title
stringlengths
12
112
published
stringlengths
19
23
url
stringlengths
28
28
video_id
stringlengths
11
11
channel_id
stringclasses
5 values
id
stringlengths
16
31
text
stringlengths
0
596
start
float64
0
37.8k
end
float64
2.18
37.8k
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t656.12
We have our data loader.
656.12
660
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t658.32
What is the name of that data loader?
658.32
660.52
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t660.0
I'm not sure.
660
665.36
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t664.24
Data loader.
664.24
667.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t665.36
Cool.
665.36
669.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t667.8
Data loader.
667.8
672.12
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t669.4
And we set leave equals true.
669.4
676.32
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t675.64
But I need that.
675.64
676.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t676.3199999999999
Sorry.
676.32
677.92
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t676.8
I need that in the same style.
676.8
682.88
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t677.92
So for batch in loop.
677.92
690.96
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t686.7199999999999
And then here, we run through each of the steps
686.72
694.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t690.9599999999999
that we're going to perform for every single training loop.
690.96
699.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t694.8
So the first thing we do is initialize the gradient
694.8
700.96
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t699.8
in our optimizer.
699.8
702.64
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t700.9599999999999
So zero grad.
700.96
706.56
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t702.64
So the reason we do this is after the first loop,
702.64
708.84
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t706.56
our optimizer is going to be assigned
706.56
710.76
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t708.8399999999999
a set of gradients, which we're going
708.84
714.24
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t710.76
to use to optimize our model.
710.76
718.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t714.24
And on the next loop, we don't want those residual gradients
714.24
720.92
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t718.8
to still be there in our optimizer.
718.8
723.48
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t720.92
We want to essentially reset it for the next loop.
720.92
726.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t723.48
So that's what we're doing here.
723.48
728.76
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t726.4
Then we want our tensors.
726.4
732.2
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t728.76
So we have input IDs.
728.76
738.64
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t732.2
And that is going to be batch input IDs.
732.2
742.92
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t738.64
And we also want to move that over to our GPU or CPU
738.64
746.24
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t742.9200000000001
if you're on that.
742.92
750.76
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t746.24
And this is pretty much the same for our three.
746.24
754.6
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t750.76
So mask labels.
750.76
761.04
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t758.4
And this is just attention mask.
758.4
764.68
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t761.04
So we've extracted our tensors.
761.04
767
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t764.68
And we just need to feed them into our model now.
764.68
770.56
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t767.0
So we're going to get outputs from the model.
767
772.2
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t770.56
We just do model.
770.56
778.52
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t772.2
Input IDs, attention mask, which is going to be equal to mask.
772.2
783.16
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t778.52
And our labels equal to labels.
778.52
789.56
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t786.28
So everything has been fed into our model.
786.28
790.6
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t789.56
We have our outputs.
789.56
794.04
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t790.6
Now we need to extract a few things from the output.
790.6
796.52
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t794.04
So we need the loss.
794.04
800.28
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t796.52
So we write loss equals outputs.loss.
796.52
804.36
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t800.28
And from that, we want to calculate
800.28
807.92
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t804.36
all of the different parameters in our model.
804.36
810.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t807.92
We need to calculate the loss for each one of those parameters.
807.92
815.44
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t810.4
So we do this loss.backwards to backpropagate
810.4
818.44
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t815.4399999999999
through all of those different values and get that loss.
815.44
825.88
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t820.9599999999999
After we've done that, we use our optimizer, take a step,
820.96
831.16
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t825.88
and optimize all those parameters based on that loss.
825.88
833.72
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t831.16
Then that's everything we need to train the model.
831.16
834.96
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t833.7199999999999
And there's just a few things.
833.72
837
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t834.9599999999999
So for the progress bar, I just want
834.96
840.04
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t837.0
a little bit of information there just so I know what's going on.
837
845.36
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t840.04
And I just write loop set description.
840.04
848.48
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t845.36
And I just want to print out the epoch.
845.36
850.08
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t848.48
So write that.
848.48
856.44
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t853.68
And then I want to set the postfix as well.
853.68
861
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t856.44
So loop.set postfix.
856.44
862.48
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t861.0
And here I just want to see the loss.
861
866.08
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t862.48
So we'll just do loss item like that.
862.48
869.6
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t866.08
So that should be everything.
866.08
875.16
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t869.6
Yeah, let's run that, see what happens.
869.6
878
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t875.1600000000001
Hopefully it should work.
875.16
880.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t878.0
No, it didn't work.
878
882.24
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t880.4000000000001
OK.
880.4
883.08
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t882.24
Let me see.
882.24
887.84
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t886.64
Oh, no, it's a CUDA error.
886.64
895.6
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t890.8000000000001
So probably just need to refresh everything.
890.8
897.04
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t895.6
I hate CUDA errors.
895.6
898.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t897.0400000000001
One moment.
897.04
901.72
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t898.8000000000001
OK, so finally figured it out.
898.8
904.2
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t901.72
Took so long.
901.72
906.48
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t904.2
So a few tips anyway.
904.2
910.68
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t906.48
When you do get a CUDA error, switch your device to CPU,
906.48
912.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t910.6800000000001
and then re-run everything.
910.68
915.64
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t912.8000000000001
And you should get a more understandable error.
912.8
918.16
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t915.64
So if we come down here, I've changed it to CPU.
915.64
919.96
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t918.16
We see that we get an index error.
918.16
922.6
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t919.96
Scroll down, index out of range itself.
919.96
928.48
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t922.6
So the reason for this is, so you get this error
922.6
932.52
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t928.48
if you don't add the extra two tokens onto the end of here.
928.48
935.08
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t932.52
But we added them.
932.52
938.08
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t935.08
So I was pretty confused about that.
935.08
940.12
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t938.08
And then it took me a really long time
938.08
943.28
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t940.12
to realize that this argument is wrong.
940.12
944.8
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t943.28
And there should be an S on the end.
943.28
948.12
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t944.8000000000001
So that was the error.
944.8
951.12
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t948.12
So yeah, super, super cool.
948.12
952.72
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t951.12
That was literally it.
951.12
957.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t952.72
It took me so long to figure that out.
952.72
958.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t957.4
But now we have it.
957.4
959.64
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t958.4
That's good.
958.4
961.32
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t959.64
We just need to run everything again.
959.64
964.28
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t961.32
So I'm just going to run through everything.
961.32
967.68
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t964.28
Remove this cell here where I changed it to CPU,
964.28
968.68
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t967.68
because I don't need it now.
967.68
973.76
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t972.04
And just re-execute all that.
972.04
978.92
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t977.64
OK, so we're back.
977.64
981.4
Training and Testing an Italian BERT - Transformers From Scratch #4
2021-07-06 13:00:03 UTC
https://youtu.be/35Pdoyi6ZoQ
35Pdoyi6ZoQ
UCv83tO5cePwHMt1952IVVHw
35Pdoyi6ZoQ-t978.92
And we've finished training our model now.
978.92
983.76