subtitles/zh-CN/01_the-pipeline-function.srt

1
00:00:00,069 --> 00:00:01,341
（屏幕嗖嗖声）
(screen whooshes)

2
00:00:01,341 --> 00:00:02,449
（面部标志呼啸而过）
(face logo whooshes)

3
00:00:02,449 --> 00:00:05,880
（屏幕嗖嗖声）
(screen whooshes)

4
00:00:05,880 --> 00:00:07,080
- 本节课内容是: pipeline 函数
- The pipeline function.

5
00:00:09,540 --> 00:00:12,020
pipeline 函数是 Transformers 库中的
The pipeline function is the most high level API

6
00:00:12,020 --> 00:00:14,010
最顶层的 API
of the Transformers library.

7
00:00:14,010 --> 00:00:16,050
它将所有步骤重新组合在一起
It regroups together all the steps

8
00:00:16,050 --> 00:00:18,873
从而实现从原始文本到可用预测的转换。
to go from raw texts to usable predictions.

9
00:00:20,228 --> 00:00:22,980
使用的模型是 pipeline 的核心，
The model used is at the core of a pipeline,

10
00:00:22,980 --> 00:00:24,390
但 pipeline 还包括
but the pipeline also include

11
00:00:24,390 --> 00:00:26,610
所有必要的预处理，
all the necessary pre-processing,

12
00:00:26,610 --> 00:00:30,240
因为模型不期望得到文本，而是数字，
since the model does not expect texts, but number,

13
00:00:30,240 --> 00:00:32,040
以及一些后期处理，
as well as some post-processing,

14
00:00:32,040 --> 00:00:34,533
使模型的输出可读。
to make the output of the model human-readable.

15
00:00:35,910 --> 00:00:37,593
让我们通过一个情绪分析 pipeline
Let's look at a first example

16
00:00:37,593 --> 00:00:39,693
的例子来解释一下。
with the sentiment analysis pipeline.

17
00:00:40,740 --> 00:00:44,670
此 pipeline 对给定的输入执行文本分类
This pipeline performs text classification on a given input

18
00:00:44,670 --> 00:00:46,953
并确定它的情绪是正面的还是负面的。
and determines if it's positive or negative.

19
00:00:47,910 --> 00:00:51,750
在这里，它为给定文本标记正面情绪的标签，
Here, it attributed the positive label on the given text,

20
00:00:51,750 --> 00:00:54,413
置信度为 95%。
with a confidence of 95%.

21
00:00:55,650 --> 00:00:58,470
您可以将多个文本传递到同一个 pipeline，
You can pass multiple texts to the same pipeline,

22
00:00:58,470 --> 00:01:00,270
它们将被处理并
which will be processed and passed

23
00:01:00,270 --> 00:01:02,673
作为一个批次传递给模型。
through the model together as a batch.

24
00:01:03,570 --> 00:01:05,970
输出是包含单个结果的列表
The output is a list of individual results

25
00:01:05,970 --> 00:01:07,923
与输入文本的顺序相同。
in the same order as the input texts.

26
00:01:08,790 --> 00:01:12,270
在这里，我们为第一个文本找到了相同的标签和分数，
Here we find the same label and score for the first text,

27
00:01:12,270 --> 00:01:14,443
第二个文本被判断为否定情绪
and the second text is judged negative

28
00:01:14,443 --> 00:01:17,243
置信度为 99.9%。
with a confidence of 99.9%.

29
00:01:18,720 --> 00:01:20,700
零样本分类 pipeline
The zero-shot classification pipeline

30
00:01:20,700 --> 00:01:23,610
是一个更通用的文本分类 pipeline，
is a more general text-classification pipeline,

31
00:01:23,610 --> 00:01:26,370
它允许您提供所需的标签。
it allows you to provide the labels you want.

32
00:01:26,370 --> 00:01:29,850
在这里，我们想根据标签对输入文本进行分类，
Here we want to classify our input text along the labels,

33
00:01:29,850 --> 00:01:32,643
标签有教育、政治和商业。
education, politics, and business.

34
00:01:33,540 --> 00:01:35,580
pipeline 都能够成功识别
The pipeline successfully recognizes

35
00:01:35,580 --> 00:01:38,280
与其他标签相比，它更多地是关于教育，
it's more about education than the other labels,

36
00:01:38,280 --> 00:01:40,643
置信度为 84%。
with a confidence of 84%.

37
00:01:41,670 --> 00:01:43,110
继续执行其他任务，
Moving on to other tasks,

38
00:01:43,110 --> 00:01:45,030
文本生成 pipeline 将
the text generation pipeline will

39
00:01:45,030 --> 00:01:46,533
自动完成给定的提示。
auto-complete a given prompt.

40
00:01:47,460 --> 00:01:49,980
输出带有一点随机性，
The output is generated with a bit of randomness,

41
00:01:49,980 --> 00:01:52,800
所以每次在给定的提示上调用生成器对象时
so it changes each time you call the generator object

42
00:01:52,800 --> 00:01:53,763
这个结果都会改变。
on a given prompt.

43
00:01:54,990 --> 00:01:57,123
到目前为止，我们已经使用了 pipeline API
Up until now, we've used the the pipeline API

44
00:01:57,123 --> 00:02:00,360
结合与每个任务关联的默认模型来演示，
with the default model associated to each task,

45
00:02:00,360 --> 00:02:02,880
但您可以将它与任何经过预训练的模型一起使用
but you can use it with any model that has been pretrained

46
00:02:02,880 --> 00:02:04,263
或根据此任务进行微调。
or fine-tuned on this task.

47
00:02:06,540 --> 00:02:10,350
进入模型中心 huggingface.co/models
Going on the model hub, huggingface.co/models

48
00:02:10,350 --> 00:02:13,350
您可以按任务过滤可用模型。
you can filter the available models by task.

49
00:02:13,350 --> 00:02:17,190
我们之前示例中使用的默认模型是 gpt2，
The default model used in our previous example was gpt2,

50
00:02:17,190 --> 00:02:19,290
但还有更多模型可供选择，
but there are many more models available,

51
00:02:19,290 --> 00:02:20,523
且不仅仅是英语。
and not just in English.

52
00:02:21,450 --> 00:02:23,670
让我们回到文本生成 pipeline
Let's go back to the text generation pipeline

53
00:02:23,670 --> 00:02:26,193
并用另一个模型 distilgpt2 加载它。
and load it with another model, distilgpt2.

54
00:02:27,060 --> 00:02:28,950
这是 gpt2 的轻量级版本
This is a lighter version of gpt2

55
00:02:28,950 --> 00:02:30,603
由 Hugging Face 团队创建。
created by the Hugging Face team.

56
00:02:31,740 --> 00:02:34,110
将 pipeline 应用于给定提示时，
When applying the pipeline to a given prompt,

57
00:02:34,110 --> 00:02:36,360
我们可以指定几个参数
we can specify several arguments

58
00:02:36,360 --> 00:02:39,240
例如生成文本的最大长度，
such as the maximum length of the generated texts,

59
00:02:39,240 --> 00:02:41,700
或者我们想要返回的句子数量，
or the number of sentences we want to return,

60
00:02:41,700 --> 00:02:44,150
因为这一代有一些随机性。
since there is some randomness in the generation.

61
00:02:46,080 --> 00:02:48,750
通过猜测句子中的下一个单词来生成文本
Generating texts by guessing the next word in a sentence

62
00:02:48,750 --> 00:02:51,450
是 GPT-2 的预训练目标。
was the pretraining objective of GPT-2.

63
00:02:51,450 --> 00:02:55,140
掩码填充 pipeline 是 BERT 的预训练目标，
The fill mask pipeline is the pretraining objective of BERT,

64
00:02:55,140 --> 00:02:57,363
这是猜测掩码词的值。
which is to guess the value of masked word.

65
00:02:58,260 --> 00:03:01,020
在这种情况下，我们询问两个最可能的值
In this case, we ask the two most likely values

66
00:03:01,020 --> 00:03:03,660
对于缺失的词，根据模型，
for the missing words, according to the model,

67
00:03:03,660 --> 00:03:07,053
并通过数学计算推测可能的答案。
and get mathematical or computational as possible answers.

68
00:03:08,280 --> 00:03:10,170
Transformers 模型可以执行的另一项任务
Another task Transformers model can perform

69
00:03:10,170 --> 00:03:12,660
就是对句子中的每一个词进行分类
is to classify each word in the sentence

70
00:03:12,660 --> 00:03:14,970
而不是整个句子。
instead of the sentence as a whole.

71
00:03:14,970 --> 00:03:18,390
其中一个例子是命名实体识别，
One example of this is Named Entity Recognition,

72
00:03:18,390 --> 00:03:20,820
这是识别实体的任务，
which is the task of identifying entities,

73
00:03:20,820 --> 00:03:25,323
例如句子中的人、组织或地点。
such as persons, organizations or locations in a sentence.

74
00:03:26,400 --> 00:03:30,570
在这里，模型正确地找到了人 Sylvain，
Here, the model correctly finds the person, Sylvain,

75
00:03:30,570 --> 00:03:32,453
组织，是 Hugging Face
the organization, Hugging Face,

76
00:03:32,453 --> 00:03:35,010
以及位置，布鲁克林，
as well as the location, Brooklyn,

77
00:03:35,010 --> 00:03:36,303
在输入文本中。
inside the input text.

78
00:03:37,661 --> 00:03:40,230
使用的 grouped_entities=True 参数
The grouped_entities=True argument used

79
00:03:40,230 --> 00:03:42,330
就是把 pipeline 组合在一起
is to make the pipeline group together

80
00:03:42,330 --> 00:03:44,790
链接到同一实体的不同词，
the different words linked to the same entity,

81
00:03:44,790 --> 00:03:46,353
比如这里的 Hugging 和 Face。
such as Hugging and Face here.

82
00:03:48,270 --> 00:03:50,670
pipeline API 可用的另一个任务
Another task available with the pipeline API

83
00:03:50,670 --> 00:03:52,920
是抽取式问答。
is extractive question answering.

84
00:03:52,920 --> 00:03:55,380
提供上下文和问题，
Providing a context and a question,

85
00:03:55,380 --> 00:03:58,290
该模型将识别上下文中的文本范围
the model will identify the span of text in the context

86
00:03:58,290 --> 00:04:00,190
包含问题的答案。
containing the answer to the question.

87
00:04:01,650 --> 00:04:03,960
获取长文的简短摘要
Getting short summaries of very long articles

88
00:04:03,960 --> 00:04:06,540
也是 Transformers 库可以提供的功能，
is also something the Transformers library can help with,

89
00:04:06,540 --> 00:04:08,140
这是摘要 pipeline 提供的功能。
with the summarization pipeline.

90
00:04:09,480 --> 00:04:12,570
pipeline API 支持的最后一个任务
Finally, the last task supported by the pipeline API

91
00:04:12,570 --> 00:04:14,130
是翻译。
is translation.

92
00:04:14,130 --> 00:04:16,170
这里我们使用在模型中心 (Model Hub) 提供的
Here we use a French/English model

93
00:04:16,170 --> 00:04:17,460
法语/英语模型
found on the model hub

94
00:04:17,460 --> 00:04:19,893
获取我们输入文本的英文版本。
to get the English version of our input text.

95
00:04:21,600 --> 00:04:23,490
这是所有任务的简要总结
Here is a brief summary of all the tasks

96
00:04:23,490 --> 00:04:25,500
我们在这段视频中进行了调查。
we've looked into in this video.

97
00:04:25,500 --> 00:04:27,390
然后通过模型中心中提供的
Try then out through the inference widgets

98
00:04:27,390 --> 00:04:28,327
推理小部件进行尝试。
in the model hub.

99
00:04:30,459 --> 00:04:33,475
（屏幕嗖嗖声）
(screen whooshes)

100
00:04:33,475 --> 00:04:35,175
（徽标嗖嗖声）
(logo whooshes)