Overview:

This hands-on workshop is designed to provide you with a comprehensive understanding of Natural Language Processing (NLP) concepts and practical skills to build a project that predicts the next word based on user input. The workshop will cover using KerasNLP for NLP tasks, training a BERT (Bidirectional Encoder Representations from Transformers) model locally, and deploying the trained model as a REST API using FastAPI.

Prerequisites:

Participants are expected to have a basic understanding of Python programming and familiarity with machine learning concepts. Additionally, a basic knowledge of neural networks and NLP will be beneficial. 

Participants should bring their laptops with the following software installed:

• Python (preferably Anaconda distribution)
• Jupyter Notebooks
• Git (optional)
• Code editor (e.g. VS Code)

Workshop Outline:

1. Introduction to NLP:

Brief overview of NLP and its applications

Introduction to key NLP concepts: tokenization, word embeddings, and transformers

2. KerasNLP Basics:

Overview of KerasNLP library

Hands-on exercises on text preprocessing and tokenization

3. Understanding BERT:

Introduction to BERT and its architecture

Overview of pre-trained BERT models

Fine-tuning BERT for custom tasks

4. Hands-on BERT Training:

Data preparation for training

Building and training a BERT model locally using KerasNLP

Evaluation of the trained model

5. Introduction to FastAPI:

Overview of FastAPI and its advantages

Setting up a FastAPI project

6. Deploying BERT Model with FastAPI:

Integrating the trained BERT model with FastAPI

Creating REST endpoints for model inference

Testing the deployed model locally

Outcome:

By the end of this workshop you will have gained practical experience in NLP concepts, used KerasNLP for text processing, trained a BERT model locally, and deployed it as a REST API using FastAPI. 

You will be equipped with the knowledge and skills to undertake similar NLP projects and integrate machine learning models into production environments.

Trainer:

Karthik Muthuswamy is a Google Developer Expert in Machine Learning.

Time & Location:

The in-person workshop will take place at Die Kabine in Königsworther Str. 37, 30167 Hannover on Saturday, January 27, at 3PM (15 Uhr) and take about three hours. If you need more time it is no problem to stay a bit longer.

We will have a limited amount of drinks and snacks. Several kiosks are close by just in case.

Social:

Feel more than welcome to connect with us on Mastodon, Telegram, LinkedIn or X.