GH-12: updated tutorials with new class names

README.md

@@ -59,7 +59,7 @@ a pre-trained model and use it to predict tags for the sentence:
 
 ```python
 from flair.data import Sentence
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 # make a sentence
 sentence = Sentence('I love Berlin .')

flair/embeddings.py

@@ -458,10 +458,6 @@ def __init__(self, word_embeddings: List[TokenEmbeddings], reproject_words: bool
 
         self.word_reprojection_map = torch.nn.Linear(self.__embedding_length, self.__embedding_length)
 
-    @property
-    def embedding_type(self):
-        return 'sentence-level'
-
     @property
     def embedding_length(self) -> int:
         return self.__embedding_length
@@ -544,10 +540,6 @@ def __init__(self, word_embeddings: List[TokenEmbeddings], hidden_states=128, nu
                                  bidirectional=self.bidirectional)
         self.dropout = torch.nn.Dropout(0.5)
 
-    @property
-    def embedding_type(self):
-        return 'sentence-level'
-
     @property
     def embedding_length(self) -> int:
         return self.__embedding_length
@@ -640,10 +632,6 @@ def __init__(self, charlm_embeddings: List[CharLMEmbeddings], detach: bool = Tru
     def embedding_length(self) -> int:
         return self._embedding_length
 
-    @property
-    def embedding_type(self):
-        return 'sentence-level'
-
     def embed(self, sentences: List[Sentence]):
 
         for embedding in self.embeddings:

resources/docs/EXPERIMENTS.md

@@ -35,8 +35,9 @@ Now, select 'ner' as the tag you wish to predict and init the embeddings you wis
 The full code to get a state-of-the-art model for English NER is as follows: 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -52,7 +53,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     # GloVe embeddings
     WordEmbeddings('glove')
@@ -67,7 +68,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -79,9 +80,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-ner', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)
@@ -108,8 +109,9 @@ FastText embeddings (they work better on this dataset). The full code then is as
 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -125,7 +127,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('ft-crawl')
     ,
@@ -137,8 +139,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
-
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -150,9 +151,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-ner', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)
@@ -176,8 +177,9 @@ Once you have the data, reproduce our experiments exactly like for CoNLL-03, jus
 FastText word embeddings and German contextual string embeddings. The full code then is as follows: 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -193,7 +195,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('ft-german')
     ,
@@ -205,7 +207,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -217,9 +219,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-ner', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)
@@ -242,8 +244,9 @@ get the dataset and place train, test and dev data in `/resources/tasks/germeval
 Once you have the data, reproduce our experiments exactly like for the German CoNLL-03: 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -259,7 +262,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('ft-german')
     ,
@@ -271,7 +274,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -283,9 +286,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-ner', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)
@@ -311,8 +314,9 @@ so the algorithm knows that POS tags and not NER are to be predicted from this d
 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -328,7 +332,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('extvec')
     ,
@@ -340,7 +344,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -352,9 +356,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-pos', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)
@@ -378,8 +382,9 @@ Run the code with extvec embeddings and our proposed contextual string embedding
 so the algorithm knows that chunking tags and not NER are to be predicted from this data. 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import  NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -395,7 +400,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('extvec')
     ,
@@ -407,7 +412,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -419,9 +424,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=False)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=False)
 
 trainer.train('resources/taggers/example-pos', mini_batch_size=32, max_epochs=150, save_model=True,
               train_with_dev=True, anneal_mode=True)

resources/docs/TUTORIAL_BASICS.md

@@ -111,7 +111,7 @@ Simply point the `NLPTaskDataFetcher` to the file containing the parsed sentence
 list of `Sentence`
 
 ```python
-import NLPTaskDataFetcher
+from flair.data_fetcher import NLPTaskDataFetcher
 
 # use your own data path
 data_folder = 'path/to/conll/formatted/data'
@@ -142,7 +142,7 @@ To read a file containing text classification data simply point the `NLPTaskData
 It will read the sentences into a list of `Sentence`
 
 ```python
-import NLPTaskDataFetcher
+from flair.data_fetcher import NLPTaskDataFetcher
 
 # use your own data path
 data_folder = 'path/to/text-classification/formatted/data'

resources/docs/TUTORIAL_TAGGING.md

@@ -9,7 +9,7 @@ This model was trained over the English CoNLL-03 task and can recognize 4 differ
 types.
 
 ```python
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger = SequenceTagger.load('ner')
 ```

resources/docs/TUTORIAL_TEXT_EMBEDDINGS.md

@@ -33,15 +33,15 @@ So if you want to create a text embedding using GloVe embeddings together with C
 use the following code:
 
 ```python
-from flair.embeddings import WordEmbeddings, CharLMEmbeddings, TextMeanEmbedder
+from flair.embeddings import WordEmbeddings, CharLMEmbeddings, DocumentMeanEmbeddings
 
 # initialize the word embeddings
 glove_embedding = WordEmbeddings('glove')
 charlm_embedding_forward = CharLMEmbeddings('news-forward')
 charlm_embedding_backward = CharLMEmbeddings('news-backward')
 
 # initialize the text embeddings
-text_embeddings = TextMeanEmbedder([glove_embedding, charlm_embedding_backward, charlm_embedding_forward])
+text_embeddings = DocumentMeanEmbeddings([glove_embedding, charlm_embedding_backward, charlm_embedding_forward])
 ```
 
 Now, create an example sentence and call the embedding's `embed()` method. 
@@ -82,11 +82,11 @@ If you want, you can also specify some other parameters:
 So if you want to create a text embedding using only GloVe embeddings, use the following code:
 
 ```python
-from flair.embeddings import WordEmbeddings, TextLSTMEmbedder
+from flair.embeddings import WordEmbeddings, DocumentLSTMEmbeddings
 
 glove_embedding = WordEmbeddings('glove')
 
-text_embeddings = TextLSTMEmbedder([glove_embedding])
+text_embeddings = DocumentLSTMEmbeddings([glove_embedding])
 ```
 
 Now, create an example sentence and call the embedding's `embed()` method. 

resources/docs/TUTORIAL_TRAINING_A_MODEL.md

@@ -85,8 +85,9 @@ Here is example code for a small NER model trained over CoNLL-03 data, using sim
 In this example, we downsample the data to 10% of the original data. 
 
 ```python
-from flair.data import NLPTaskDataFetcher, TaggedCorpus, NLPTask
-from flair.embeddings import TextEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
+from flair.data import TaggedCorpus
+from flair.data_fetcher import NLPTaskDataFetcher, NLPTask
+from flair.embeddings import TokenEmbeddings, WordEmbeddings, StackedEmbeddings, CharLMEmbeddings, CharacterEmbeddings
 from typing import List
 import torch
 
@@ -102,7 +103,7 @@ tag_dictionary = corpus.make_tag_dictionary(tag_type=tag_type)
 print(tag_dictionary.idx2item)
 
 # initialize embeddings
-embedding_types: List[TextEmbeddings] = [
+embedding_types: List[TokenEmbeddings] = [
 
     WordEmbeddings('glove')
 
@@ -119,7 +120,7 @@ embedding_types: List[TextEmbeddings] = [
 embeddings: StackedEmbeddings = StackedEmbeddings(embeddings=embedding_types)
 
 # initialize sequence tagger
-from flair.tagging_model import SequenceTagger
+from flair.models import SequenceTagger
 
 tagger: SequenceTagger = SequenceTagger(hidden_size=256,
                                         embeddings=embeddings,
@@ -130,9 +131,9 @@ if torch.cuda.is_available():
     tagger = tagger.cuda()
 
 # initialize trainer
-from flair.trainer import TagTrain
+from flair.trainers import SequenceTaggerTrainer
 
-trainer: TagTrain = TagTrain(tagger, corpus, test_mode=True)
+trainer: SequenceTaggerTrainer = SequenceTaggerTrainer(tagger, corpus, test_mode=True)
 
 trainer.train('resources/taggers/example-ner', mini_batch_size=32, max_epochs=150, save_model=False,
               train_with_dev=False, anneal_mode=False)