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-t1665.36
|
a few of these ending points as well.
| 1,665.36 | 1,676.64 |
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-t1673.9199999999998
|
OK, so I think that looks pretty good.
| 1,673.92 | 1,682.64 |
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-t1677.84
|
And that means we can move on to actually encoding our text.
| 1,677.84 | 1,696.88 |
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-t1682.64
|
To tokenize or encode our text, this is where we bring in a BERT tokenizer.
| 1,682.64 | 1,700.48 |
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-t1698.16
|
So we need to import the Transformers library for this.
| 1,698.16 | 1,706.72 |
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-t1702.4
|
And from Transformers, we are going to import the Distilbert.
| 1,702.4 | 1,712.96 |
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-t1706.72
|
So Distilbert is a smaller version of BERT, which is just going to run a bit quicker,
| 1,706.72 | 1,714.32 |
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-t1712.96
|
but it will take a very long time.
| 1,712.96 | 1,723.44 |
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-t1716.72
|
And we're going to import the FAST version of this tokenizer because this allows us to more
| 1,716.72 | 1,731.36 |
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-t1723.44
|
easily adjust our character and then start locations to token and start locations later on.
| 1,723.44 | 1,740 |
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-t1731.36
|
So first, we need to actually initialize our tokenizer, which is super easy.
| 1,731.36 | 1,760.48 |
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-t1740.0
|
All we're doing is loading it from a pre-trained model.
| 1,740 | 1,768.8 |
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-t1761.28
|
And then all we do to create our encodings is to load the
| 1,761.28 | 1,774.8 |
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-t1768.8
|
tokenizer. So we'll do the training set first.
| 1,768.8 | 1,778.96 |
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-t1777.68
|
Let's call it tokenizer.
| 1,777.68 | 1,782.72 |
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-t1780.1599999999999
|
And in here, we include our training context.
| 1,780.16 | 1,788.32 |
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-t1785.9199999999998
|
And the training questions.
| 1,785.92 | 1,792.16 |
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-t1789.76
|
So what this will do
| 1,789.76 | 1,796 |
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-t1792.16
|
is actually merge these two strings together.
| 1,792.16 | 1,802.08 |
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-t1796.0
|
So what we will have is our context and then there will be a separator token
| 1,796 | 1,804.32 |
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-t1802.88
|
followed by the question.
| 1,802.88 | 1,807.68 |
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-t1804.3200000000002
|
And this will be fed into Distilbert during training.
| 1,804.32 | 1,813.92 |
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-t1811.68
|
I just want to add padding there as well.
| 1,811.68 | 1,817.76 |
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-t1813.92
|
And then we'll copy this and do the same for our relation set.
| 1,813.92 | 1,825.44 |
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-t1817.76
|
Okay, and this will convert our data into encoding objects.
| 1,817.76 | 1,842.64 |
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-t1827.04
|
So what we can do here is
| 1,827.04 | 1,852.2 |
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-t1842.64
|
So what we can do here is print out different parts that we have within our
| 1,842.64 | 1,864.52 |
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-t1852.2
|
encodings. So in here you have the input IDs so let's access that and you'll find
| 1,852.2 | 1,870.84 |
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-t1864.5200000000002
|
in here we have a big list of all of our samples so check that we have
| 1,864.52 | 1,879.68 |
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-t1870.84
|
130k and let's open one of those okay and we have these token IDs and this is
| 1,870.84 | 1,885.72 |
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-t1879.6799999999998
|
what Bert will be reading. Now if we want to have a look at what this actually is
| 1,879.68 | 1,892.64 |
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-t1885.72
|
in sort of human readable language we can use the tokenizer to just decode it for
| 1,885.72 | 1,904.44 |
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-t1892.64
|
us. Okay this is what we're feeding in so we have a couple of these special
| 1,892.64 | 1,911.28 |
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-t1904.44
|
tokens this just means it's the sort of sequence and in here we have a process
| 1,904.44 | 1,918.8 |
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-t1911.2800000000002
|
form of our original context. Now you find that the context actually ends here
| 1,911.28 | 1,923.64 |
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-t1918.8
|
and like I said before we have this separated token and then after that we
| 1,918.8 | 1,930.56 |
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-t1923.6399999999999
|
have our actual question and this is what is being fed into Bert but obviously
| 1,923.64 | 1,937.16 |
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-t1930.56
|
the token ID version. So it's just good to be aware of what is actually being
| 1,930.56 | 1,941.68 |
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-t1937.1599999999999
|
fed in and what we're actually using here but this is a format that Bert is
| 1,937.16 | 1,945.36 |
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-t1941.68
|
expecting and then after that we have another separated token followed by all
| 1,941.68 | 1,952.28 |
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-t1945.36
|
of our padding tokens because Bert is going to be expecting 512 tokens to be
| 1,945.36 | 1,957.64 |
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-t1952.28
|
fed in for every one sample so we just need to fill that space essentially so
| 1,952.28 | 1,967.84 |
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-t1957.6399999999999
|
that's all that is doing. So let's remove those and we can continue. So the next
| 1,957.64 | 1,975.44 |
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-t1967.84
|
thing we need to add to our encodings is the start and end positions because at the
| 1,967.84 | 1,983.16 |
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-t1975.4399999999998
|
moment we just don't have them in there. So to do that we need to add a additional
| 1,975.44 | 1,990.72 |
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-t1983.1599999999999
|
bit of logic. We use this character to token method so if we just take out one
| 1,983.16 | 2,004.8 |
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-t1990.72
|
of these. Let's take the first one. Okay we have this and what we can do is
| 1,990.72 | 2,011.76 |
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-t2004.8
|
actually modify this to use the character token method. Remove the input
| 2,004.8 | 2,017.08 |
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-t2011.76
|
IDs because we just need to pass it the index of whichever encoding we are
| 2,011.76 | 2,024.24 |
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-t2017.08
|
wanting to modify or get the start and end position of and in here all we're
| 2,017.08 | 2,030.22 |
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-t2024.24
|
doing is converting from the character that we have found a position for to the
| 2,024.24 | 2,036.4 |
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-t2030.22
|
token that we want to find a position for and what we need to add is train
| 2,030.22 | 2,043.8 |
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-t2036.3999999999999
|
answers. We have our position again because the answers and encodings the
| 2,036.4 | 2,047.56 |
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-t2043.8
|
context and question that needs to match up to the answer of course that we're
| 2,043.8 | 2,056.44 |
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-t2047.56
|
asking about and we do answers start. So here we're just feeding in the position
| 2,047.56 | 2,064.24 |
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-t2056.44
|
of the character and this is answer. Okay so feeding in the position of the character and
| 2,056.44 | 2,072.72 |
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-t2064.24
|
we're expecting to return the position of the token which is position 64. So
| 2,064.24 | 2,078.04 |
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-t2072.72
|
all we need to do now is do this for both of those so for the start position
| 2,072.72 | 2,091.68 |
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-t2078.04
|
and end position. See here we should get a different value. Okay but this is one
| 2,078.04 | 2,098.28 |
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-t2091.68
|
limitations of this. Sometimes this is going to return nothing as you can see
| 2,091.68 | 2,103.52 |
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-t2098.28
|
it's not returning anything here and that is because sometimes it is actually
| 2,098.28 | 2,110.04 |
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-t2103.52
|
returning the space and when it looks at the space and the tokenizer see
| 2,103.52 | 2,113.88 |
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-t2110.0400000000004
|
that and they say okay that's nothing we're not concerned about spaces and it
| 2,110.04 | 2,120.92 |
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-t2113.88
|
returns this non value that you can see here. So this is something that we need
| 2,113.88 | 2,127.48 |
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-t2120.92
|
to consider and build in some added logic for. So to do that again we're
| 2,120.92 | 2,138.96 |
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-t2127.48
|
going to use a function to contain all this and call it add token positions.
| 2,127.48 | 2,145.8 |
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-t2138.96
|
Here we'll have our encodings and our answers and then we just modify this code so we have
| 2,138.96 | 2,155.92 |
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-t2145.8
|
the encodings we have the answers and because we're collecting all of the
| 2,145.8 | 2,162.68 |
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-t2155.92
|
token positions we also need to initialize a list to containers. So we
| 2,155.92 | 2,174.32 |
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-t2162.6800000000003
|
do start positions empty list and end positions. And now we just want to loop
| 2,162.68 | 2,188.6 |
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-t2174.32
|
through every single answer and encoding that we have. Like so. And here we have
| 2,174.32 | 2,199.2 |
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-t2188.6000000000004
|
our start position so we need to append that to our start positions list.
| 2,188.6 | 2,212.72 |
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-t2199.2
|
And we just do the same for our end positions which is here. Now here we can
| 2,199.2 | 2,219.56 |
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-t2212.72
|
deal with this problem that we had. So if we find that the end positions the most
| 2,212.72 | 2,227 |
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-t2219.56
|
recent one so the negative one index is non that means it wasn't found and it
| 2,219.56 | 2,234.2 |
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-t2227.0
|
means there is a space. So what we do is we change it to instead use the minus one
| 2,227 | 2,241.36 |
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-t2234.2
|
version. And all this needs to do is update the end positions here. Okay
| 2,234.2 | 2,248.84 |
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-t2241.36
|
that's great but in some cases this also happens with the start position but that
| 2,241.36 | 2,252.2 |
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-t2248.84
|
is for a different reason. The reason that will occasionally happen with start
| 2,248.84 | 2,258.64 |
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-t2252.2
|
position is when the passage of data that we're adding in here so you saw
| 2,252.2 | 2,263.96 |
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-t2258.64
|
before we had the context that separated token and then the question. Sometimes
| 2,258.64 | 2,270.88 |
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-t2263.96
|
the context passage is truncated in order to fit in the question. So some of
| 2,263.96 | 2,276.96 |
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-t2270.8799999999997
|
it will be cut off and in that case we do have a bit of a problem but we still
| 2,270.88 | 2,284.84 |
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-t2276.96
|
need to just allow our code to run without any problems. So what we do is we
| 2,276.96 | 2,292.36 |
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-t2284.84
|
just modify the start positions again just like we did with the end positions.
| 2,284.84 | 2,302.76 |
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-t2292.36
|
Obviously only if it's a non and we just set it to be equal to the maximum length
| 2,292.36 | 2,307.4 |
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-t2302.76
|
that has been defined by the tokenizer.
| 2,302.76 | 2,319.26 |
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-t2313.0
|
It's as simple as that. Now the only final thing we need to do which is because we're
| 2,313 | 2,324.96 |
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-t2319.26
|
using the encodings is actually update those encodings to include this
| 2,319.26 | 2,331.64 |
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-t2324.96
|
data because as of yet we haven't added that back in. So to do that we can use
| 2,324.96 | 2,340.12 |
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-t2331.64
|
this quite handy update method and just add in our data as a dictionary. So you
| 2,331.64 | 2,358 |
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-t2340.12
|
have start positions, start positions and we also have our end positions. And then
| 2,340.12 | 2,363.88 |
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-t2358.0
|
again we just need to apply this to our training and validation sets and let's
| 2,358 | 2,368.08 |
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-t2363.88
|
just modify that.
| 2,363.88 | 2,379.68 |
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-t2373.8
|
Let's add the training encodings here and train answers.
| 2,373.8 | 2,391.6 |
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-t2379.68
|
We do that again the validation set.
| 2,379.68 | 2,407.52 |
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-t2398.7599999999998
|
So now let's take a look at our encodings and here we can see great now
| 2,398.76 | 2,413.78 |
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-t2407.52
|
have those start positions and end positions. We can even so a quick look
| 2,407.52 | 2,417.12 |
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-t2413.78
|
what they look like.
| 2,413.78 | 2,433.6 |
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-t2427.04
|
What we've done is actually not included the index here so we're just taking it
| 2,427.04 | 2,442.4 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.