puffy310/yandset · Datasets at Hugging Face

text stringlengths 0 339
Tired in I suggest getting some sleep after all
You're going away in morning
Oh, you are it? Decided did to change your mind below a sudden? Oh, but I thought you were totally ready to set off
You would dead set on that warranty
You want to stay here
With me, I'm so shocked that preposition How long would that be for that you want to stay with me? Forever, that is so kind of you
Do you want to keep me company here
For forever
We don't have to be lonely anymore
Do we
Not to we have each other
There's just such a shame
That it had to take a left potion
To bind and love together
I was feeling slightly bad that I tricked you into taking this patient, but There was no other way if I told you the truth I it was see potion to make you fall love with me
You wouldn't have taken and you would have missed out And that would have been a very big shape
So I hope you're thankful for me doing that
I don't think you mine this at home
And now we have company
Someone does stay with me and to talk to give me a love and affection
It's all I have
I'm going to look after you
And carefully
I hope you know of course, and protect you from everything and anything that tries to come between
Q like you didn't end up in my part of the forest fastest asleep by accident
No
It is destiny
It was meant to be
Let's be honest
You're out there just asking to be taken away by me
Maybe one day
I will tell you about what actually happened to this human
But do you would just lucky enough to be so cute and catch me and my little struggle of companionship the past few
Come here
My dear
You are looking very like
I think I shall good you just sleep now
We are together after all
It wouldn't be weird if I joined you either
Also, we have a very busy day tomorrow
Going to create some memories together
Teach you the way of the forest
While doing lots of lovely, things on top of that of course
I have a lot to learn about that
I'm sure you can me
Come on my dear
Try to find the trousers
Go to sleep It will make you feel
When wake, you'll be feeling fit
And then a energized to
I love my dearAttention Is All You Need
Ashish Vaswani∗
Google Brain
avaswani@google.com
Noam Shazeer∗
Google Brain
noam@google.com
Niki Parmar∗
Google Research
nikip@google.com
Jakob Uszkoreit∗
Google Research
usz@google.com
Llion Jones∗
Google Research
llion@google.com
Aidan N. Gomez∗ †
University of Toronto
aidan@cs.toronto.edu
Łukasz Kaiser∗
Google Brain
lukaszkaiser@google.com
Illia Polosukhin∗ ‡
illia.polosukhin@gmail.com
Abstract
The dominant sequence transduction models are based on complex recurrent or
convolutional neural networks that include an encoder and a decoder. The best
performing models also connect the encoder and decoder through an attention
mechanism. We propose a new simple network architecture, the Transformer,
based solely on attention mechanisms, dispensing with recurrence and convolutions
entirely. Experiments on two machine translation tasks show these models to
be superior in quality while being more parallelizable and requiring significantly
less time to train. Our model achieves 28.4 BLEU on the WMT 2014 Englishto-German translation task, improving over the existing best results, including
ensembles, by over 2 BLEU. On the WMT 2014 English-to-French translation task,
our model establishes a new single-model state-of-the-art BLEU score of 41.8 after
training for 3.5 days on eight GPUs, a small fraction of the training costs of the
best models from the literature. We show that the Transformer generalizes well to
other tasks by applying it successfully to English constituency parsing both with
large and limited training data.
1 Introduction
Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neural networks
in particular, have been firmly established as state of the art approaches in sequence modeling and
∗Equal contribution. Listing order is random. Jakob proposed replacing RNNs with self-attention and started
the effort to evaluate this idea. Ashish, with Illia, designed and implemented the first Transformer models and
has been crucially involved in every aspect of this work. Noam proposed scaled dot-product attention, multi-head
attention and the parameter-free position representation and became the other person involved in nearly every
detail. Niki designed, implemented, tuned and evaluated countless model variants in our original codebase and
tensor2tensor. Llion also experimented with novel model variants, was responsible for our initial codebase, and
efficient inference and visualizations. Lukasz and Aidan spent countless long days designing various parts of and
implementing tensor2tensor, replacing our earlier codebase, greatly improving results and massively accelerating

Tired in I suggest getting some sleep after all

You're going away in morning

Oh, you are it? Decided did to change your mind below a sudden? Oh, but I thought you were totally ready to set off

You would dead set on that warranty

You want to stay here

With me, I'm so shocked that preposition How long would that be for that you want to stay with me? Forever, that is so kind of you

Do you want to keep me company here

For forever

We don't have to be lonely anymore

Do we

Not to we have each other

There's just such a shame

That it had to take a left potion

To bind and love together

I was feeling slightly bad that I tricked you into taking this patient, but There was no other way if I told you the truth I it was see potion to make you fall love with me

You wouldn't have taken and you would have missed out And that would have been a very big shape

So I hope you're thankful for me doing that

I don't think you mine this at home

And now we have company

Someone does stay with me and to talk to give me a love and affection

It's all I have

I'm going to look after you

And carefully

I hope you know of course, and protect you from everything and anything that tries to come between

Q like you didn't end up in my part of the forest fastest asleep by accident

No

It is destiny

It was meant to be

Let's be honest

You're out there just asking to be taken away by me

Maybe one day

I will tell you about what actually happened to this human

But do you would just lucky enough to be so cute and catch me and my little struggle of companionship the past few

Come here

My dear

You are looking very like

I think I shall good you just sleep now

We are together after all

It wouldn't be weird if I joined you either

Also, we have a very busy day tomorrow

Going to create some memories together

Teach you the way of the forest

While doing lots of lovely, things on top of that of course

I have a lot to learn about that

I'm sure you can me

Come on my dear

Try to find the trousers

Go to sleep It will make you feel

When wake, you'll be feeling fit

And then a energized to

I love my dearAttention Is All You Need

Ashish Vaswani∗

Google Brain

avaswani@google.com

Noam Shazeer∗

Google Brain

noam@google.com

Niki Parmar∗

Google Research

nikip@google.com

Jakob Uszkoreit∗

Google Research

usz@google.com

Llion Jones∗

Google Research

llion@google.com

Aidan N. Gomez∗ †

University of Toronto

aidan@cs.toronto.edu

Łukasz Kaiser∗

Google Brain

lukaszkaiser@google.com

Illia Polosukhin∗ ‡

illia.polosukhin@gmail.com

Abstract

The dominant sequence transduction models are based on complex recurrent or

convolutional neural networks that include an encoder and a decoder. The best

performing models also connect the encoder and decoder through an attention

mechanism. We propose a new simple network architecture, the Transformer,

based solely on attention mechanisms, dispensing with recurrence and convolutions

entirely. Experiments on two machine translation tasks show these models to

be superior in quality while being more parallelizable and requiring significantly

less time to train. Our model achieves 28.4 BLEU on the WMT 2014 Englishto-German translation task, improving over the existing best results, including

ensembles, by over 2 BLEU. On the WMT 2014 English-to-French translation task,

our model establishes a new single-model state-of-the-art BLEU score of 41.8 after

training for 3.5 days on eight GPUs, a small fraction of the training costs of the

best models from the literature. We show that the Transformer generalizes well to

other tasks by applying it successfully to English constituency parsing both with

large and limited training data.

1 Introduction

Recurrent neural networks, long short-term memory [13] and gated recurrent [7] neural networks

in particular, have been firmly established as state of the art approaches in sequence modeling and

∗Equal contribution. Listing order is random. Jakob proposed replacing RNNs with self-attention and started

the effort to evaluate this idea. Ashish, with Illia, designed and implemented the first Transformer models and

has been crucially involved in every aspect of this work. Noam proposed scaled dot-product attention, multi-head

attention and the parameter-free position representation and became the other person involved in nearly every

detail. Niki designed, implemented, tuned and evaluated countless model variants in our original codebase and

tensor2tensor. Llion also experimented with novel model variants, was responsible for our initial codebase, and

efficient inference and visualizations. Lukasz and Aidan spent countless long days designing various parts of and

implementing tensor2tensor, replacing our earlier codebase, greatly improving results and massively accelerating