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
|
|---|---|---|---|---|---|---|---|---|
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t652.88
|
So like query like context vector, similar thing to the
| 652.88 | 663.52 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t661.84
|
query vector that we used before with XQ.
| 661.84 | 669.92 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t665.4399999999999
|
But this time I'm going to write index.query.
| 665.44 | 674 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t670.72
|
And we're going to pass XQ, so our query vector.
| 670.72 | 678 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t674.0
|
And we're going to say how many results we want to return.
| 674 | 684.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t678.0
|
Now, later on, we're going to use Streamlit, a little like a slider bar to decide how many
| 678 | 685.28 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t684.48
|
we would like to return.
| 684.48 | 687.44 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t685.28
|
But for now, we will hard code it.
| 685.28 | 695.52 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t688.72
|
And another thing that we want to include here is we want to tell Pinecone to return
| 688.72 | 698.88 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t695.52
|
the metadata because by default, it will not return metadata.
| 695.52 | 702.56 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t699.68
|
So return metadata equals true.
| 699.68 | 705.68 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t702.56
|
So these are like the extra little bits I mentioned before.
| 702.56 | 708.64 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t705.68
|
So included the title.
| 705.68 | 715.44 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t708.64
|
So like the topic, Wikipedia topic that the context is coming from and also the text itself.
| 708.64 | 722.88 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t717.1999999999999
|
OK, so we're going to return the relevant context and then we're going to look through
| 717.2 | 723.92 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t722.88
|
each of those.
| 722.88 | 731.76 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t723.92
|
Now, when we do this, there's a particular format that we need to follow.
| 723.92 | 740.72 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t731.76
|
So our context are actually going to be stored for context in XC results.
| 731.76 | 748.4 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t742.56
|
And results is going to return a list and we just want the first item in that list.
| 742.56 | 754.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t748.4
|
The reason it returns a list is because if you are querying Pinecone with multiple queries,
| 748.4 | 759.04 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t754.48
|
it will return lists of your answers for each query.
| 754.48 | 764.96 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t759.04
|
But in this case, we are only ever going to query with one query vector.
| 759.04 | 774.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t765.4399999999999
|
So we always enter at position zero here and then in there we will have all of our returned
| 765.44 | 781.12 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t774.48
|
matches inside this matches key value value.
| 774.48 | 790.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t781.12
|
So for context in there, all we're going to do is write st.write context.
| 781.12 | 794.72 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t790.48
|
And then we want to go into the metadata that we were returning.
| 790.48 | 797.84 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t794.72
|
And we have title and text here.
| 794.72 | 799.68 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t797.84
|
We don't want title, we want text.
| 797.84 | 803.92 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t799.68
|
OK, so let's save that and check that actually works.
| 799.68 | 809.6 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t804.64
|
So again, this is going to take a little while to load because we're initializing like the
| 804.64 | 813.28 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t809.6
|
full pipeline of our vector database and the tree model.
| 809.6 | 818.56 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t813.28
|
Every time we run this, it's downloading the full tree model, which takes quite a bit of time.
| 813.28 | 821.92 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t818.5600000000001
|
OK, so this is just rerun our app.
| 818.56 | 825.76 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t821.9200000000001
|
Now we can say who are the Normans?
| 821.92 | 827.76 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t825.76
|
OK.
| 825.76 | 831.84 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t828.72
|
Again, it's going to reload everything, so it's going to take a while.
| 828.72 | 834.16 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t831.84
|
We're going to fix this in the next video.
| 831.84 | 840.16 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t834.16
|
So we should be returning five contexts and if we scroll down, we can see we have these
| 834.16 | 840.96 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t840.16
|
five paragraphs.
| 840.16 | 844.56 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t840.9599999999999
|
Each one of these paragraphs is a single context.
| 840.96 | 848.96 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t844.56
|
So we could maybe we can respect the element.
| 844.56 | 856.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t852.48
|
OK, so we can see down here.
| 852.48 | 861.68 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t860.0799999999999
|
Pretty horrific to look at.
| 860.08 | 871.44 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t861.68
|
So if I zoom up, we can see each one of these is a single one of our contexts.
| 861.68 | 873.44 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t871.4399999999999
|
Right.
| 871.44 | 876.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t874.4799999999999
|
These here.
| 874.48 | 878.48 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t876.4799999999999
|
Cool.
| 876.48 | 883.76 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t878.4799999999999
|
So I think that is that's it for this video.
| 878.48 | 890.08 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t883.76
|
So now we have these backend working and the next one, what we'll do is we'll do a little
| 883.76 | 895.36 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t890.08
|
bit of a fix and the next one, what we'll do is fix this issue with it taking forever
| 890.08 | 900.64 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t895.36
|
to load, reload everything every time, which is actually super easy.
| 895.36 | 904.32 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t900.64
|
But we'll make a big difference to our app.
| 900.64 | 909.6 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t904.32
|
So thank you very much for watching and I will see you in the next one.
| 904.32 | 921.76 |
Streamlit for ML #2 - ML Models and APIs
|
2022-01-26 16:30:36 UTC
|
https://youtu.be/U0EoaFFGyTg
|
U0EoaFFGyTg
|
UCv83tO5cePwHMt1952IVVHw
|
U0EoaFFGyTg-t909.6
| 909.6 | 921.76 |
|
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t0.0
|
Okay, so we're going to take a look at Unicode normalization.
| 0 | 11.84 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t4.8
|
Unicode normalization is something that we use when we have those weird font variants
| 4.8 | 13.8 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t11.84
|
that people always use on the internet.
| 11.84 | 20.4 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t13.8
|
So if you've ever seen people using those odd characters, I think they use it to express
| 13.8 | 24.94 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t20.400000000000002
|
some form of individuality or to catch your attention.
| 20.4 | 31.76 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t24.94
|
And then we also have another issue where we have weird glyphs in text.
| 24.94 | 35.72 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t31.76
|
And this is more reasonable because it's actually a part of language.
| 31.76 | 39.44 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t35.72
|
That's like little glyphs, so you have the accents above the E's and stuff in Italian
| 35.72 | 41.12 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t39.44
|
or Spanish.
| 39.44 | 46 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t41.120000000000005
|
And those little glyphs all together, they're called diacritics.
| 41.12 | 51.52 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t46.0
|
And whenever we come across diacritics or that weird text, we can get issues when we're
| 46 | 53.48 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t51.52
|
building models.
| 51.52 | 59.6 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t53.48
|
The issues with the weird text is obviously if we have someone who's got hello world in
| 53.48 | 64.12 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t59.599999999999994
|
normal text and we're comparing it to someone who's got hello world in some weird text with
| 59.6 | 70.56 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t64.12
|
circles around every letter, we can't actually compare them like for like because our models
| 64.12 | 76.36 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t70.56
|
or code in general is not going to be able to compare those two different Unicode character
| 70.56 | 77.68 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t76.36
|
sets.
| 76.36 | 83.44 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t77.68
|
And the issue with diacritics is that those characters always have this hidden property
| 77.68 | 89.88 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t83.44000000000001
|
in that we have one Unicode character, which is the capital C with Cedilla.
| 83.44 | 97.38 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t89.88000000000001
|
But then we have an identical set of characters, which is, for example, the Latin capital C
| 89.88 | 101.96 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t97.38000000000001
|
immediately followed by something called the combining Cedilla character.
| 97.38 | 109.56 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t101.96
|
And they together look exactly like the other Unicode character.
| 101.96 | 114.52 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t109.55999999999999
|
And this is quite difficult to deal with.
| 109.56 | 120.72 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t114.52
|
So we have these two problems and we use Unicode normalization to actually deal with those
| 114.52 | 123.2 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t120.72
|
when we're building NLP models.
| 120.72 | 130.16 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t123.19999999999999
|
So I kind of said there's two forms of equivalent characters that are not really equivalent.
| 123.2 | 132.64 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t130.16
|
The first of those is the compatibility equivalence.
| 130.16 | 135.64 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t132.64
|
That's where we have stuff like font variants.
| 132.64 | 141.04 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t135.64
|
We have different line break sequences, circled variants, superscripts, subscripts, fractions
| 135.64 | 143.8 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t141.04
|
and a few other things as well.
| 141.04 | 151.24 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t143.8
|
Now we want our model to see both hello world with those weird circles and also just hello
| 143.8 | 158.04 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t151.24
|
world as one because that's how we read it and that's how it's supposed to be interpreted.
| 151.24 | 163.68 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t158.04
|
And that is what the compatibility equivalence is for.
| 158.04 | 168.52 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t163.67999999999998
|
And we'll look at how we actually deal with that pretty soon.
| 163.68 | 174.44 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t168.51999999999998
|
And then we also have the canonical equivalence, which is the thing with the accents and the
| 168.52 | 176.72 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t174.44
|
glyphs I mentioned before.
| 174.44 | 179.12 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t176.72
|
So you have a few different reasons for that.
| 176.72 | 185.36 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t179.12
|
But two that I think are most relevant is where you have the combined characters.
| 179.12 | 192 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t185.36
|
So we have that C with cedilla character and then we also have the capital C plus the combining
| 185.36 | 195.88 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t192.0
|
cedilla characters merged together.
| 192 | 203 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t195.88000000000002
|
And then we also have conjoined creating characters, which I think are pretty common as well.
| 195.88 | 208.08 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t203.0
|
Canonical equivalence is much more to do with characters that we can't really see that they
| 203 | 211.32 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t208.08
|
are different, but they are in fact different.
| 208.08 | 217.12 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t211.32
|
Whereas compatibility equivalence is more to do with the purpose that made them different
| 211.32 | 221.42 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t217.12
|
and in reality their meaning is the same.
| 217.12 | 228.68 |
Unicode Normalization for NLP in Python
|
2021-03-17 13:30:00 UTC
|
https://youtu.be/9Od9-DV9kd8
|
9Od9-DV9kd8
|
UCv83tO5cePwHMt1952IVVHw
|
9Od9-DV9kd8-t221.42
|
So we have two different directions for how we can transform our text between these two
| 221.42 | 231.32 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.