Training spaCy’s Statistical Models訓練spaCy模型
This guide describe show to train new statistical models for spaCy's part-of-speech tagger, named entity recognizer and dependency parser. Once the model is trained, you can then save and load it.
本指南介紹如何訓練新的spaCy模型：詞性標注器，NER和依存句法分析模型。模型訓練完成后可以存儲和加載。
Training basics 訓練基礎
spaCy's models are statistical and every "decision" they make – for example, which part-of-speech tag to assign, or whether a word is a named entity – is a prediction. This prediction is based on the examples the model has seen during training. To train a model, you first need training data – examples of text, and the labels you want the model to predict. This could be a part-of-speech tag, a named entity or any other information.
spaCy的模型是統計學的，作出的每一個“判別”都是預測，例如：詞性標注，或者是否命名實體。其預測基于模型在訓練過程中見過的樣本。訓練一個模型，首先需要訓練數據（文本樣本），以及希望模型預測出的標記。可以是詞性標簽，命名實體或其他信息。
The model is then shown the unlabelled text and will make a prediction. Because we know the correct answer, we can give the model feedback on its prediction in the form of an error gradient of the loss function that calculates the difference between the training example and the expected output. The greater the difference, the more significant the gradient and the updates to our model.
之后，模型會找出未標記的文本并作出預測。因為我們知道真確答案，就可以給模型計算輸出的錯誤結果反饋其與預期輸出的偏差。差異越大，對模型的提升更重要。
Training data: Examples and their annotations. 樣本及其注釋。
Text: The input text the model should predict a label for. 模型應預測出的標記內容。
Label: The label the model should predict. 模型應預測出的標記。
Gradient: Gradient of the loss function calculating the difference between input and expected
output. 損失函數計算的輸入值和預期輸出之間的差異度
When training a model, we don't just want it to memorise our examples – we want it to come up with theory that can be generalised across other examples. After all, we don't just want the model to learn that this one instance of "Amazon" right here is a company – we want it to learn that "Amazon", in contexts like this, is most likely a company. That's why the training data should always be representative of the data we want to process. A model trained on Wikipedia, where sentences in the first person are extremely rare, will likely perform badly on Twitter. Similarly, a model trained on romantic novels will likely perform badly on legal text.
訓練模型時，不僅僅希望其記住樣本，還希望模型能夠進行廣義的跨樣本推測。畢竟我們不僅僅希望模型學到Amazon在這里公司這么一個實例，還希望它能夠學到Amazon在這樣的上下文語境中，最可能是一個公司。這就是為什么訓練數據對于要處理的數據來說應該具有代表性。用維基數據訓練的模型，在句子中第一人稱極為罕見，那么該模型在Twitter中很可能會表現不佳。同樣的，用言情小說訓練出的模型，在法律文本中也很可能表現不佳。
This also means that in order to know how the model is performing, and whether it's learning the right things, you don't only need training data – you'll also need evaluation data. If you only test the model with the data it was trained on, you'll have no idea how well it's generalising. If you want to train a model from scratch, you usually need at least a few hundred examples for both training and evaluation. To update an existing model, you can already achieve decent results with very few examples – as long as they're representative.
這也意味著為了了解模型的效果如何，是否學習了正確的內容，不僅需要訓練數據，還需要評估數據。如果只用其訓練數據進行測試，不會知道其表現如何。如果想從零開始訓練模型，通常需要至少幾百個訓練和評估樣本。要更新一個已有模型，可以用很少的樣本（只要具有代表性）即可獲得好的效果。
How do I get training data? 如何獲取訓練數據
Collecting training data may sound incredibly painful – and it can be, if you're planning a large-scale annotation project. However, if your main goal is to update an existing model's predictions – for example, spaCy's named entity recognition –the hard part is usually not creating the actual annotations. It's finding representative examples and extracting potential candidates. The good news is, if you've been noticing bad performance on your data, you likely already have some relevant text, and you can use spaCy to bootstrap a first set of training examples. For example, after processing a few sentences, you may end up with the following entities, some correct, some incorrect.
收集訓練數據可能聽起來非常的痛苦(而且是如果計劃大規模的標注目標的話，確實痛苦)。然而，如果主要目標是升級已有模型的預測能力，例如：spaCy的NER，最難的部分通常不是創建實際的標注。好消息是，如果曾注意到過數據的不良效果，那么看來已經有了一些相關文本，就可以用spaCy來構建第一組訓練樣本。例如：處理一批段落后，可以以一些正確和不正確的實體結束。
HOW MANY EXAMPLES DO I NEED? 需要多少樣本
As a rule of thumb, you should allocate at least 10% of your project resources to creating training and evaluation data. If you're looking to improve an existing model, you might be able to start off with only a handful of examples. Keep in mind that you'll always want a lot more than that for evaluation – especially previous errors the model has made. Otherwise, you won't be able to sufficiently verify that the model has actually made the correct generalisations required for your use case.
按經驗估計，應該用項目數據資源的10%來創建訓練和評估數據。如果要升級已有模型，或許可以僅從極少數樣本開始。要記住始終需要更多評估，特別是模型之前發生過的錯誤。否則，不能夠充分驗證模型確實對指定情況做出了正確處理。
Alternatively, the rule-based matcher can be a useful tool to extract tokens or combinations of tokens, as well as their start and end index in a document. In this case, we'll extract mentions of Google and assume they're an ORG.
或者，提取tokens或tokens組及其在文檔中開始和結束的位置時，基于規則的匹配是比較有用的。下例中是對涉及Google的提取，并識別為ORG的結果。
Based on the few examples above, you can already create six training sentences with eight entities in total. Of course, what you consider a "correct annotation" will always depend on what you want the model to learn. While there are some entity annotations that are more or less universally correct –like Canada being a geopolitical entity – your application may have its very own definition of the NER annotation scheme.
基于上述樣例，已經能夠創建用8個實體創建6條訓練語句了。當然，想要“正確的注釋”，始終取決于想要模型學習的是什么。雖然有些實體很容易正確標注（例如Canada為地理實體），應用還是可以有自己私有的NER標注體系定義。
樣例：
train_data = [
???("Uber blew through $1 million a week", [(0, 4, 'ORG')]),
???("Android Pay expands to Canada", [(0, 11, 'PRODUCT'), (23,30, 'GPE')]),
???("Spotify steps up Asia expansion", [(0, 8, "ORG"),(17, 21, "LOC")]),
???("Google Maps launches location sharing", [(0, 11,"PRODUCT")]),
???("Google rebrands its business apps", [(0, 6,"ORG")]),
???("look what i found on google! ??", [(21, 27,"PRODUCT")])]
TIP: TRY THE PRODIGY ANNOTATION TOOL
If you need to label a lot of data, check out Prodigy, a new, active learning-powered annotation tool we've developed. Prodigy is fast and extensible, and comes with a modern web application that helps you collect training data faster. It integrates seamlessly with spaCy, pre-selects the most relevant examples for annotation, and lets you train and evaluate ready-to-use spaCy models.
ExplosionAI的自推廣告：試試標注工具Prodigy（https://prodi.gy/）
確實是個很牛的東西，不過收費。是Explosion自己做的收費工具，必要情況下買吧，個人版$349，商業版$449。
Training with annotations 用標注訓練
The GoldParse object collects the annotated training examples, also called the gold standard. It's initialised with the Doc object it refers to, and keyword arguments specifying the annotations, like tags or entities. Its job is to encode the annotations, keep them aligned and create the C-level data structures required for efficient access. Here's an example of a simple GoldParse for part-of-speech tags:
GoldParser對象收集了標注訓練樣本，也叫goldstandard。其以Doc對象進行初始化，并且關鍵詞參數指定了標注，比如標簽和實體。其工作是將標注進行編碼，對齊并創建高效訪問需要的C級數據結構。
vocab = Vocab(tag_map={'N': {'pos':'NOUN'}, 'V': {'pos': 'VERB'}})
doc = Doc(vocab, words=['I', 'like','stuff'])
gold = GoldParse(doc, tags=['N', 'V','N'])
Using the Doc and its gold-standard annotations, the model can be updated to learn a sentence of three words with their assigned part-of-speech tags. The tag map is part of the vocabulary and defines the annotation scheme. If you're training a new language model, this will let you map the tags present in the treebank you train on to spaCy's tag scheme.
使用Doc對象及其gold-standard標注，模型可以用其詞性標簽升級到學習三字短語。tag map是詞匯和定義標注體系的一部分。如果訓練新語言模型，tag map會以spaCy標簽體系映射訓練的treebank里的標簽。
doc = Doc(Vocab(), words=['Facebook','released', 'React', 'in', '2014'])
gold = GoldParse(doc, entities=['U-ORG','O', 'U-TECHNOLOGY', 'O', 'U-DATE'])
The same goes for named entities. The letters added before the labels refer to the tags of the BILUO scheme – O is a token outside an entity, U an single entity unit, B the beginning of an entity, I a token inside an entity and L the last token of an entity.
命名實體也同樣，參照BILUO標簽體系在標記之前加一個字母（B/I/L/U/O）。
BILUO說明：
TAG????????????? DESCRIPTION
BEGIN?? The first token of a multi-tokenentity.
IN????????? An inner token of a multi-tokenentity.
LAST????? The final token of a multi-tokenentity.
UNIT????? A single-token entity.
OUT????? A non-entity token.
WHYBILUO, NOT IOB? 為啥是BILUO而不是IOB
There are several coding schemes for encoding entity annotations as token tags. These coding schemes are equally expressive, but not necessarily equally learnable. Ratinov and Roth showed that the minimal Begin, In, Out scheme was more difficult to learn than the BILUO scheme that we use, which explicitly marks boundary tokens.
有很多實體標注為標簽的編碼體系，這些編碼體系其實效果都一樣，但是可學習型不同。Ratinov和Roth表示最小的IOB(In/Out/Begin)體系在學習是比BILUO體系要復雜困難，因為BILUO明確標識了邊界tokens。
Training data: The training examples.
Text and label: The current example.
Doc: A Doc object created from theexample text.
GoldParse: A GoldParse object of the Docand label.
nlp: The nlp object with the model.
Optimizer: A function that holds statebetween updates.
Update: Update the model's weights.
Of course, it's not enough to only show a model a single example once. Especially if you only have few examples, you'll want to train for a number of iterations. At each iteration, the training data is shuffled to ensure the model doesn't make any generalisations based on the order of examples. Another technique to improve the learning results is to set a dropout rate, a rate at which to randomly"drop" individual features and representations. This makes it harder for the model to memorise the training data. For example, a 0.25 dropout means that each feature or internal representation has a 1/4 likelihood of being dropped.
當然，一次只有一個模型一個獨立樣本。特別是如果只有少量樣本的時候，想要訓練一堆迭代器。每個迭代器都清洗訓練數據，以確保模型不基于樣本順序作任何輸出。另一個提升學習結果的技術是設置流失率，一個隨機“扔下”個體特征和表述的頻率。這樣的話模型很難做到記住訓練數據。例如：0.25的流失率意味著每個特征或內部表述有1/4的可能性被扔掉。
begin_training() : Start the training and return an optimizer function to update the model's weights. Can take an optional function converting the training data to spaCy's training format.
update() : Update the model with the training example and gold data.
to_disk() : Save the updated model to adirectory.
EXAMPLE TRAINING LOOP
optimizer = nlp.begin_training(get_data)
for itn in range(100):
???random.shuffle(train_data)
???for raw_text, entity_offsets in train_data:
???????doc = nlp.make_doc(raw_text)
???????gold = GoldParse(doc, entities=entity_offsets)
???????nlp.update([doc], [gold], drop=0.5, sgd=optimizer)
nlp.to_disk('/model')
nlp.update函數有以下參數：
docs：?? Doc objects. The update method takes a sequence of them, so you can batch up your training examples. Alternatively, you can also pass in a sequence of raw texts.
Golds： GoldParse objects. The update method takes a sequence of them, so you can batch up your training examples. Alternatively, you can also pass in a dictionary containing the annotations.
Drop：?? Dropout rate. Makes it harder for the model to just memorise the data.
Sgd：??? An optimizer, i.e. a callable to update the model's weights. If not set, spaCy will create a new one and save it for further use.
Instead of writing your own training loop, you can also use the built-in train? command, which expects data in spaCy's JSON format. On each epoch, a model will be saved out to the directory. After training, you can use the package?command to generate an installable Python package from your model.
也可以直接使用內置的train命令進行訓練，需要數據是spaCy要求的JSON格式。每次，一個模型會被存儲于目錄中。訓練后，可以用package命令生成模型的python安裝包。
python -m spacy convert /tmp/train.conllu/tmp/data
python -m spacy train en /tmp/model/tmp/data/train.json -n 5
Simple training style
Instead of sequences of Doc and GoldParse objects, you can also use the "simple training style" and pass raw texts and dictionaries of annotations to nlp.update .The dictionaries can have the keys entities, heads, deps, tags and cats. This is generally recommended, as it removes one layer of abstraction, and avoids unnecessary imports. It also makes it easier to structure and load your training data.
可以使用simpletrain style替代Doc和GoldParse對象序列，然后把原始文本和標注字典傳遞給nlp.update。字典可以含有關鍵實體，heads，deps，tags和cats。通常推薦此做法，因為它去除了一個抽象層，而且避免了不必要的import。它也簡化了構建和加載訓練數據。
EXAMPLE ANNOTATIONS
{
???'entities': [(0, 4, 'ORG')],
???'heads': [1, 1, 1, 5, 5, 2, 7, 5],
???'deps': ['nsubj', 'ROOT', 'prt', 'quantmod', 'compound', 'pobj', 'det','npadvmod'],
???'tags': ['PROPN', 'VERB', 'ADP', 'SYM', 'NUM', 'NUM', 'DET', 'NOUN'],
???'cats': {'BUSINESS': 1.0}
}
SIMPLE TRAINING LOOP
TRAIN_DATA = [
????("Uber blew through $1 million a week", {'entities': [(0, 4,'ORG')]}),
????("Google rebrands its business apps", {'entities': [(0, 6,"ORG")]})]
nlp = spacy.blank('en')
optimizer = nlp.begin_training()
for i in range(20):
???random.shuffle(TRAIN_DATA)
???for text, annotations in TRAIN_DATA:
???????nlp.update([text], [annotations], sgd=optimizer)
nlp.to_disk('/model')
The above training loop leaves out a few details that can really improve accuracy – but the principle really is that simple. Once you've got your pipeline together and you want to tune the accuracy, you usually want to process your training examples in batches, and experiment with minibatch sizes and dropout rates, set via the drop keyword argument. See the Language?and Pipe? API docs for available options.
上述訓練環節遺漏了一些可以真正提升精度的細節，但是原理就是那么簡單。有時整理好pipline就想調整精度，想要批量處理訓練樣本就試試minibatch，大小和流失率呢可以通過drop參數來設置。參見Languige和pip API文檔。
Training the named entity recognizer 訓練命名實體識別（NER）
All spaCy models support online learning, so you can update a pre-trained model with new examples. You'll usually need to provide many examples to meaningfully improve the system — a few hundred is a good start, although more is better.
spaCy的所有模型都支持在線學習，所以你可以用新樣本更新預訓練模型。通常需要提供很多樣本來提升系統，幾百個起步，多多益善。
You should avoid iterating over the same few examples multiple times, or the model is likely to"forget" how to annotate other examples. If you iterate over the same few examples, you're effectively changing the loss function. The optimizer will find a way to minimize the loss on your examples, without regard for the consequences on the examples it's no longer paying attention to. One way to avoid this "catastrophic forgetting" problem is to "remind"the model of other examples by augmenting your annotations with sentences annotated with entities automatically recognised by the original model.Ultimately, this is an empirical process: you'll need to experiment on your data to find a solution that works best for you.
應該避免同一批樣本的多次迭代，不然模型可能會“忘記”如何標注其他樣本的。迭代同一批樣本，會有效改變損失率。優化器會為樣本找到最小化損失的方法，不顧后果的忽視樣本。避免這種“災難性遺忘”問題的方法是通過用普通模型自動識別實體的段落標注增加標注，來提醒其他樣本的模型，最后，經驗之談：需要數據試驗找到最好的解決方案。
TIP:CONVERTING ENTITY ANNOTATIONS 轉換實體標注
You can train the entity recognizer with entity offsets or annotations in the BILUO scheme. The spacy.gold module also exposes two helper functions to convert offsets to BILUO tags, and BILUO tags to entity offsets.
可以用實體集或BILUO標注體系來訓練實體識別。spaCy.gold模塊也有兩個函數可以進行實體集和BILUO標簽的相互轉換。
Updating the Named Entity Recognizer 更新NER
This example shows how to update spaCy's entity recognizer with your own examples, starting off with an existing, pre-trained model, or from scratch using a blank Language class. To do this, you'll need example texts and the character offsets and labels of each entity contained in the texts.
下例為如何用自己的樣本更新spaCy的實體識別，基于已有的預訓練模型或以空白語言類從零開始。為此，需要樣本文本和字符集以及文本中每個實體的標記。
spacy/examples/training/train_ner.py
Step by step guide
1、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using a blank model, don't forget to add the entity recognizer to the pipeline. If you're using an existing model, make sure to disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the entity recognizer.
2、Shuffle and loop over the examples. For each example, update the model by calling nlp.update ,which steps through the words of the input. At each word, it makes a prediction. It then consults the annotations to see whether it was right. If it was wrong, it adjusts its weights so that the correct action will score higher next time.
3、Save the trained model using nlp.to_disk .
4、Test the model to make sure the entities in the training data are recognised correctly.
1、加載模型，或者用spacy.blank和語言ID創建一個空的模型。如果使用空模型，別忘了在pipline中加入實體識別。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練實體識別。
2、shuffle和loopover樣本。對于每個樣本，調用nlp.update來升級模型（詞單步輸入）。對每個詞都會做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
3、nlp.to_disk，保存模型。
4、測試。
Training an additional entity type
This script shows how to add a new entity type ANIMAL to an existing pre-trained NER model, or an empty Language class. To keep the example short and simple, only a few sentences are provided as examples. In practice, you'll need many more — a few hundred would be a good start. You will also likely need to mix in examples of other entity types, which might be obtained by running the entity recognizer over unlabelled sentences, and adding their annotations to the training set.
下面的腳本是給已有的NER模型（或空語言類）添加一個ANIMAL的實體類型。樣例為了保持簡短，僅使用了很少的樣本。實操中需要很多，幾百個起步吧。此外，最好在樣本中混入其他實體類型，以及未標記的句子，并在訓練集中加入其標注。
spacy/examples/training/train_new_entity_type.py
IMPORTANTNOTE
If you're using an existing model, make sure to mix in examples of other entity types that spaCy correctly recognized before. Otherwise, your model might learn the new type, but "forget" what it previously knew. This is also referred to as the"catastrophic forgetting" problem.
如果使用已有模型，一定要在樣本中加入經spaCy正確識別的其他類型實體。否則，模型可能會狗熊掰棒子，屬于“災難性遺忘”問題。
Step by step guide
1、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using a blank model, don't forget to add the entity recognizer to the pipeline. If you're using an existing model, make sure to disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the entity recognizer.
2、Add the new entity label to the entity recognizer using the add_label? method. You can access the entity recognizer in the pipeline via nlp.get_pipe('ner').
3、Loop over the examples and call nlp.update , which steps through the words of the input. At each word, it makes a prediction. It then consults the annotations, to see whether it was right. If it was wrong, it adjusts its weights so that the correct action will score higher next time.
4、Save the trained model using nlp.to_disk .
5、Test the model to make sure the new entity is recognised correctly.
1、加載模型，或者用spacy.blank和語言ID 創建一個空的模型。如果使用空模型，別忘了在pipline中加入實體識別。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練實體識別。
2、用add_label函數給實體識別添加新的實體標記。通過nlp.get_pip(‘ner’)可在pipline中獲得實體識別。
3、loop over樣本，調用nlp.update，單步輸入詞。對每個詞都會做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
4、nlp.to_disk，保存模型。
5、測試模型。
Training the tagger and parser
Updating the Dependency Parser
This example shows how to train spaCy's dependency parser, starting off with an existing model ora blank model. You'll need a set of training examples and the respective heads and dependency label for each token of the example texts.
訓練spaCy依存句法分析的樣例，基于已有或空的模型都可以。需要一組訓練樣本，還有每個樣本文本中的每個token的head和依存關系標記。
spacy/examples/training/train_parser.py
Step by step guide
1、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using a blank model, don't forget to add the parser to the pipeline. If you're using an existing model, make sure to disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the parser.
2、Add the dependency labels to the parser using the add_label?method. If you're starting off with a pre-trained spaCy model, this is usually not necessary – but it doesn't hurt either, just to be safe.
3、Shuffle and loop over the examples. For each example, update the model by calling nlp.update ,which steps through the words of the input. At each word, it makes a prediction. It then consults the annotations to see whether it was right. If it was wrong, it adjusts its weights so that the correct action will score higher next time.
4、Save the trained model using nlp.to_disk .
5、Test the model to make sure the parser works as expected.
1、加載模型，或者用spacy.blank和語言ID創建一個空的模型。如果使用空模型，別忘了在pipline中加入parser。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練parser。
2、用add_label函數為parser添加依存關系標記。用已經預訓練的spaCy模型也沒啥關系，無傷害。
3、shuffle和loopover樣本，對于每個樣本，調用nlp.update來升級模型（詞單步輸入）。對每個詞都會做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
4、nlp.to_disk，保存模型。
5、測試模型。
Updating the Part-of-speech Tagger
In this example, we're training spaCy's part-of-speech tagger with a custom tag map. We start off with a blank Language class, update its defaults with our custom tags and then train the tagger. You'll need a set of training examples and the respective custom tags, as well as a dictionary mapping those tags to theUniversal Dependencies scheme.
下面的樣例用自定義的tagmap訓練spaCy的詞性標注器。用一個空語言類來開始，再用自定義標簽更新默認的，之后訓練tagger。需要一組訓練樣本及其自定義標簽，此外還要一個標簽與UD體系的映射字典。
spacy/examples/training/train_tagger.py
Step by step guide
1、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using a blank model, don't forget to add the tagger to the pipeline. If you're using an existing model, make sure to disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the tagger.
2、Add the tag map to the tagger using the add_label? method.The first argument is the new tag name, the second the mapping to spaCy's coarse-grained tags, e.g. {'pos': 'NOUN'}.
3、Shuffle and loop over the examples. For each example, update the model by calling nlp.update ,which steps through the words of the input. At each word, it makes a prediction. It then consults the annotations to see whether it was right. If it was wrong, it adjusts its weights so that the correct action will score higher next time.
4、Save the trained model using nlp.to_disk .
5、Test the model to make sure the parser works as expected.
1、加載模型，或者用spacy.blank和語言ID創建一個空的模型。如果使用空模型，別忘了在pipline中加入tagger。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練tagger。
2、用add_label函數為tagger添加tag
map。第一個參數是新tag名，第二個是對spaCy的coarse-grained tag的映射，即{‘pos’:’NOUN’}。
3、shuffle和loopover樣本，對于每個樣本，調用nlp.update來升級模型（詞單步輸入）。對每個詞都會做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
4、nlp.to_disk，保存模型。
5、測試模型。
Training a parser for custom semantics
spaCy's parser component can be used to be trained to predict any type of tree structure over your input text – including semantic relations that are not syntactic dependencies. This can be useful to for conversational applications, which need to predict trees over whole documents or chat logs, with connections between the sentence roots used to annotate discourse structure. For example, you can train spaCy's parser to label intents and their targets, like attributes, quality, time and locations. The result could look like this:
spaCy的parser組件可以用來訓練成預測輸入文本中的任何樹結構，包括非句法依賴的語義關系。這對于會話應用很有用，可以對整個文檔或聊天記錄進行樹預測（用連接線連接并標注語義）。例如，可以訓練spaCy的parser來標記目的及其目標，比如屬性、數量、時間和位置，效果如下：
doc = nlp(u"find a hotel with good wifi")
print([(t.text, t.dep_, t.head.text) fort in doc if t.dep_ != '-'])
# [('find', 'ROOT', 'find'), ('hotel','PLACE', 'find'),
#?('good', 'QUALITY', 'wifi'), ('wifi', 'ATTRIBUTE', 'hotel')]
The above tree attaches "wifi" to "hotel" and assigns the dependency labelATTRIBUTE. This may not be a correct syntactic dependency – but in this case, it expresses exactly what we need: the user is looking for a hotel with the attribute "wifi" of the quality "good". This query can then be processed by your application and used to trigger the respective action –e.g. search the database for hotels with high ratings for their wifi offerings.
上面的樹中，將wifi附給hotel，并標注依存關系標記ATTRIBUTE。這不一定是正確的依存關系，但是在這個情況下，確切表述了需要的東西：用戶想要找一個wifi質量good的hotel。這個檢索就會被應用處理，且出發各自的動作，即：在數據庫中搜索wifi評級高的hotels。
TIP:MERGE PHRASES AND ENTITIES
To achieve even better accuracy, try merging multi-word tokens and entities specific to your domain into one token before parsing your text. You can do this by running the entity recognizer or rule-based matcher to find relevant spans, and merging them using Span.merge . You could even add your own custom pipeline component to do this automatically – just make sure to add it before='parser'.
合并短語和實體
為達到更高精度，可以在parsing文本之前將多詞（字）tokens和相關實體合并為一個token。可以通過運行基于規則的匹配或實體識別來找到有關段落，然后用Span.merge合并他們。還可以添加自定義pipline組件來自動合并，注意要添加在 =’parser’前面。
The following example shows a full implementation of a training loop for a custom message parser fora common "chat intent": finding local businesses. Our message semantics will have the following types of relations: ROOT, PLACE, QUALITY,ATTRIBUTE, TIME and LOCATION.
下例是自定義一個完整的聊天內容parser的訓練：查找當地商業信息。信息語義包括如下關系類型：ROOT，PLACE，QUALITY，ATTRIBUTE，TIME以及LOCATION。
spacy/examples/training/train_intent_parser.py
Step by step guide
1、Create the training data consisting of words, their heads and their dependency labels in order. A token's head is the index of the token it is attached to. The heads don't need to be syntactically correct – they should express the semantic relations you want the parser to learn. For words that shouldn't receive a label, you can choose an arbitrary placeholder, for example -.
2、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using a blank model, don't forget to add the custom parser to the pipeline. If you're using an existing model, make sure to remove the old parser from the pipeline, and disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the parser.
3、Add the dependency labels to the parser using the add_label?method.
4、Shuffle and loop over the examples. For each example, update the model by calling nlp.update ,which steps through the words of the input. At each word, it makes a prediction. It then consults the annotations to see whether it was right. If it was wrong, it adjusts its weights so that the correct action will score higher next time.
5、Save the trained model using nlp.to_disk .
6、Test the model to make sure the parser works as expected.
1、創建由詞組成的訓練數據，及其head和依存關系label。一個token的head是token歸屬的索引。head不要求語法準確，只需能表述出語義關系即可。對于不應有label的字詞，隨便找個占位符就行了，比如“-”。
2、加載模型，或者用spacy.blank和語言ID創建一個空的模型。如果使用空模型，別忘了在pipline中加入自定義parser。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練parser。
3、用add_label函數為parser添加依存關系label。
4、shuffle和loopover樣本，對于每個樣本，調用nlp.update來升級模型（詞單步輸入）。對每個詞都會做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
5、nlp.to_disk，保存模型。
6、測試模型。
Training a text classification model
Adding a text classifier to a spaCy modelV2.0
This example shows how to train a multi-label convolutional neural network text classifier on IMDB movie reviews, using spaCy's new TextCategorizer component. The dataset will be loaded automatically via Thinc's built-in dataset loader. Predictions are available via Doc.cats .
下例是關于如何訓練一個對IMDB影評進行多標簽卷積神經網絡的文本分類器，使用spaCy的TextCategorizer組件。數據集通過Thinc的內置數據集加載器自動加載。預測通過Doc.cats實現。
spacy/examples/training/train_textcat.py
Step by step guide
1、Load the model you want to start with, or create an empty model using spacy.blank? with the ID of your language. If you're using an existing model, make sure to disable all other pipeline components during training using nlp.disable_pipes . This way, you'll only be training the text classifier.
2、Add the text classifier to the pipeline, and add the labels you want to train – for example,POSITIVE.
3、Load and pre-process the dataset, shuffle the data and split off a part of it to holdback for evaluation. This way, you'll be able to see results on each training iteration.
4、Loop over the training examples and partition them into batches using spaCy's minibatch? and compounding helpers.
5、Update the model by calling nlp.update , which steps through the examples and makes a prediction.It then consults the annotations to see whether it was right. If it was wrong, it adjusts its weights so that the correct prediction will score higher next time.
6、Optionally, you can also evaluate the text classifier on each iteration, by checking how it performs on the development data held back from the dataset. This lets you print the precision, recall and F-score.
7、Save the trained model using nlp.to_disk .
8、Test the model to make sure the text classifier works as expected.
1、加載模型，或者用spacy.blank和語言ID創建一個空的模型。如果使用已有模型，確認在訓練時關閉其他的pipline組件（nlp.disable_pips）。這個方法僅訓練text classifier。
2、在pipline中添加textclassifier，在添加想要訓練的labels，比如：POSITIVE。
3、加載預處理過的數據集，清洗數據并分離出一部分做評估。這樣就能看到每個訓練迭代器的結果。
4、用spaCy的minibatch等對樣本進行分批反復訓練
5、調用nlp.update升級模型，樣本單步調試，做一個預測。之后對比標注是否正確。如果錯了，就調整比重，所以下一次得分會更高些。
6、可選，還可以在每個迭代器上做評估，檢查效果如何，輸出precision，recall和F-score。
7、nlp.to_disk保存模型
8、測試模型。
Optimization tips and advice 優化建議
There are lots of conflicting "recipes" for training deep neural networks at the moment. The cutting-edge models take a very long time to train, so most researchers can't run enough experiments to figure out what's really going on.For what it's worth, here's a recipe that seems to work well on a lot of NLP problems:
訓練深度神經網絡時存在一些矛盾的方法。前端模型訓練用時很長，所以多數人不能運行足夠的試驗來找出到底咋回事。無論怎樣，這里有個方法似乎在一些NLP問題上還湊合，如下：
1、Initialise with batch size 1, and compound to a maximum determined by your data size and problem type.
2、Use Adam solver with fixed learning rate.
3、Use averaged parameters
4、Use L2 regularization.
5、Clip gradients byL2 norm to 1.
6、On small data sizes, start at a high dropout rate, with linear decay.
1、用batch size 1初始化，并賦予最大值取決于數據尺寸和問題類型。
2、用Adam solver固定學習率。
3、用平均值。
4、用L2正則化。
5、調整norm L2梯度為1。
6、小數據量，從高流失率開始線性衰減。
This recipe has been cobbled together experimentally. Here's why the various elements of the recipe made enough sense to try initially, and what you might try changing, depending on your problem.
此方法經過拼裝試驗，所以值得一試，再根據實際情況進行變更。
Compounding batch size
The trick of increasing the batch size is starting to become quite popular (see Smith et al., 2017). Their recipe is quite different from how spaCy's models are being trained, but there are some similarities. In training the various spaCy models, we haven't found much advantage from decaying the learning rate – but starting with a low batch size has definitely helped. You should try it out on your data, and see how you go. Here's our current strategy:
提升batchsize的手段頗受歡迎。他們的方法也和spaCy的模型訓練頗有不同，但是也有很多相似。訓練不同的spaCy模型時，沒有發現學習率衰退有多好，反而低batchsize倒是很有用。你應該用自己的數據試試看咋樣。下面是我們目前的策略：
BATCH HEURISTIC
def get_batches(train_data, model_type):
???max_batch_sizes = {'tagger': 32, 'parser': 16, 'ner': 16, 'textcat': 64}
???max_batch_size = max_batch_sizes[model_type]
???if len(train_data) < 1000:
???????max_batch_size /= 2
???if len(train_data) < 500:
???????max_batch_size /= 2
???batch_size = compounding(1, max_batch_size, 1.001)
???batches = minibatch(train_data, size=batch_size)
???return batches
This will set the batch size to start at 1, and increase each batch until it reaches a maximum size. The tagger, parser and entity recognizer all take whole sentences as input, so they're learning a lot of labels in a single example. You therefore need smaller batches for them. The batch size for the text categorizer should be somewhat larger, especially if your documents are long.
在這里batch設置為從1開始，然后遞增至最大size。tagger，parser和實體識別都以整個句子作為輸入，所以他們在單樣本中學到了很多labels，因此就需要給他們小一些的batch。text categorizer的batch size應該大一些，特別是面對長文檔。
Learning rate, regularization and gradient clipping
By default spaCy uses the Adam solver, with default settings (learning rate 0.001, beta1=0.9, beta2=0.999). Some researchers have said they found these settings terrible on their problems – but they've always performed very well in training spaCy's models, in combination with the rest of our recipe. You can change these settings directly, by modifying the corresponding attributes on the optimizer object. You can also set environment variables, to adjust the defaults.
spaCy默認使用Adamsolver（learning rate 0.001, beta1=0.9, beta2=0.999）。有人說自己發現這樣設置對于他們的問題來說很糟糕，但這些設置在spaCy模型的訓練中一直表現很好，同時結合了我們其余的方法。可以直接修改這些設置，直接針對優化對象修改對應屬性。也可以設置環境變量，以調整默認值。
There are two other key hyper-parameters of the solver: L2 regularization, and gradient clipping(max_grad_norm). Gradient clipping is a hack that's not discussed often, but everybody seems to be using. It's quite important in helping to ensure the network doesn't diverge, which is a fancy way of saying "fall over during training". The effect is sort of similar to setting the learning rate low.It can also compensate for a large batch size (this is a good example of how the choices of all these hyper-parameters intersect).
有其他兩個solver的key超參，L2正則和gradient clipping(max_grad_norm)。Gradient clipping是一個不常討論的hack，但是每個人似乎都在用，它對于確保網絡不發散很重要，有一個有趣的說法“訓練中被絆倒”。結果很多類似設置低學習率。這也可以補償一個大的batch size（這是一個如何交叉選擇所有這些超參數挺好的例子）。
Dropout rate
For small datasets, it's useful to set a high dropout rate at first, and decay it down towards amore reasonable value. This helps avoid the network immediately overfitting, while still encouraging it to learn some of the more interesting things in your data. spaCy comes with a decaying?utility function to facilitate this. You might try setting:
對于小數據集，在一開始設置高流失率很有用，之后之后降低到一個更合理的值。這樣有助于從數據中持續學習更多靠譜東西時避免網絡很快過擬合。spaCy自帶一個decaying工具函數來搞這個事情，可以試試：
from spacy.util import decaying
dropout = decaying(0.6, 0.2, 1e-4)
You can then draw values from the iterator with next(dropout), which you would pass to the drop keyword argument of nlp.update . It's pretty much always a good idea to use at least some dropout. All of the models currently use Bernoulli dropout, for no particularly principled reason – we just haven't experimented with another scheme like Gaussian dropout yet.
之后可以從迭代器中用next(dropout)寫參數值了，傳遞給nlp.update的drop參數。多一些dropout幾乎一直是個好主意。所有模型目前都使用Bernoulli dropout，沒什么特殊原因，我們只是還沒有用其他體系做過試驗，比如Gaussian dropout之類的。
Parameter averaging
The last part of our optimization recipe is parameter averaging, an old trick introduced by Freundand Schapire (1999), popularised in the NLP community by Collins (2002), and explained in more detail by Leon Bottou. Just about the only other people who seem to be using this for neural network training are the SyntaxNet team (one of whom is Michael Collins) – but it really seems to work great on every problem.
最后一部分優化方法是參數平均， 從Freund and Schapire (1999)），Collins(2002)，到Leon Bottou解釋了更多細節。大概僅有Michael Collins所在團隊SyntaxNet在用，不過看起來這個方法在所有問題上都運行良好。
The trick is to store the moving average of the weights during training. We don't optimize this average – we just track it. Then when we want to actually use the model, we use the averages, not the most recent value. In spaCy (and Thinc) this is done by using a context manager, use_params , to temporarily replace the weights:
其手段是存儲訓練過程中權重的平均移動值。我們不去優化這個平均值，只是跟蹤它。然后當我們真的要用模型時，使用平均值而不是最近值。在spaCy和thinc中，用一個環境manager，use_params，去臨時替換權重來實現。
with nlp.use_params(optimizer.averages):
???nlp.to_disk('/model')
The context manager is handy because you naturally want to evaluate and save the model at various points during training (e.g. after each epoch). After evaluating and saving, the context manager will exit and the weights will be restored, so you resume training from the most recent value, rather than the average. By evaluating the model after each epoch, you can remove one hyper-parameter from consideration(the number of epochs). Having one less magic number to guess is extremely nice– so having the averaging under a context manager is very convenient.
Contextmanager很方便，因為在訓練過程中肯定想要在不通地方評估并存儲模型（例如：每個epoch之后）。評估和存儲之后，context manager會退出并重置權重值，所以恢復訓練是從最近值開始而不是平均值。通過每個epoch之后評估模型，可以考慮移除一個超參（epoch總數）。少猜一個數超爽，所以在contextmanager有個平均值很實用。
Transfer learning
Finally, if you're training from a small data set, it's very useful to start off with some knowledge already in the model. Word vectors are an easy and reliable way to do that, but depending on the application, you may also be able to start with useful knowledge from one of spaCy's pre-trained models, such as the parser, entity recogniser and tagger. If you're adapting a pre-trained model and you want it to retain accuracy on the tasks it was originally trained for, you should consider the "catastrophic forgetting" problem. See this blogpost to read more about the problem and our suggested solution, pseudo-rehearsal.
最后，如果用一個小數據集進行訓練，使用一些已在模型中存在的知識非常有用。詞向量是一個簡單而直接的方法，但是取決于實際應用，或許還可以用一個spaCy的預訓練模型，比如實體識別，parser，tagger。如果適配一個預訓練模型，而且想在實際任務中再訓練原有訓練內容，恐怕會發生狗熊掰棒子問題。關于問題的更多內容及建議解決方案，參考下文https://explosion.ai/blog/pseudo-rehearsal-catastrophic-forgetting
Saving and loading models
After training your model, you'll usually want to save its state, and load it back later. You can do this with the Language.to_disk()?method:
訓練模型后，通常會保存，以便之后加載。Language。to_disk()函數nlp.to_disk( '/home/me/data/en_example_model' )
The directory will be created if it doesn't exist, and the whole pipeline will be written out. To make the model more convenient to deploy, we recommend wrapping it as a Python package.
如果指定目標目錄不存在則會創建一個，并且整個pipline將被寫入。要使模型更實用，推薦打包為Python包。
Generating a model package
IMPORTANTNOTE
The model packages are not suitable for the public pypi.python.org directory, which is not designed for binary data and files over 50 MB. However, if your company is running an internal installation of PyPi, publishing your models on there can be a convenient way to share them with your team.
spaCy comes with a handy CLI command that will create all required files, and walk you through generating the meta data. You can also create the meta.json manually and place it in the model data directory, or supply a path to it using the --meta flag.For more info on this, see the package?docs.
spaCy自帶一個方便的CLI命令用來創建所有需要的文件，且直接生成元數據。可以手動創建meta.json并放進模型的數據目錄中，或者用 –meta flag提供一個路徑。更多內容參見package文檔：https://spacy.io/api/cli#package
META.JSON
{
???"name": "example_model",
???"lang": "en",
???"version": "1.0.0",
???"spacy_version": ">=2.0.0,<3.0.0",
???"description": "Example model for spaCy",
???"author": "You",
???"email": "you@example.com",
???"license": "CC BY-SA 3.0",
???"pipeline": ["tagger", "parser","ner"]
}
python -m spacy package/home/me/data/en_example_model /home/me/my_models
This command will create a model package directory that should look like this:
上面的命令將創建一個模型包，其目錄結構如下所示：
DIRECTORY STRUCTURE
└── /
???├── MANIFEST.in???????????????????????????????? #to include meta.json
???├── meta.json????????????????????????????????????????? #model meta data
???├── setup.py???????????????????????????????????? #setup file for pip installation
???└── en_example_model????????????????? #model directory
???????├── __init__.py???????????????????????????? #init for pip installation
???????└── en_example_model-1.0.0? # model data
You can also find templates for all files on GitHub . If you're creating the package manually, keep in mind that the directories need to be named according to the naming conventions of lang_name and lang_name-version.
所有模版文件都可以在GitHub(https://github.com/explosion/spacy-models/blob/master/template )上找到。自己創建包的時候注意目錄的命名規則，lang_name和lang_name-version。
Customising the model setup
The meta.json includes the model details, like name, requirements and license, and lets you customise how the model should be initialised and loaded. You can define the language data to be loaded and the processing pipeline to execute.
Meta.json包括模型的細節，比如name，requirements和license，也允許自定義如何初始化和加載模型。還可以定義加載語言數據和運行處理pipline。
The load() method that comes with our model package templates will take care of putting all this together and returning a Language object with the loaded pipeline and data. If your model requires custom pipeline components or a custom language class, you can also ship the code with your model. For examples of this, check out the implementations of spaCy's load_model_from_init_py? and load_model_from_path? utility functions.
Load()函數自帶了模型包模版，能夠將所有內容整合并返回一個語言對象及其加載的pipline和數據。如果模型需要自定義pipline組件或自定義語言類，也可以將編碼與模型一并封裝。例如：參看spaCy以下實現工具：load_model_from_init_py以及load_model_from_path。
Building the model package
To build the package, run the following command from within the directory. For more information on building Python packages, see the docs on Python's Setuptools.
Build包命令如下（更多內容參見Python的Setuptools https://setuptools.readthedocs.io/en/latest/）：
python setup.py sdist
This will create a.tar.gz archive in a directory /dist. The model can be installed by pointingpip to the path of the archive:
如上將在/dist目錄中創建一個.tar.gz的壓縮包。該模型可以用pip install加包路徑進行安裝：
pip install/path/to/en_example_model-1.0.0.tar.gz
You can then load the model via its name, en_example_model, or import it directly as a module and then call its load() method.
然后就可以用模型的名稱加載模型了，或者直接引入模塊后用load()函數加載。
Loading a custom model package
To load a model from a data directory, you can use spacy.load()?with the local path. This will look for a meta.json in the directory and use the lang and pipeline settings to initialise a Language class with a processing pipeline and load in the model data.
從數據目錄加載模型：spacy.load()加本地路徑。之后會到指定目錄中查找meta.json并用lang和pipline設置用pipline來初始化一個語言類，并加載模型數據。
nlp = spacy.load('/path/to/model')
If you want to load only the binary data, you'll have to create a Language class and call from_disk? instead.
如果想加載bin，需要創建一個語言類，并使用from_disk。
nlp =spacy.blank('en').from_disk('/path/to/data')
IMPORTANT NOTE: LOADING DATA IN V2.X
In spaCy 1.x, the distinction between spacy.load() and the Language class constructor was quite unclear. You could call spacy.load() when no model was present, and it would silently return an empty object. Likewise, you could pass a path to English, even if the mode required a different language. spaCy v2.0 solves this with a clear distinction between setting up the instance and loading the data.
spaCy1.x版本中，spacy.load()和語言類構造器的區別很模糊。當前沒有模型時，可以調用spacy.load()也可以直接返回一個空對象。而且還可以傳遞一個English的路徑，即使模型需要其他語言。2.0版本解決了這個問題。
正確方法：??? nlp= spacy.blank('en').from_disk('/path/to/data')
錯誤方法：??? nlp= spacy.load('en', path='/path/to/data')
Example: How we're training and packaging models for spaCy
Publishing a new version of spaCy often means re-training all available models – currently, that's 13 models for 8 languages. To make this run smoothly, we're using an automated build process and a spacy train?template that looks like this:
每發布一個新版本通常意味著重新訓練所有模型—當前為8個語言的13個模型。為了平滑過渡，我們使用一個自動build處理和一個spacytrain模版，長相如下：
python -m spacy train {lang}{models_dir}/{name} {train_data} {dev_data} -m meta/{name}.json -V {version} -g{gpu_id} -n {n_epoch} -ns {n_sents}
META.JSON TEMPLATE
{
???"lang": "en",
???"name": "core_web_sm",
???"license":"CC BY-SA 3.0",
???"author":"Explosion AI",
???"url":"https://explosion.ai",
???"email":"contact@explosion.ai",
???"sources": ["OntoNotes 5", "CommonCrawl"],
???"description":"English multi-task CNN trained onOntoNotes, with GloVe vectors trained on common crawl. Assigns word vectors, context-specific token vectors, POS tags, dependency parse and named entities."
}
In a directory meta, we keep meta.json templates for the individual models, containing all relevant information that doesn't change across versions, like the name, description, author info and training data sources. When we train the model, we pass in the file to the meta template as the --meta argument, and specify the current model version as the --version argument.
在一個目錄內，為每一個獨立的模型保留一個meta.json模版，包括所有跨版本無需變更的有關信息，比如name，description，authorinfo還有training data sources。當訓練模型時，傳遞meta模版的–meta參數，還有指定當前模型版本的—version參數。
On each epoch, the model is saved out with a meta.json using our template and added properties, like the pipeline, accuracy scores and the spacy_version used to train the model. After training completion, the best model is selected automatically and packaged using the package? command.Since a full meta file is already present on the trained model, no further setup is required to build a valid model package.
每個epoch，模型都和一個meta.json一同保存并加入屬性，比如pipline，accuracyscores以及spacy_version以用來訓練模型。訓練完成后，最好的模型被自動挑出來并用package命令打包。因為一個完整的meta文件已經在模型中存在了，所以不需要更多設置。
python -m spacy package -f {best_model}dist/
cd dist/{model_name}
python setup.py sdist
This process allows us to quickly trigger the model training and build process for all available models and languages, and generate the correct meta data automatically.
上述樣例為快速出發模型的訓練并build所有模型和語言的處理，并自動生成正確的meta數據。
