LLM

1. What is an LLM?

LLM (Large Language Model) is a language model algorithm developed using a neural network trained with millions of parameters, trained on large amounts of unlabeled text using self-supervised learning or semi-supervised learning. The current greatest exponent of this type of language models is GPT-4 from the company OpenAI.

2. What’s the function of an LLM?

The main function of an LLM is to help users in their queries, providing updated information up to the training date of that model (for example GPT-3 gives answers up to data from January 2023, with the free version). Despite the advantage that it can have in filtering the information by doing a bit of web searching, sometimes you can have “hallucinations” this is because the LLM misunderstands the user’s query, confusing the given parameters or asking for more information. updated with respect to their last training date, giving the user an answer that is incorrect. When these hallucinations do not occur, the answers almost always tend to be the correct answer to the user’s query.

3. Examples of LLM

• GPT-4 (OpenAI).
• LaMDA (Google)
• GPT-NeoX (EleutherAI)
• Chinchilla (Deep Mind)
• AlexaTM (Amazon)
• LlaMA (Meta)
• BLOOM (Hugging Face)