
that are not random.
| 226 | 230 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t228.0
|
So we're going to be taking
| 228 | 232 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t230.0
|
sentence A's and
| 230 | 234 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t232.0
|
50% of the time we're going
| 232 | 236 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t234.0
|
to be adding the genuine
| 234 | 238 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t236.0
|
sentence B for that sentence A's
| 236 | 240 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t238.0
|
e.g. the sentence that follows it in
| 238 | 242 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t240.0
|
the text. And then
| 240 | 244 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t242.0
|
the other 50% of the time we're going to be
| 242 | 246 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t244.0
|
just choosing a random sentence and
| 244 | 248 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t246.0
|
pulling that in and using
| 246 | 250 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t248.0
|
that.
| 248 | 252 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t250.0
|
So to
| 250 | 254 |
Training BERT #4 - Train With Next Sentence Prediction (NSP)
|
2021-05-27 16:15:39 UTC
|
https://youtu.be/x1lAcT3xl5M
|
x1lAcT3xl5M
|
UCv83tO5cePwHMt1952IVVHw
|
x1lAcT3xl5M-t252.0
|
do that we first
| 252 | 256 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.