subtitles/zh-CN/04_the-transformer-architecture.srt

1
00:00:00,000 --> 00:00:02,750
（徽标呼啸而过）
(logo whooshing)

2
00:00:05,010 --> 00:00:07,323
- 让我们 Transformer 的架构。
- Let's study the transformer architecture.

3
00:00:09,150 --> 00:00:12,030
该视频是编码器的介绍视频，
This video is the introductory video to the encoders,

4
00:00:12,030 --> 00:00:15,510
解码器和编码器 - 解码器系列视频。
decoders, and encoder-decoder series of videos.

5
00:00:15,510 --> 00:00:16,343
在这个系列中，
In this series,

6
00:00:16,343 --> 00:00:18,900
我们将尝试了解是什么构成了 transformer 网络，
we'll try to understand what makes a transformer network,

7
00:00:18,900 --> 00:00:22,770
我们将尝试用简单、高层次的术语来解释它。
and we'll try to explain it in simple, high-level terms.

8
00:00:22,770 --> 00:00:25,800
无需深入了解神经网络，
No advanced understanding of neural networks is necessary,

9
00:00:25,800 --> 00:00:29,343
但了解基本向量和张量可能会有所帮助。
but an understanding of basic vectors and tensors may help.

10
00:00:32,250 --> 00:00:33,270
开始，
To get started,

11
00:00:33,270 --> 00:00:34,530
我们将处理这张图
we'll take up this diagram

12
00:00:34,530 --> 00:00:36,630
从原来的变压器纸，
from the original transformer paper,

13
00:00:36,630 --> 00:00:40,140
Vaswani 等人题为 “注意力就是你所需要的”。
entitled "Attention Is All You Need", by Vaswani et al.

14
00:00:40,140 --> 00:00:41,010
正如我们将在这里看到的，
As we'll see here,

15
00:00:41,010 --> 00:00:42,780
我们只能利用它的一部分，
we can leverage only some parts of it,

16
00:00:42,780 --> 00:00:44,630
根据我们正在尝试做的事情。
according to what we're trying to do.

17
00:00:45,480 --> 00:00:47,610
我们想深入到特定的层次，
We want to dive into the specific layers,

18
00:00:47,610 --> 00:00:48,990
建立那个架构，
building up that architecture,

19
00:00:48,990 --> 00:00:51,390
但我们会尝试理解不同的方式
but we'll try to understand the different ways

20
00:00:51,390 --> 00:00:52,893
可以使用此架构。
this architecture can be used.

21
00:00:55,170 --> 00:00:56,003
让我们先开始
Let's first start

22
00:00:56,003 --> 00:00:58,260
通过将该架构分成两部分。
by splitting that architecture into two parts.

23
00:00:58,260 --> 00:00:59,910
在左边我们有编码器，
On the left we have the encoder,

24
00:00:59,910 --> 00:01:01,980
右边是解码器。
and on the right, the decoder.

25
00:01:01,980 --> 00:01:03,330
这两个可以一起使用，
These two can be used together,

26
00:01:03,330 --> 00:01:05,330
但它们也可以独立使用。
but they can also be used independently.

27
00:01:06,180 --> 00:01:08,610
让我们了解这些是如何工作的。
Let's understand how these work.

28
00:01:08,610 --> 00:01:11,460
编码器接受表示文本的输入。
The encoder accepts inputs that represent text.

29
00:01:11,460 --> 00:01:13,620
它转换这个文本，这些词，
It converts this text, these words,

30
00:01:13,620 --> 00:01:15,675
成数值表示。
into numerical representations.

31
00:01:15,675 --> 00:01:17,400
这些数值表示
These numerical representations

32
00:01:17,400 --> 00:01:20,460
也可以称为嵌入或特征。
can also be called embeddings, or features.

33
00:01:20,460 --> 00:01:23,100
我们会看到它使用了 self-attention 机制
We'll see that it uses the self-attention mechanism

34
00:01:23,100 --> 00:01:24,483
作为其主要组成部分。
as its main component.

35
00:01:25,500 --> 00:01:27,120
我们建议你查看视频
We recommend you check out the video

36
00:01:27,120 --> 00:01:29,700
关于编码器具体要了解
on encoders specifically to understand

37
00:01:29,700 --> 00:01:31,680
这个数字表示是什么，
what is this numerical representation,

38
00:01:31,680 --> 00:01:33,690
以及它是如何工作的。
as well as how it works.

39
00:01:33,690 --> 00:01:36,660
我们将更详细地研究自注意力机制，
We'll study the self-attention mechanism in more detail,

40
00:01:36,660 --> 00:01:38,913
以及它的双向属性。
as well as its bi-directional properties.

41
00:01:40,650 --> 00:01:42,780
解码器类似于编码器。
The decoder is similar to the encoder.

42
00:01:42,780 --> 00:01:45,630
它还可以接受文本输入。
It can also accept text inputs.

43
00:01:45,630 --> 00:01:48,210
它使用与编码器类似的机制，
It uses a similar mechanism as the encoder,

44
00:01:48,210 --> 00:01:51,150
这也是掩蔽的自我关注。
which is the masked self-attention as well.

45
00:01:51,150 --> 00:01:52,590
它不同于编码器
It differs from the encoder

46
00:01:52,590 --> 00:01:54,990
由于其单向特性
due to its uni-directional feature

47
00:01:54,990 --> 00:01:58,590
并且传统上以自回归方式使用。
and is traditionally used in an auto-regressive manner.

48
00:01:58,590 --> 00:02:01,650
在这里，我们也建议你查看有关解码器的视频
Here too, we recommend you check out the video on decoders

49
00:02:01,650 --> 00:02:04,000
特别是要了解所有这些是如何工作的。
especially to understand how all of this works.

50
00:02:06,810 --> 00:02:07,890
结合两部分
Combining the two parts

51
00:02:07,890 --> 00:02:10,200
结果就是所谓的编码器 - 解码器，
results in what is known as an encoder-decoder,

52
00:02:10,200 --> 00:02:12,720
或序列到序列转换器。
or a sequence-to-sequence transformer.

53
00:02:12,720 --> 00:02:14,280
编码器接受输入
The encoder accepts inputs

54
00:02:14,280 --> 00:02:17,850
并计算这些输入的高级表示。
and computes a high-level representation of those inputs.

55
00:02:17,850 --> 00:02:20,252
然后将这些输出传递给解码器。
These outputs are then passed to the decoder.

56
00:02:20,252 --> 00:02:22,860
解码器使用编码器的输出，
The decoder uses the encoder's output,

57
00:02:22,860 --> 00:02:26,370
与其他输入一起生成预测。
alongside other inputs to generate a prediction.

58
00:02:26,370 --> 00:02:27,900
然后它预测输出，
It then predicts an output,

59
00:02:27,900 --> 00:02:30,248
它将在未来的迭代中重复使用，
which it will re-use in future iterations,

60
00:02:30,248 --> 00:02:32,662
因此，术语自回归。
hence the term, auto-regressive.

61
00:02:32,662 --> 00:02:34,740
最后，为了理解
Finally, to get an understanding

62
00:02:34,740 --> 00:02:36,690
编码器 - 解码器作为一个整体，
of the encoder-decoders as a whole,

63
00:02:36,690 --> 00:02:39,670
我们建议你查看有关编码器 - 解码器的视频。
we recommend you check out the video on encoder-decoders.

64
00:02:39,670 --> 00:02:42,420
（徽标呼啸而过）
(logo whooshing)