Large Language Models Course assignments, More information about the course is available here

Fall 2023

1. LoRa and Adapters:

   • Exploring Full and Parameter Effecient fine-tuning of LLMs using Adapters and LoRa as well as Soft Prompting.

2. In-Context Learning:

   • Analysing the effects of In-Context Learning based on number and order of examples. Studying the effects of dataset and altering it to see the results on generalization.

3. Captioning and RAG:

   • Using LLM's as captioners by combining the embeddings of image and text. Studying the applications of Retrieval-augmented generation (RAG) model in retriving data and generating proper response.

4. Evaluation, Decoding and Machine Translation:

   • Comparing different decoding methods used in generating text with LLMs. Getting familiar with different evaluation methods for LLMs, and exploring their capabilities for machine translation.