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 Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4176.72
zero. So when you consider that with our 63.6% accuracy here that means that this
4,176.72
4,190.24
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4186.04
model is actually probably doing pretty well. It's not perfect of course but it's
4,186.04
4,196.24
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4190.240000000001
doing pretty well. So overall that's I mean that's everything for this video
4,190.24
4,201.52
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4196.240000000001
we've gone all the way through this. If you do want to code for this I'm gonna
4,196.24
4,206.64
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4201.52
make sure I keep a link to it in the description so check that out if you
4,201.52
4,212
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4206.64
just want to sort of copy this across. But for now that's everything so thank
4,206.64
4,232.8
How to Build Custom Q&A Transformer Models in Python
2021-02-12 13:30:03 UTC
https://youtu.be/ZIRmXkHp0-c
ZIRmXkHp0-c
UCv83tO5cePwHMt1952IVVHw
ZIRmXkHp0-c-t4212.0
4,212
4,232.8
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-t0.0
Hi and welcome to this video on question answering with Bert.
0
8.56
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-t4.0
So firstly we're going to have a look at the transformers library
4
11.76
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-t8.56
and we're going to look at how we can find a Q&A model.
8.56
14.88
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-t12.4
And then we're going to look at the Q&A pipeline.
12.4
20
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-t14.88
So we're going to look at actually loading a model in python using the transformers library.
14.88
25.28
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-t20.0
We're going to look at tokenization, how we load a tokenizer and what exactly tokenizer is actually doing.
20
29.52
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-t25.28
And then we're going to take a look at the pipeline class,
25.28
34.96
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-t29.520000000000003
which is essentially a wrapper made available by the Hugging Face Transformers library.
29.52
40.96
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-t34.96
And it basically just makes our job in terms of building a Q&A pipeline incredibly easy.
34.96
45.6
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-t40.96
So we're going to cover all those, it's going to be quite straightforward and quite simple.
40.96
47.36
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-t45.6
So let's just get straight into it.
45.6
54.16
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-t47.36
Okay so when we're doing question answering, we're essentially asking the model a question
47.36
60.64
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-t54.16
and passing a context, which is what you can see here for the model to use to answer that question.
54.16
65.68
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-t62.0
So you can see down here we have these three questions.
62
70.16
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-t65.67999999999999
So what organization is the IPCC a part of?
65.68
76.4
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-t70.16
And then the model will read through this and use its language modeling to figure out
70.16
82.4
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-t76.4
which organization the IPCC is part of, which is not inherently clear from reading this.
76.4
88.64
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-t82.4
We can see we've got IPCC here and is a scientific intergovernmental body
82.4
91.2
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-t88.64
under the auspices of the United Nations.
88.64
98.64
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-t92.48
So clearly the IPCC is a part of the United Nations, but it's not clear.
92.48
104.08
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-t98.64000000000001
It's not definitively saying that in this, but once we've actually built this model,
98.64
108.24
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-t104.08000000000001
it will quite easily be able to answer each one of these questions without any issues.
104.08
114
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-t108.24
So the first thing we want to do is go over to the Hugging Face website.
108.24
123.84
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-t114.47999999999999
And on the Hugging Face website, we just want to go over to the Models page.
114.48
124.64
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-t123.83999999999999
So it's here.
123.84
130.64
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-t125.84
Okay and on this Models page, the thing that we want to be looking at
125.84
133.28
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-t130.64
is this question and answering task.
130.64
136.8
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-t133.28
So here we have all these tasks because when you're working with transformers,
133.28
139.28
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-t136.8
they can work with a lot of different things.
136.8
145.28
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-t140.24
Text summarization, text classification, generation, loads of different things.
140.24
147.68
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-t145.28
But what we want to do is question answering.
145.28
153.76
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-t147.68
So we click on here and this filters all of the models that are available to us
147.68
156.8
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-t154.24
just purely for question and answering.
154.24
164.4
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-t158.24
So this is the sort of power of using the Hugging Face Transformers library.
158.24
169.6
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-t164.4
It already has all these pre-trained models that we can just download and start using.
164.4
175.04
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-t171.20000000000002
Now, when you want to go and apply these to specific use cases,
171.2
179.44
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-t175.04000000000002
you probably want to fine tune it, which means you want to train it a little bit more
175.04
180.8
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-t179.44
than what it is already trained.
179.44
186.4
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-t181.68
But for actually getting used to how all of this works,
181.68
190.32
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-t186.4
all you need to do is download this model and start asking questions
186.4
193.36
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-t190.32
and understanding how everything is actually functioning.
190.32
195.92
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-t193.36
So obviously there's a lot of models here.
193.36
199.12
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-t195.92000000000002
We've got 262 models for question answering,
195.92
201.12
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-t199.12
and there's new ones being added all the time.
199.12
205.44
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-t201.84
A few of the ones that I would recommend using are the DeepSets models.
201.84
211.2
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-t207.44000000000003
So here are the DeepSets models, eight of them for question answering.
207.44
214.96
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-t211.20000000000002
The one that we will be using is this BERT Base Case Squad 2.
211.2
219.28
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-t214.96
Another one that I would definitely recommend trying out is this Electra Base Squad 2.
214.96
222.64
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-t219.28
But we will be sticking with BERT Base.
219.28
227.36
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-t222.64000000000001
Now, it's called DeepSet here because it's from the DeepSet AI company,
222.64
230.96
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-t227.36
and this model is being pulled directly from their GitHub repository.
227.36
234.64
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-t230.96
So DeepSet is actually the GitHub organization,
230.96
237.68
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-t234.64
and then this is the repository BERT Base Case Squad 2.
234.64
241.2
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-t238.4
BERT is obviously the model BERT from Google AI.
238.4
245.04
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-t242.0
Base is the base version of BERT.
242
249.84
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-t245.04
So you can see here we have BERT large, that's just a large model. We're using the base model.
245.04
254.88
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-t249.84
Case just refers to the fact that this model will differentiate between
249.84
258.24
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-t254.88
uppercase and lowercase characters or words.
254.88
262.64
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-t258.24
The alternative to this would be uncased here where there's no differentiation
258.24
264.4
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-t262.64
between uppercase and lowercase.
262.64
270.88
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-t264.4
And then Squad 2 refers to the question answering data set that this model has been trained on,
264.4
272.96
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-t270.88
which is the Squad 2 version.
270.88
275.2
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-t272.96
So we're going to take this model.
272.96
280.96
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-t275.2
So you see DeepSet BERT Base Case Squad 2, and we are going to load it into here.
275.2
287.6
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-t282.0
And all we need to do to do that is from transformers.
282
290.48
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-t287.59999999999997
So this is the Hug & Face Transformers library.
287.6
298.48
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-t292.24
We want to import BERT for question and answer.
292.24
305.6
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-t298.48
So this is a specific class and using this class we can initialize a few different models,
298.48
307.36
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-t305.6
not just this specific model.
305.6
309.68
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-t307.36
So you can see here we have this BERT base case.
307.36
314.24
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-t309.68
We can also initialize this BERT large uncased Roberta.
309.68
318.16
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-t314.24
And if there's a Distill BERT as well, we can also load those in.
314.24
324.4
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-t319.92
And what this does is it loads that BERT base case.
319.92
332.48
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-t324.4
And what this does is it loads that specific model with its question and answering layer
324.4
333.76
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-t332.47999999999996
added on there as well.
332.48
338.4
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-t333.76
So this model has been trained with the extra layer specifically for question answering.
333.76
341.92
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-t338.4
And we need to use BERT for question answering to load that.
338.4
347.76
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-t341.91999999999996
Otherwise, if you are not using it with a specific use case and you're just wanting
341.92
352.56
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-t347.76
to get the model itself, you can just use the auto model class like that.
347.76
355.68
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-t352.56
But we want it for question answering, so we load this one.
352.56
361.12
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-t355.68
Another thing to note is that we are using the PyTorch implementation of BERT here.
355.68
368.48
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-t361.92
So Transformers works by having both TensorFlow and PyTorch as alternative frameworks working
361.92
369.68
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-t368.48
behind the scenes.
368.48
371.6
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-t369.68
In this case, we're using PyTorch.
369.68
376.56
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-t371.6
If you want to switch over to TensorFlow, all you do is add TF in front of that class.
371.6
380.48
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-t378.4
So that is our model.
378.4
389.84
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-t380.48
And to actually load that in, all we do is copy this and we use the from pre-trained
380.48
390.34
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-t389.84000000000003
method.
389.84
396.32
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-t392.16
And then this is where the model name from over here comes into play.
392.16
398.96
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-t396.32
So we've got deep set BERT base case squad 2.
396.32
402.88
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-t401.36
And we just enter that in there.
401.36
405.76
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-t402.88
And we just enter that in there.
402.88
417.52
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-t414.4
Okay, and with that, we've actually just loaded the model.
414.4
418.64