This repository houses several Practical Data Science assignements relating to LLM Prompting (use cases, modern prompting techniques, fail conditions).
The repository is organized as follows:
-
ChatGPT prompting fail cases where we explore two cases of ChatGPT code generation failing.
-
LLM prompting for annotation where we leverage LLMs to help us in a difficult annotation task, which we then handle by using pandas.
-
Annotation analysis where we analyze the clustering between different annotators.
-
Secure-GPT a group presentation for a new LLM use-case to make the internet safer - includes fake data generation for demonstration purposes.
-
Webscraping where we create a webscraper for a Greek gaming forum.
-
Greek Embedding Plot, in which we create an interquartile range for all the posts in a Greeklish2Greek dataset.
-
Greek Embedding Creation in which we create and test our own Greek Embeddings.