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-t2433.6
|
for the very first item every single time. So let's just update that. So
| 2,433.6 | 2,454.92 |
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-t2442.4
|
obviously that won't get us very far. And just update that as well. And now this
| 2,442.4 | 2,467.24 |
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-t2454.92
|
should look a little bit better. So it's lucky we checked. Okay so our data at the
| 2,454.92 | 2,473.04 |
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-t2467.2400000000002
|
moment is in the right format. We just need to use it to create a PyTorch
| 2,467.24 | 2,484.64 |
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-t2473.04
|
dataset object. So to do that obviously we need to import PyTorch. And we define
| 2,473.04 | 2,497.64 |
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-t2484.64
|
that dataset using a class. I'm just passing the torch. utils data dataset.
| 2,484.64 | 2,519.88 |
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-t2497.64
|
We need to initialize that. Like so. And this is coming from the Houden Face
| 2,497.64 | 2,536.36 |
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-t2519.88
|
Transformers documentation. Don't take credit for this. And we essentially need
| 2,519.88 | 2,541.12 |
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-t2536.36
|
to do this so that we can load in our data using the PyTorch data loader
| 2,536.36 | 2,566.48 |
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-t2541.12
|
later on. Which makes things incredibly easy. And then we just have one more
| 2,541.12 | 2,592.52 |
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-t2566.48
|
function here. Or method. Okay. And return. And also this as well. That should be
| 2,566.48 | 2,609.6 |
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-t2592.52
|
okay. So we apply this to our datasets to create dataset objects. Now our encodings.
| 2,592.52 | 2,623.48 |
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-t2609.6
|
And then the same again for the validation set. Okay so that is our data.
| 2,609.6 | 2,629.24 |
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-t2623.48
|
Almost fully prepared. All we do now is load it into a data loader object. But
| 2,623.48 | 2,634.6 |
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-t2629.24
|
this is everything on the data side done. Which is great because I know this bit
| 2,629.24 | 2,639.8 |
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-t2634.6
|
does take some time. And I know it's not the most interesting part of it. But it's
| 2,634.6 | 2,644.8 |
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-t2639.7999999999997
|
just something that we need to do. And need to understand what we're doing as
| 2,639.8 | 2,653.44 |
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-t2644.7999999999997
|
well. So now we get to the more interesting bit. So we'll just add the
| 2,644.8 | 2,666.36 |
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-t2653.44
|
imports in here. So we need our data loader.
| 2,653.44 | 2,680.96 |
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-t2673.56
|
We're going to import the Adam optimizer with weighted decay. Which is pretty
| 2,673.56 | 2,686.48 |
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-t2680.96
|
commonly used for transformer models when you are fine-tuning. Because
| 2,680.96 | 2,692.12 |
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-t2686.48
|
transformer models are generally very large models. And they can over fit very
| 2,686.48 | 2,699.32 |
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-t2692.12
|
easily. So this Adam optimizer with weighted decay essentially just reduces
| 2,692.12 | 2,706.96 |
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-t2699.32
|
the chances of that happening. Which is supposed to be very useful and quite
| 2,699.32 | 2,723.76 |
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-t2706.96
|
important. So obviously we're going to use that. And then final bit is TQDM. So
| 2,706.96 | 2,730.6 |
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-t2725.56
|
TQDM is a progress bar that we are going to be using. So that we can actually see
| 2,725.56 | 2,735.36 |
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-t2730.6
|
the progress of our training. Otherwise we're just going to sit there for
| 2,730.6 | 2,740.92 |
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-t2735.36
|
probably quite a long time not knowing what is actually happening. And trust me
| 2,735.36 | 2,744.12 |
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-t2740.92
|
it won't take long before you start questioning whether anything is
| 2,740.92 | 2,750.56 |
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-t2744.1200000000003
|
happening. Because it takes a long time to train these models. So they are our
| 2,744.12 | 2,758.6 |
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-t2750.56
|
imports. And I'm being stupid again here. That's from, did that twice. Okay so
| 2,750.56 | 2,764.92 |
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-t2758.6
|
that's all good. So now we just need to do a few little bits for the setup. So
| 2,758.6 | 2,771.88 |
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-t2764.92
|
we need to tell Pytorch whether we're using CPU or GPU. In my case it will be a
| 2,764.92 | 2,777.52 |
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-t2771.88
|
GPU. If you're using CPU this is going to take you a very long time to train. And
| 2,771.88 | 2,786 |
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-t2777.52
|
it's still going to take you a long time on GPU. So just be aware of that. But what
| 2,777.52 | 2,804.72 |
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-t2786.0
|
we're going to do here is say device. It's CUDA. If CUDA is available. Otherwise
| 2,786 | 2,815.6 |
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-t2804.72
|
we are going to use the CPU. And good luck if that is what you're doing. So once
| 2,804.72 | 2,825.96 |
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-t2815.6
|
we've defined the device we want to move our model over to it. So we just.model.to
| 2,815.6 | 2,831.04 |
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-t2825.96
|
device. So this.to method is essentially a way of transferring data between
| 2,825.96 | 2,837.44 |
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-t2831.04
|
different hardware components. So your CPU or GPU. It's quite useful. And then we
| 2,831.04 | 2,844.6 |
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-t2837.44
|
want to activate our model for training. So there's two things we have here. So we
| 2,837.44 | 2,851.96 |
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-t2844.6
|
have.train and eval. So when we're in train mode there's a lot of
| 2,844.6 | 2,855.4 |
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-t2851.96
|
different layers and different parts of your model that will behave differently
| 2,851.96 | 2,860 |
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-t2855.4
|
depending on whether you are using the model for training or you're using it
| 2,855.4 | 2,865.16 |
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-t2860.0
|
for inference which is predictions. So we just need to make sure our model is in
| 2,860 | 2,870.92 |
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-t2865.16
|
the right mode for whatever we're doing. And later on we'll switch it to eval to
| 2,865.16 | 2,876.64 |
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-t2870.92
|
make some predictions. So that's almost everything. So we just need to
| 2,870.92 | 2,885.36 |
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-t2876.64
|
initialize the optimizer. And here we're using the weighted decay Adam optimizer.
| 2,876.64 | 2,894.76 |
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-t2885.64
|
We need to pass in our model parameters and also give it a learning rate. And
| 2,885.64 | 2,902.92 |
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-t2894.76
|
we're going to use this value here. All these are the recommended parameters for
| 2,894.76 | 2,913.28 |
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-t2902.92
|
what we are doing here. So the one thing that I have somehow missed is defining
| 2,902.92 | 2,919.16 |
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-t2913.28
|
the actual initializing the model. So let's just add that in. And all we're
| 2,913.28 | 2,923.96 |
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-t2919.1600000000003
|
doing here is loading again a pre-trained one. So like we did before
| 2,919.16 | 2,936.68 |
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-t2923.96
|
when we were loading the transformers tokenizer. This time it's for question
| 2,923.96 | 2,944.6 |
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-t2936.68
|
answering. So this the Silbert of question answering is a the Silbert model with a
| 2,936.68 | 2,950.4 |
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-t2944.6
|
question and answering head added on to the end of it. So essentially with
| 2,944.6 | 2,953.68 |
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-t2950.4
|
transformers you have all these different heads that you add on and they
| 2,950.4 | 2,962.68 |
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-t2953.68
|
will do different things depending on what head it has on there. So let's
| 2,953.68 | 2,973.6 |
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-t2962.68
|
initialize that from pre-trained. And we're using the same one we use up here
| 2,962.68 | 2,989.2 |
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-t2973.6
|
which is the Silbert base uncased. And sometimes you will need to download that.
| 2,973.6 | 2,993.12 |
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-t2989.2
|
Fortunately I don't need to as I've already done that but this can also take
| 2,989.2 | 2,996.92 |
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-t2993.12
|
a little bit of time. Not too long though and you get a nice progress bar
| 2,993.12 | 3,005.52 |
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-t2996.92
|
hopefully as well. Okay so now that is all set up we can initialize our data
| 2,996.92 | 3,020.36 |
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-t3005.52
|
loader. So all we're doing here is using the PyTorch data loader object and we
| 3,005.52 | 3,027.68 |
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-t3020.36
|
just pass in our training data set. The batch size so how many we want to train
| 3,020.36 | 3,033.52 |
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-t3027.6800000000003
|
on at once in parallel before updating the model weights which will be 16. And
| 3,027.68 | 3,039.04 |
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-t3033.52
|
we also would like to shuffle the data because we don't want to train the model
| 3,033.52 | 3,042.4 |
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-t3039.04
|
on a single batch and it just learned about Beyonce and then the next one
| 3,039.04 | 3,047.6 |
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-t3042.4
|
it's learning about Chopin and it will keep switching between those but never
| 3,042.4 | 3,058.56 |
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-t3047.6
|
within a single batch having a good mix of different things to learn about. So it
| 3,047.6 | 3,065.8 |
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-t3058.56
|
is data set seems a bit of a weird name to me so I'm just going to change it.
| 3,058.56 | 3,081.12 |
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-t3065.8
|
And they also can't spell. There we go. And that is everything we can actually
| 3,065.8 | 3,095.36 |
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-t3081.1200000000003
|
begin our training loop. So we're gonna go for three parts and what we want to
| 3,081.12 | 3,103.36 |
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-t3095.36
|
start with here is a loop object. So we do this mainly because we're using TQDM
| 3,095.36 | 3,108.96 |
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-t3103.36
|
as a progress bar otherwise we wouldn't need to do this. There would be no point in
| 3,103.36 | 3,116.84 |
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-t3108.96
|
doing it and all this is doing is kind of like pre-initializing our loop that we
| 3,108.96 | 3,120.16 |
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-t3116.84
|
are going to go through. So we're going to obviously loop through every batch
| 3,116.84 | 3,128.12 |
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-t3120.16
|
within the train loader so we just add that in here and then there's this other
| 3,120.16 | 3,134.64 |
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-t3128.12
|
parameter which I don't know if we... So let's leave it but essentially you can
| 3,128.12 | 3,138.64 |
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-t3134.64
|
add a leave equals true in order to leave your progress bar in the same
| 3,134.64 | 3,143.48 |
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-t3138.64
|
place with every epoch. Whereas at the moment with every epoch what it will do
| 3,138.64 | 3,148.6 |
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-t3143.48
|
is create a new progress bar. We are going to create a new progress bar but
| 3,143.48 | 3,151.68 |
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-t3148.6
|
if you don't do that and you want it to just stay in the same place you add
| 3,148.6 | 3,160.24 |
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-t3151.68
|
leave equals true into this function here. So after that we need to go through
| 3,151.68 | 3,168.16 |
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-t3160.24
|
each batch within our loop and the first thing that we need to do is set all of
| 3,160.24 | 3,177.36 |
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-t3168.16
|
our calculated gradients to zero. So with every iteration that we go through here
| 3,168.16 | 3,181.48 |
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-t3177.36
|
or every batch at the end of it we are going to calculate gradients which tells
| 3,177.36 | 3,188.72 |
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-t3181.48
|
the model in which direction to change the weights within the model and
| 3,181.48 | 3,193.48 |
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-t3188.7200000000003
|
obviously when we go into the next iteration we don't want those gradients
| 3,188.72 | 3,198.84 |
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-t3193.48
|
to still be there. So all we're doing here is reinitializing those gradients
| 3,193.48 | 3,202.92 |
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-t3198.84
|
at the start of every loop so we have a fresh set of gradients to work with
| 3,198.84 | 3,211.44 |
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-t3202.92
|
every time and here we just want to pull in our data. So this is everything that
| 3,202.92 | 3,218.72 |
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-t3211.44
|
is relevant that we're going to be feeding into the training process. So
| 3,211.44 | 3,225.8 |
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-t3218.7200000000003
|
everything within our batch and then in here we have all of our different items.
| 3,218.72 | 3,242.12 |
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-t3225.8
|
So we can actually see go here we want to add in all of these and we also want
| 3,225.8 | 3,249.16 |
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-t3242.1200000000003
|
to move them across to the GPU in my case or whatever device you are working
| 3,242.12 | 3,257 |
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-t3249.16
|
on. I would do that for the attention mass start positions and end positions.
| 3,249.16 | 3,282.08 |
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-t3257.0
|
So these start and end positions are essentially the labels that are targets
| 3,257 | 3,289.32 |
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-t3282.08
|
that we want our model to optimize for and the input IDs and attention mass are
| 3,282.08 | 3,311.32 |
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-t3289.3199999999997
|
the inputs. So now we have those defined we just need to feed them into our model
| 3,289.32 | 3,318.56 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.