Using Transformer Pipelines

The HuggingFace Transformers library provides pipelines to help developers benefit from transformer code immediately without any custom training. A pipeline is a tokenizer and a pre-trained model combined.

HuggingFace provides a variety of models for a variety of NLP tasks. Here are some tasks that HuggingFace pipelines offer:

• Sentiment analysis

• Question answering

• NER

• Text summarization

• Translation

You can see the full list of tasks at the Huggingface documentation. In this section, we'll explore the pipelines for sentiment analysis and question answering (the use of pipelines with other tasks is similar).

Let's go through some examples. We'll start with sentiment analysis: