Overview

This course teaches advanced techniques in Natural Language Processing (NLP) using spaCy and spaCy-LLM. You’ll learn how to integrate LLMs into your NLP workflows, creating custom components and models. Designed for both beginners and experienced developers, this course provides hands-on examples to help you master spaCy’s core functionalities. It equips you to apply transformer models and fine-tune them for specialized NLP tasks. What sets this course apart is its practical approach. You’ll learn to build end-to-end NLP workflows and gain skills in deploying production-ready solutions, making it ideal for real-world applications. This course is perfect for NLP engineers, machine learning developers, and software engineers. A basic understanding of Python and NLP concepts is recommended for the best experience.

Syllabus

Getting Started with spaCy

In this section, we install spaCy and its language models, configure the environment, employ displaCy to visualize entities and dependencies, and assess spaCy's suitability for production-level Python NLP workflows.

Core Operations with spaCy

In this section, we build a spaCy NLP pipeline, customize the Tokenizer, segment sentences, apply lemmatization, and explore Doc, Span, and Token containers to strengthen everyday language processing skills.

Extracting Linguistic Features

In this section, we walk through spaCy workflows for Part-of-Speech tagging, dependency parsing, and Named Entity Recognition, then merge or split tokens to supply clean linguistic features to applications.

Mastering Rule-Based Matching

In this section, we design token and phrase patterns using spaCy's Matcher, PhraseMatcher, and SpanRuler, employ POS, morphology, and regex operators, then integrate rules with NER to extract domain-specific entities.

Extracting Semantic Representations with spaCy Pipelines

In this section, we build SpanRuler rules for LOCATION extraction, craft DependencyMatcher intent patterns, and assemble a custom spaCy pipeline leveraging Language.pipe() to efficiently process large ATIS datasets.

Utilizing spaCy with Transformers

In this section, we integrate transformer-based transfer learning into spaCy, examine BERT and RoBERTa architectures, and prepare config files to train an accurate TextCategorizer for production NLP pipelines.

Enhancing NLP Tasks Using LLMs with Spacy-LLM

In this section, we integrate spaCy and LLM components, build a summarization pipe, design Jinja-based prompts for context-aware extraction, and embed these custom tasks to enhance NLP performance.

Training an NER Component with Your Own Data

In this section, we assess spaCy's default NER on domain texts, annotate entities using Prodigy and nertk, then configure, train and integrate multiple custom NER components for accurate, specialized pipelines.

Creating End-to-End spaCy Workflows with Weasel

In this section, we clone a Weasel spaCy template, customize it for varied NLP tasks, then integrate DVC Studio to version data, track experiments, and enable reproducible production pipelines.

Training an Entity Linker Model with spaCy

In this section, we configure spaCy's pipeline to train an EntityLinker, craft high-quality annotated corpora, and evaluate linking accuracy with a custom reader for knowledge-base integration.

Integrating spaCy with Third-Party Libraries

In this section, we connect spaCy models to Streamlit and FastAPI, building an interactive NER web app and a type-hinted REST API that serves entity extraction for production use.