Are you interested in using ChatGPT professionally? At the Deep Learning Reading Seminar meetup a few days ago, our co-founder Gyula Kovács excitedly shared his thoughts on the technology and its importance, and highlighted some examples and strategies for using it. Read what he talked about and watch the recording for more details!
From the presentation:
NLP and LLMs – Natural language processing and large language models
Gyula started the presentation with the professional background and the milestones, to sketch the picture of where we are now in the field of natural language processing and what large language models have emerged recently. He showed what a small slice of the new language models have entered the public domain.
What is the new technology good for?
To give just a few examples, ChatGPT and similar models that are taking the public spotlight can be used for
- information retrieval/search: the familiar Google searches can be an effective trigger for the new technology, as the answer is much more relevant to us than the Google results, from which we have to retrieve the necessary details. In contrast, language models immediately return these details.
- semantic classification: Gyula showed through a simple example that ChatGPT can classify things, so with a little creativity we can use language models to perform certain classification tasks.
- summarisation: Don’t have time to write briefly? No time to read the whole thing? – If you have a lot of text and little time, you can summarise the article or text you want to know using language models to get to grips with the content quickly. But it also works if you’re writing a not-so-condensed, long-winded piece of text and you want to send a compact, well-written email to your client.
- text generation: Or do you just have the gist in your head and need to add the surrounding text? You should write an article and you’d rather sit back? Ask the right questions from ChatGPT!
But what are the right questions? What input should we give the language model to produce the right output?
The art of this is called prompt engineering, which requires background knowledge and experience to achieve the desired result.
Gyula gave us a little insight into the world of prompt engineering and a quick overview of its theory:
- Definition of Prompt
- The purpose of a prompt
- The anatomy of a prompt
- Formalizing a prompt
- Prompting strategies – non-exhaustive list of main approaches
- Beyond classical prompting strategies
Watch the recording of the presentation for details:
How to take advantage of new technology?
Would you use ChatGPT to boost your business? Are you already clear on exactly how you want to use it? Or do you need expert help to explore the potential of the technology? The Neuron Solutions team can help you identify the areas where the new technology can have the greatest impact on your business and guarantee an efficient and rapid implementation.
You can find out more about how we can help by following the link below: