Token Classification

Token classification is a task in which a label is assigned to some tokens in a text. Some popular token classification subtasks are Named Entity Recognition (NER) and Part-of-Speech (PoS) tagging.

For more details about the token-classification task, check out its dedicated page! You will find examples and related materials.

Recommended models

dslim/bert-base-NER: A robust performance model to identify people, locations, organizations and names of miscellaneous entities.
FacebookAI/xlm-roberta-large-finetuned-conll03-english: A strong model to identify people, locations, organizations and names in multiple languages.
blaze999/Medical-NER: A token classification model specialized on medical entity recognition.
flair/ner-english: Flair models are typically the state of the art in named entity recognition tasks.

Explore all available models and find the one that suits you best here.

Using the API

Language

Client

Provider

Settings

from huggingface_hub import InferenceClient

client = InferenceClient(
    provider="hf-inference",
    api_key="hf_xxxxxxxxxxxxxxxxxxxxxxxx",
)

result = client.token_classification(
    inputs="My name is Sarah Jessica Parker but you can call me Jessica",
    model="dslim/bert-base-NER",
)

API specification

Request

Payload
inputs*	string	The input text data
parameters	object
ignore_labels	string[]	A list of labels to ignore
stride	integer	The number of overlapping tokens between chunks when splitting the input text.
aggregation_strategy	string	One of the following:
(#1)	’none’	Do not aggregate tokens
(#2)	’simple’	Group consecutive tokens with the same label in a single entity.
(#3)	’first’	Similar to “simple”, also preserves word integrity (use the label predicted for the first token in a word).
(#4)	’average’	Similar to “simple”, also preserves word integrity (uses the label with the highest score, averaged across the word’s tokens).
(#5)	’max’	Similar to “simple”, also preserves word integrity (uses the label with the highest score across the word’s tokens).

Some options can be configured by passing headers to the Inference API. Here are the available headers:

Headers
authorization	string	Authentication header in the form `'Bearer: hf_**'` when `hf_**` is a personal user access token with Inference API permission. You can generate one from your settings page.
x-use-cache	boolean, default to `true`	There is a cache layer on the inference API to speed up requests we have already seen. Most models can use those results as they are deterministic (meaning the outputs will be the same anyway). However, if you use a nondeterministic model, you can set this parameter to prevent the caching mechanism from being used, resulting in a real new query. Read more about caching here.
x-wait-for-model	boolean, default to `false`	If the model is not ready, wait for it instead of receiving 503. It limits the number of requests required to get your inference done. It is advised to only set this flag to true after receiving a 503 error, as it will limit hanging in your application to known places. Read more about model availability here.

For more information about Inference API headers, check out the parameters guide.

Response

Body
(array)	object[]	Output is an array of objects.
entity_group	string	The predicted label for a group of one or more tokens
entity	string	The predicted label for a single token
score	number	The associated score / probability
word	string	The corresponding text
start	integer	The character position in the input where this group begins.
end	integer	The character position in the input where this group ends.

< > Update on GitHub

Inference Providers