OpenAI launched a new model in the GPT-3 family of AI-enabled large language models, called Davinci-003, on Monday. Davinci-003 ensures high quality, more extended output, and better instruction-following.
Soon after the launch of the Davinci-003, people discovered that Davinci-003 could generate rhyming songs, limericks, and poetry, which GPT-3 could not do previously.
However, OpenAI has not yet commented anything on the rhyming songs and poetry of Davinci-003.
According to OpenAI, Davinci-003 can perform all the tasks that the other models can and even need fewer instructions. It works well with tasks requiring in-depth knowledge of the subject, like summarizing texts for specific audiences and creative content development. As Davinci-003 produces the best results for tasks that need an in-depth understanding of content, it will require more computing resources. Therefore, Davinci-003 costs more per API call and is also a bit slower than the other models.
Davinci-003 excels at understanding the intent of the text. For example, it is good at providing solutions to various logical problems and explaining the character’s motivation. Davinci-300 has been capable of solving most of the challenging AI problems containing cause and effect.