What tasks are LLMs capable of doing?

Question

Other than they being used to get answers when a question is prompted, and to write (source code, stories, etc), are there any tasks that they can do (specially when fine tuned)? Are there some obscure or unusual tasks that they can be used?

Answer 1

The only task an LLM is capable of doing is predicting the continuation of the input text.

The greater the similarity between your desired task and text prediction, the better the LLM will do at it. For example, creative writing is quite similar, so an LLM is good at things like producing text in a given style. Likewise, answering questions is similar (although answering questions correctly is not).

At the other end of the spectrum, LLMs are horrible at being original. I'm sure you've seen the complaints about ChatGPT always producing the same half-dozen "dad jokes". This is an inevitable consequence of being text predictors: "creativity" is pretty much the exact opposite of probability maximization.

Answer 2

Large Language Models (LLMs) are capable of performing a wide range of tasks that are as follows:-

1. Text Generation
2. Text Completion
3. Summarization
4. Translation
5. Question Answering
6. Text Classification
7. Conversational Agents
8. Content Generation for Specific Purposes
9. Text-Based Games and Simulations
10. Code Generation and Debugging
11. Educational Tools
12. Data Extraction
13. Paraphrasing
14. Personalization
15. Sentiment and Opinion Analysis
16. Visual assistance

Answer 3

Other than they being used to get answers when a question is prompted, and to write (source code, stories, etc), are there any tasks that they can do (specially when fine tuned)?

The only task an LLM is capable of doing is predicting the continuation of the input text.

Note that all tasks can be phrased as QA or continuation of the input text.