GPT 3 is currently one of the most state-of-the-art Natural Language Processing models out there in the world.

Fortunate enough to get hold of a GPT3 API key, I’ve been playing with it and building demos for some promising use cases, including:

  • Job Requirements Extractor

logo

It can extract job requirements quite accurately and also output the years of experience needed to apply.

The latter is insanely important because even tho I’m a junior data scientist Linkedin recommends me more-senior jobs all the time! If Linkedin were to set up a year-of-experience filter, think about how much frustration, disappointment and wasted time would be saved!

  • Job Requirements Extractor logo

It can generate a list of pros and cons from customer reviews on e-commerce sites(Ipad Pro in our case), which can save buyers from having to trudge through all the reviews to find constructive stuff (i.e. facts useful for the purchase decision making process).

Skills covered:

  • Prompt writing for GPT3 (literally only skill needed but it took me quite a while to wrap my mind around it..)

Limitations with GPT3:

  • It often repeats itself
  • Unstable output(the output may change drastically even with a slight variation of the prompt, such as paraphrasing or even just adding a new space)
  • Input size is limited (difficult to pack in more than a few long text examples in the prompt. However, it’s possible to fine-tune/re-train the model with customized examples)

The code can be found here