Sequence Model Workshop Series - Part 1
Intro to Sequence Models

Date: April 17, 2024 (Wednesday)

Time: 5:30pm – 7:30 pm

Venue: Digital Learning Studio, Tam Wing Fan Innovation Wing One, HKU

Intro to Sequence Models

Do you want to know how ChatGPT works?

Interested in how Google Translate operates?

Many modern technologies, such as ChatGPT, Google Translate, AutoCorrect, and email filtering, are all built on top of sequence models! ᕦ(ò_óˇ)ᕤ

In this workshop series, we will introduce you to the theory and applications of sequence models. By learning how to use powerful machine learning models, you will eventually be able to create your own language models. (=´∀`)人(´∀`=)

In this first workshop, two basic sequence models: RNNs and LSTMs will be covered and you will have the skills to do language modelling, prediction problems, automated translations, and much more! (p_-)

Date: April 17, 2024 (Wednesday)

Time: 5:30pm- 17:00pm

Venue: Digital Learning Studio, Tam Wing Fan Innovation Wing One, HKU

Class size: A limited quota of 18 will be offered on first-come-first-served basis. Prior knowledge on Python programming is a MUST.

Eligibility: Priority to members of Innovation Wing! All HKU students are welcome to join.

Instructor: AI student research assistants of Innovation Wing

Workshop Outline
Session Duration Details
Introduction and Recap of Machine Learning Basics
10 minutes

1. Workshop Overview and Expectations
2. A short revision of machine learning

    • Machine learning and its branches
    • General aspects of machine learning model training process
    • Gradient descent and cost function
Introduction to Sequence Models
3 minutes
1. Introduction to common sequence models
2. Differences between the models
3. Applications of sequence models
General Overview of Recurrent Neural Networks (RNNs)
12 minutes
1. Introduction to RNNs
2. General applications of RNNs
3. Limitations of RNNs
Coding: Character-based RNNs
35 minutes
1. Implementation of a basic RNN using Python
2. Preparation of dataset for training
Break
5 minutes
Coding Practice
(using Character-based RNNs)
15 minutes
1. Workshop attendees prepare and use a custom dataset on the model
Introduction to Long Short-Term Memory (LSTMs)
10 minutes
1. Introduction to LSTMs
(an improved version of RNNs with longer memory)
Coding: LSTMs
25 minutes
1. Implementation of an LSTM for sentiment analysis on Python
2. Usage of predefined functions in libraries like Datasets and PyTorch
Wrap-up and Q&A
5 minutes
1. Workshop wrap-up
2. Q&A Session
Event Photos