Create Your Own GPT-3 Chatbot Based on Your Social Media Data
UPDATE: The workshop is now full. For future workshops, please sign up for my newsletter.
Have you ever wondered what it would be like to chat with a chat bot developed from your social media data? In this workshop, part of the study Exploring Embedded Values in Natural Language Processing through Fine-Tuning Foundational Models on Social Media Data, participants will learn how to fine-tune the large language model GPT-3 using their own social media data to create a functional chat bot. Participants will then converse with their bot and compare this to conversations with the foundation model GPT-3.
This workshop is limited to 60 participants: 30 in-person participants (Copenhagen, Denmark – location at the University of Copenhagen exact building TBD) and 30 online participants (you will be asked to specify your preference for online or offline at sign up). Participants must be 18 years of age or older. Participation is free.
There are two workshops offered: one for experienced NLP programmers (those who have familiarity with NLP at the programming level) and one for those who may be new to NLP. If you are confused about which workshop to sign up for, please feel free to email us. We will compare our findings between the two groups for our study. We will provide in-depth instructions and assistance for anyone who is new to programming or worried they do not have the necessary skills for the workshop including online office hours. The workshop will be recorded as part of the study.
Date / Time
Participation Requirements – PLEASE READ
We will close sign-ups as soon as we reach capacity, so please register early provided you have accepted the commitment to prepare and attend. If you are unsure as to whether or not you can attend or meet the preparation requirements, we ask you to refrain from signing up so as to allow the opportunity for others to participate. We will maintain a waitlist following the closure of the registration period in case others are unable to participate and spots open up.
Sign Up Deadline: April 7, 2023 Midnight AOE
Preparation Deadline (see below for details): April 10th, 2023 Midnight AOE
Preparation Deadline Requirement: Participants are asked to complete preparatory steps by April 10th, 2023. You will receive written and video instructions upon acceptance into the workshop. If you are unable to complete these steps by the deadline, please message us to see if it is still possible to attend the workshop. These steps cannot be completed on the day of the workshop as they require some time to complete. The steps include downloading your Facebook Messenger conversations (unless you are using Twitter posts – these will be collected during the workshop). Each participant will also need to sign up for a free account with OpenAI in advance (please note: free accounts come with a limited amount of tokens – this should be sufficient for the workshop but might require using less data than your complete profile if you have a lot of data).
This workshop is a part of the study Exploring Embedded Values in Natural Language Processing through Fine-Tuning Foundational Models on Social Media Data. For more details on the study please see the informed consent letter.
- Do I need to know how to program? No, but please sign up for the appropriate workshop. We will show you through videos and text instruction how to use the command line and install the various tools you will need.
- To participate do I have to share my social media data with you? Your social media data is used for the fine-tuning process, but we (the researchers running the workshop) will not see your data. You run the fine-tuning process by yourself and we help you to complete it. For this, seeing individual messages is not necessary. If you have any worries about this, you can also decide what data you want to include or not include in the process by deleting anything you do not want to use from your data prior to the workshop.
- How does OpenAI use my data?
- Is this workshop open to students only? No, but because we are running the workshop under the University of Copenhagen and University of Twente we are interested in involving the student body and therefor interested in your affiliation (if any). If you are unaffiliated with a university that is perfectly fine.
- What will I need to do to prepare for this workshop? You will need to 1) sign up for an OpenAI account (free) 2) download your social media data of your choosing. We will provide you instructions on how to do this upon acceptance.
- Who is running the workshop? The workshop is run by Mirabelle Jones (PhD Fellow University of Copenhagen) and Nastasia Griffioen (PostDoc University of Twente).
- What is the study all about? We are running a study in tandem with the workshop about social media, chatbots, and natural language processing focusing on the difference between fine-tuned models and foundation models. The study information sheet is located here. All details in our research findings including personal data will be anonymized in accordance with GDPR regulations as we are both from European institutions.
By signing up, you acknowledge that you have read the aforementioned instructions carefully and are committed to both attending the workshop for the full duration as well as completing the preparation requirements prior to the deadline.