spacy-transformers 1.0.0a15

spaCy pipelines for pre-trained BERT and other transformers

spacy-transformers

This package (previously spacy-pytorch-transformers) provides spaCy model pipelines that wrap Hugging Face's transformers package, so you can use them in spaCy. The result is convenient access to state-of-the-art transformer architectures, such as BERT, GPT-2, XLNet, etc.

Features

• Use pretrained transformer models like BERT, RoBERTa and XLNet to power your spaCy pipeline.
• Easy multi-task learning: backprop to one transformer model from several pipeline components.
• Train using spaCy v3's powerful and extensible config system.
• Automatic alignment of transformer output to spaCy's tokenization.
• Easily customize what transformer data is saved in the Doc object.
• Easily customize how long documents are processed.
• Out-of-the-box serialization and model packaging.

🚀 Installation

Installing the package from pip will automatically install all dependencies, including PyTorch and spaCy. Make sure you install this package before you install the models. Also note that this package requires Python 3.6+ and spaCy v3.

pip install spacy-transformers

For GPU installation, find your CUDA version using nvcc --version and add the version in brackets, e.g. spacy-transformers[cuda92] for CUDA9.2 or spacy-transformers[cuda100] for CUDA10.0.

If you are having trouble installing PyTorch, follow the instructions on the official website for your specific operation system and requirements.

📖 Usage

⚠️ Important note: This package has been extensively refactored to take advantage of spaCy v3. Previous versions that were built for spaCy v2 worked considerably differently. Please see previous tagged versions of this readme for documentation on prior versions.

spaCy v3 lets you use almost any statistical model to power your pipeline. You can use models implemented in a variety of frameworks, including Tensorflow, PyTorch and MXNet. To keep things sane, spaCy expects models from these frameworks to be wrapped with a common interface, using our machine learning library Thinc. A transformer model is just a statistical model, so the spacy-transformers package actually has very little work to do: we just have to provide a few functions that do the required plumbing. We also provide a pipeline component, Transformer, that lets you do multi-task learning and lets you save the transformer outputs for later use.

Training usage

The recommended workflow for training is to use spaCy v3's new config system, usually via the spacy train-from-config command-line command. See here for an end-to-end example. The config system lets you describe a tree of objects by referring to creation functions, including functions you register yourself. Here's a config snippet for the Transformer component, along with matching Python code.

[nlp.pipeline.transformer]
factory = "transformer"
set_extra_annotations = null
max_batch_size = 32

[nlp.pipeline.transformer.model]
@architectures = "spacy-transformers.TransformerModel.v1"
name = "bert-base-cased"
tokenizer_config = {"use_fast": true}

[nlp.pipeline.transformer.model.get_spans]
@span_getters = "spacy-transformers.get_doc_spans.v1"

trf = Transformer(
    nlp.vocab,
    TransformerModel(
        "bert-base-cased",
        get_spans=get_doc_spans,
        tokenizer_config={"use_fast": True},
    ),
    set_extra_annotations=null_annotation_setter,
    max_batch_size=32,
)
nlp.add_pipe("transformer", trf, first=True)

The nlp.pipeline.transformer block adds the transformer component to the pipeline, and the nlp.pipeline.transformer.model block describes the creation of a Thinc Model object that will be passed into the component. The block names a function registered in the @architectures registry. This function will be looked up and called using the provided arguments. You're not limited to just that function --- you can write your own or use someone else's. The only limitation is that it must return an object of type Model[List[Doc], FullTransformerBatch]: that is, a Thinc model that takes a list of Doc objects, and returns a FullTransformerBatch object with the transformer data.

The same idea applies to task models that power the downstream components. Most of spaCy's built-in model creation functions support a tok2vec argument, which should be a Thinc layer of type Model[List[Doc], List[Floats2d]]. This is where we'll plug in our transformer model, using the Tok2VecTransformer layer, which sneakily delegates to the Transformer pipeline component.

[nlp.pipeline.ner]
factory = "ner"

[nlp.pipeline.ner.model]
@architectures = "spacy.TransitionBasedParser.v1"
nr_feature_tokens = 3
hidden_width = 128
maxout_pieces = 3
use_upper = false

[nlp.pipeline.ner.model.tok2vec]
@architectures = "spacy-transformers.TransformerListener.v1"
grad_factor = 1.0

[nlp.pipeline.ner.model.tok2vec.pooling]
@layers = "reduce_mean.v1"

The TransformerListener layer expects a pooling layer, which needs to be of type Model[Ragged, Floats2d]. This layer determines how the vector for each spaCy token will be computed from the zero or more source rows the token is aligned against. Here we use the reduce_mean layer, which averages the wordpiece rows. We could instead use reduce_last, reduce_max, or a custom function you write yourself.

You can have multiple components all listening to the same transformer model, and all passing gradients back to it. By default, all of the gradients will be equally weighted. You can control this with the grad_factor setting, which lets you reweight the gradients from the different listeners. For instance, setting grad_factor = 0 would disable gradients from one of the listeners, while grad_factor = 2.0 would multiply them by 2. This is similar to having a custom learning rate for each component. Instead of a constant, you can also provide a schedule, allowing you to freeze the shared parameters at the start of training.

Runtime usage

Transformer models can be used as drop-in replacements for other types of neural networks, so your spaCy pipeline can include them in a way that's completely invisible to the user. Users will download, load and use the model in the standard way, like any other spaCy pipeline.

Instead of using the transformers as subnetworks directly, you can also use them via the Transformer pipeline component. This sets the doc._.trf_data extension attribute, which lets you access the transformers outputs at runtime via the doc._.trf_data extension attribute. You can also customize how the Transformer object sets annotations onto the Doc, by customizing the Transformer.set_extra_annotations function. This callback will be called with the raw input and output data for the whole batch, along with the batch of Doc objects, allowing you to implement whatever you need.

import spacy

nlp = spacy.load("en_core_trf_lg")
for doc in nlp.pipe(["some text", "some other text"]):
    doc._.trf_data.tensors
    tokvecs = doc._.trf_data.tensors[-1]

The nlp object in this example is just like any other spaCy pipeline, so see the spaCy docs for more details about what you can do.