In what might be considered a baby step towards the much-anticipated GPT-4, OpenAI has released an updated version of the text generation model behind its popular GPT-3 platform.
According to an email reviewed by Synthetic Engineers, OpenAI says “This model builds on top of our previous InstructGPT models, and improves on a number of behaviors that we’ve heard are important to you as developers.”
Specifically, they say the following about Davinci-003’s new capabilities:
- It produces higher quality writing. This will help your applications deliver clearer, more engaging, and more compelling content.
- It can handle more complex instructions, meaning you can get even more creative with how you make use of its capabilities now.
- It’s better at longer form content generation, allowing you to take on tasks that would have previously been too difficult to achieve.
In early testing, Synthetic Engineers confirmed this. Prompts that would have previously yielded a short, formulaic response now yield several paragraphs of text.
The prompt:
I wrote a blog post about the release of the new text-davinci-004 model from OpenAI.Yields the following with the new model:
It’s been a long time coming, but OpenAI has finally released its new text-davinci-004 model. The model is a large-scale, unsupervised model for natural language processing (NLP), and it has been designed to enable researchers to create more sophisticated and powerful language models. The text-davinci-004 model has been trained on a massive collection of text from a variety of sources, including books, articles, blogs, and more. This training data allows the model to understand the nuances of language and provide better results when it comes to tasks such as text summarization, question answering, and text generation. The model is also able to generate text, which could be used for a variety of applications. For example, the model could be used to generate realistic-sounding dialogue for video games or to generate creative writing for stories. It could also be used for machine translation, helping to bridge the gap between different languages. OpenAI believes that the text-davinci-004 model could be a major step forward in the field of NLP, paving the way for more advanced models in the future. With its massive training data set, powerful algorithms, and sophisticated capabilities, it could revolutionize
Still, the new text often fails to pass even rudimentary GPT-3 detection programs. Although the text may be useful for certain applications, search engines can likely still tell that it’s AI-generated.
OpenAI users can access the new model through their existing Playground interface by selecting text-davinci-003 under the Model dropdown. The model is also available via the OpenAI API, which means that services built on GPT-3 will be able to access it and improve the quality of their textual output almost immediately.