Datasets:
jamescalam
/
youtube-transcriptions

Dataset card Data Studio Files
xet
Community
2
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
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t873.54
And the output of that tokenizer, our token IDs will be fed into BERT.
873.54
882.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t877.46
BERT will return us a span start and span end,
877.46
887.06
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t882.02
which is essentially two numbers, which signify the start position
882.02
890.18
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t887.06
and end position of our answer within the context.
887.06
894.34
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t890.18
And this pipeline will take those two numbers and apply them to our context
890.18
898.26
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t894.34
to get the text, which is our answer from that.
894.34
902.98
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t898.98
So it's essentially just a little wrapper and it adds a few functionalities
898.98
906.34
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t902.98
so that we don't have to worry about converting all of these things.
902.98
910.82
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t907.78
So now we just need to pass in our model.
907.78
912.34
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t910.82
And the tokenizer as well.
910.82
916.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t914.98
And it's as simple as that.
914.98
918.18
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t916.0200000000001
That's our pipeline set up.
916.02
923.22
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t918.82
So if we want to use that now, all we need to do is write NLP.
918.82
926.66
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t924.74
And then here we pass a dictionary.
924.74
933.22
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t927.62
And this dictionary, like I said before, needs to contain our question and context.
927.62
934.9
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t934.1800000000001
So the question.
934.18
940.82
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t934.9
And for this, we will just pass the first of our questions up here again.
934.9
946.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t940.8199999999999
So this questions at the index zero.
940.82
950.18
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t948.66
And then we also pass our context,
948.66
954.58
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t951.22
which is inside the context variable up here.
951.22
957.94
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t957.38
Okay.
957.38
966.5
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t957.94
And this will output a dictionary containing the, well, we can see the score of the answer.
957.94
970.18
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t966.5
So that is the model's confidence that this is actually an answer.
966.5
980.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t970.9000000000001
Like I said before, the start index and end index and what those start index and end index map to,
970.9
981.38
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t980.0200000000001
which is United Nations.
980.02
984.82
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t981.3800000000001
So our question was, what is the answer?
981.38
988.58
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t984.82
And we got United Nations, which is correct.
984.82
993.78
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t988.58
So let me just show you what I mean with this start and end.
988.58
1,001.94
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t993.7800000000001
So if we do 118 here, we get the first letter of our answer because we are going through here
993.78
1,005.3
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1001.94
and it is pulling out this specific character.
1,001.94
1,011.06
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1007.3000000000001
If we then add the first letter of our answer,
1,007.3
1,019.38
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1011.06
if we then add this and go all the way up to our end, which is at 132,
1,011.06
1,026.66
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1020.66
we get the full set because what we're doing here is pulling out all the characters from you
1,020.66
1,035.06
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1026.6599999999999
or character 118 all the way up to character 132, which is actually this comma here.
1,026.66
1,039.62
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1035.06
But obviously with Python list indexing, we get the character before.
1,035.06
1,042.58
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1039.62
And that gives us United Nations, which is our answer.
1,039.62
1,046.42
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1044.4199999999998
So let's ask another question.
1,044.42
1,053.22
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1048.9799999999998
We have what UN organizations established the IPCC?
1,048.98
1,060.82
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1055.54
And we get this WMO and United Nations Environment Program, UNEP.
1,055.54
1,066.5
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1062.6599999999999
So if we go in here, we can see it was first established in 1988
1,062.66
1,072.82
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1066.5
by two United Nations organizations, the World Meteorological Organization,
1,066.5
1,076.74
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1072.82
WMO, and the United Nations Environment Program, UNEP.
1,072.82
1,084.1
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1077.78
So here we have two organizations and it is only actually pulling out one of those.
1,077.78
1,091.14
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1085.7
So I think the reason for that is all that is reading is WMO and United Nations Environment
1,085.7
1,097.06
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1091.14
Program. So it is pulling out those two organizations in the end, just not the full name of the first one.
1,091.14
1,099.46
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1097.8600000000001
So it's still a pretty good result.
1,097.86
1,103.78
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1100.1000000000001
And let's go down to this final question.
1,100.1
1,109.06
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1105.7800000000002
So what does the UN want to stabilize?
1,105.78
1,114.74
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1110.74
And here we're getting the answer of greenhouse gas concentrations in the atmosphere.
1,110.74
1,122.74
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1114.74
So if we go down here, we can see the ultimate objective of the UNFCCC
1,114.74
1,128.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1122.74
is to stabilize greenhouse gas concentrations in the atmosphere at a level that would prevent
1,122.74
1,131.54
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1128.02
dangerous anthropogenic interference with the climate system.
1,128.02
1,137.78
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1132.5
So again, we are getting the answer, stabilize greenhouse gas concentrations.
1,132.5
1,143.7
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1137.78
So our model has gone through each one of those questions and successfully answered them.
1,137.78
1,145.78
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1143.7
And all we've done is written a few lines of code.
1,143.7
1,149.54
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1147.06
And this is without us fine tuning them at all.
1,147.06
1,154.98
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1150.1
Now, when you do go and apply these to your own problems, sometimes you won't need to do
1,150.1
1,159.22
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1154.98
any fine tuning and the model as is will be more than enough.
1,154.98
1,161.86
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1159.22
But a lot of the time you will need to fine tune it.
1,159.22
1,164.98
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1161.86
And in that case, the answer is yes.
1,161.86
1,169.54
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1164.98
You fine tune it and in that case, there are a few extra steps.
1,164.98
1,174.66
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1170.5
But for this introduction, that's everything I wanted to cover there.
1,170.5
1,177.38
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1174.66
In terms of fine tuning, I have covered that in another video.
1,174.66
1,179.94
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1177.38
So I will put a link to that in the description.
1,177.38
1,182.5
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1180.58
But that's everything for this video.
1,180.58
1,184.74
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1182.5
So thank you very much for watching.
1,182.5
1,188.02
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1184.74
I hope you enjoyed and I will see you again next time.
1,184.74
1,188.5
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1188.02
Thanks.
1,188.02
1,195.22
How to Build Q&A Models in Python (Transformers)
2021-02-19 15:00:21 UTC
https://youtu.be/scJsty_DR3o
scJsty_DR3o
UCv83tO5cePwHMt1952IVVHw
scJsty_DR3o-t1188.5
1,188.5
1,195.22
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t0.0
Hi and welcome to the video. We're going to go through language generation using GPT-2.
0
14.64
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t6.8
Now this is actually incredibly easy to do and we can build this entire model including the imports,
6.8
22
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t14.64
the tokenizer model and outputting our generated text with just seven lines of code which is
14.64
29.6
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t22.0
pretty insane. Now the only libraries we need for this are PyTorch and Transformers. So we'll go
22
49.76
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t29.6
ahead and import them now. Now all we need from the Transformers library are the GPT-2 LMHead
29.6
63.6
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t49.76
model and GPT-2 tokenizer. So we can initialize both of those as well now and both will be from
49.76
85.92
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t63.6
pre-trained. So now we have initialized our tokenizer and model. We just need a sequence of
63.6
96.72
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t85.92
text to feed in and get our model going. So I've taken a snippet of text from the Wikipedia page
85.92
105.76
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t96.72
of Winston Churchill which is here and it's just a small little snippet talking about when he took
96.72
112.56
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t105.76
office during World War II. Now from this I've tested it briefly and it seems to give some
105.76
119.84
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t112.56
pretty interesting results. So we will go ahead use this. All we need to do is tokenize it.
112.56
134.16
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t126.56
Now all we're doing here is taking each of these words, splitting them into tokens. So that would
126.56
143.28
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t134.16
be a list where each word is its own item. So he began his premiership. Each one of those would be
134.16
151.36
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t143.28
a separate value within that list. Once we have them in that tokenized format our tokenizer will
143.28
164.08
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t151.36
then convert them into numerical IDs which map to a word vector that's been trained to work with GPT-2.
151.36
169.68
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t164.08
Now because we're using PyTorch we just need to remember to return PT tensors here.
164.08
179.44
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t169.68
So now we have our inputs we just need to feed them into our model. So we can do that using
169.68
192.96
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t179.44
model.generate. We add our inputs. Now we also need to tell PyTorch how long we want our generate
179.44
201.76
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t192.96
sequence to be. So all we do for that is add a max length. And this will act as the cut off point.
192.96
212.48
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t201.76000000000002
Anything longer than this will simply be cut off. And now here we are just generating our output.
201.76
223.04
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t212.48
We also need to pass this into the outputs variable here so that we can actually read from it and
212.48
232
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t223.04
decode it. So to decode our output IDs because it will output numerical IDs representing words just
223.04
236.64
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t232.0
like we fed into it we need to use the tokenizer decode method.
232
246
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t236.64
And our output IDs are in the zero index of the outputs object. And we also want to skip any
236.64
254.08
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t246.0
special tokens. So this would be stuff like end of sequence tokens, padding tokens, unknown word
246
265.2
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t254.08
tokens and so on. And then we can print the text. Now we can see here that it's basically just going
254.08
271.44
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t265.2
over and over again saying the same things which is not really what we want. So this is a pretty
265.2
277.44
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t271.44
common problem and all we need to do to fix this is add a new output. So we can just add a new
271.44
287.84
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t277.44
argument to our generator method here. So we simply do sample equals true. And then we can rerun this.
277.44
292.32
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t290.48
And this looks pretty good now.
290.48
300.24
Language Generation with OpenAI's GPT-2 in Python
2020-11-24 14:22:46 UTC
https://youtu.be/YvVQgvAz9dY
YvVQgvAz9dY
UCv83tO5cePwHMt1952IVVHw
YvVQgvAz9dY-t295.2
So we can add more randomness and restrict the number of possible texts.
295.2
307.04