In this article, we will discuss the key concepts of how you can use OpenAI within your Zammo business and gain an understanding of best practices, and common use cases.
OpenAI is an AI research company that has a mission to ensure artificial general intelligence (AGI) benefits all of humanity.
OpenAI has trained a number of models that are available for use as a service and can be used via API.
<aside> š” At the time of writing this article, OpenAI has come out with GPT3.5. We can expect additional improvements and versioning to be released as well in the future
</aside>
GPT3 represents the version of the series of models that OpenAI has trained.
There are 4 key models that have been trained by OpenAI and are available for use via API:
| LATEST GPT3 MODEL | DESCRIPTION | MAX REQUEST | TRAINING DATA |
|---|---|---|---|
| text-davinci-003 | Most capable GPT-3 model. Can do any task the other models can do, often with higher quality, longer output and better instruction-following. Also supportsĀ https://platform.openai.com/docs/guides/completion/inserting-textĀ completions within text. | 4,000 tokens | Up to Jun 2021 |
| text-curie-001 | Very capable, but faster and lower cost than Davinci. | 2,048 tokens | Up to Oct 2019 |
| text-babbage-001 | Capable of straightforward tasks, very fast, and lower cost. | 2,048 tokens | Up to Oct 2019 |
| text-ada-001 | Capable of very simple tasks, usually the fastest model in the GPT-3 series, and lowest cost. | 2,048 tokens | Up to Oct 2019 |
source: https://platform.openai.com/docs/models/gpt-3
<aside>
š” There are four model groups: DaVinci > Curie > Babbage > Ada
There are three tasks that the models can do: natural language (text), codex
(code), and embeddings (text-search,text-similarity)
The models have the following structure: <task>-<group>-<version>
</aside>