Improved-Language-Model-Instructions-Tuning-using-Alpaca-Dataset

In this project, I explored different prompt types for Large Language Models.

Alpaca Dataset

I utilized the "Alpaca" dataset, which comprises 52,000 instructions and demonstrations generated by OpenAI's text-davinci-003 engine. This instruction data is ideal for conducting instruction-tuning for language models, enhancing their ability to follow instructions effectively.

Dataset Link: Alpaca Dataset

Large Language Models (LLMs)

I employed six different types of Large Language Models for this task. Here are the details along with their respective links:

1. GPT2

   Model Link: GPT2 Documentation

2. GPT-Medium

   Model Link: GPT-Medium

3. Mistral-7B-v0.1

   Model Link: Mistral-7B-v0.1

4. TinyLlama-1.1B-Chat-v1.0

   Model Link: TinyLlama-1.1B-Chat-v1.0

5. Mistral-7B-Instruct-v0.2

   Model Link: Mistral-7B-Instruct-v0.2

6. Starling-LM-7B-alpha

   Model Link: Starling-LM-7B-alpha

Feel free to explore these models and the Alpaca dataset for a deeper understanding of the project's advancements in language model instruction tuning.