Introduction to Large Language Models (LLMs)

Front

What are LLMs?

Back

LLMs are ‘token generators’ that process units of text, such as words, characters, or subwords.

Front

What is the main purpose of LLMs?

Back

The main purpose of LLMs is to understand and generate human-like language.

Front

What is the training process for LLMs?

Back

The training process includes pre-training on a diverse dataset, fine-tuning on a specific dataset, and reinforcement learning from human feedback (RLHF).

Front

What does 'pre-training' involve?

Back

Pre-training involves training the model on a huge, diverse dataset to learn general language patterns, grammar, and world knowledge.

Front

What is 'fine-tuning' in the context of LLMs?

Back

Fine-tuning is the process of training a pre-trained model on a smaller, more specific dataset to improve its performance for a particular task.

Front

What is RLHF?

Back

Reinforcement Learning from Human Feedback (RLHF) involves humans ranking different model outputs to align the model's responses with human preferences.

Front

What are tokens in LLMs?

Back

Tokens are sequences of characters that represent a word, a part of a word, or punctuation.

Front

What are the three main architectures of LLMs?

Back

The three main architectures are Decoder-only, Encoder-only, and Encoder-Decoder.

Front

What is the function of Decoder-only models?

Back

Decoder-only models predict the next token based only on the preceding tokens and are excellent for text generation.

Front

What are Encoder-only models used for?

Back

Encoder-only models are great at understanding and analyzing text but not generating new content.