With the introduction of their robust and user-friendly APIs, OpenAI has had a huge influence on the artificial intelligence industry. Without requiring a thorough grasp of machine learning, these APIs enable developers and organizations to access the cutting-edge capabilities of OpenAI's models.
In this blog post, we will explore a couple of OpenAI APIs and the capabilities they offer: Text completion API and JavaScript helper chatbot. We will also share an example of how these APIs have been used in real-world projects and highlight their potential for future innovation.
The for-profit OpenAI LP and its parent organization, the nonprofit OpenAI Inc., make up the artificial intelligence research facility known as OpenAI.
Through its API (Application Programming Interface), OpenAI is one of the primary means through which it makes its research and resources accessible to the general public. Modern AI models from the firm may be accessed and used by developers using the OpenAI API, a cloud-based service, to carry out activities like computer vision, natural language processing, and other tasks in their own apps.
The OpenAI API can be accessed using a simple RESTful interface, which allows developers to send requests to the API and receive responses in JSON format. The API also provides a variety of different models that can be used for different tasks, such as text generation, language translation, and sentiment analysis. To use the OpenAI API, you just need to create an API key and use it in your requests to the API.
Text completion
There are several tasks that may be performed with the completion endpoint. It offers any of our models a straightforward yet effective user interface. When you provide a text prompt, the model will provide a text completion that makes an effort to match the context or pattern you provided.
Collection of models with various features and pricing ranges power the OpenAI API. Additionally, you may fine-tune the fundamental models to fit your unique use case.
- GPT-3: A set of models that can understand and generate natural language.
- Codex (Limited beta): A set of models that can understand and generate code, including translating natural language to code.
- Content filter: A fine-tuned model that can detect whether text may be sensitive or unsafe.
GPT-3 is the most popular model.
Natural language may be produced and understood by the GPT-3 models. OpenAI provides four primary types with varying degrees of power appropriate for various purposes. The most competent model is Davinci, while the quickest is Ada.
- 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 supports inserting completions within text.
- text-curie-001 Very capable, but faster and lower cost than Davinci.
- text-babbage-001 Capable of straightforward tasks, very fast, and lower cost.
- text-ada-001 Capable of very simple tasks, usually the fastest model in the GPT-3 series, and lowest cost.
Here’s a node.js example
const { Configuration, OpenAIApi } = require("openai");
const c Configuration({
apiKey: process.env.OPENAI_API_KEY,
});
const openai = new OpenAIApi(configuration);
const resp openai.createCompletion({
model: "text-davinci-003",
prompt: "Say this is a test",
max_tokens: 7,
temperature: 0,
});
model string Required
JavaScript helper chatbot
This chatbot can respond to inquiries about utilizing JavaScript in the form of messages. It starts the discussion with a few examples.
model: "code-davinci-002"
Prompt
You: How do I combine arrays?
JavaScript chatbot: You can use the concat() method.
You: How do you make an alert appear after 10 seconds?
JavaScript chatbot
Sample response
const { Configuration, OpenAIApi } = require("openai");
const c Configuration({
apiKey: process.env.OPENAI_API_KEY,
});
const openai = new OpenAIApi(configuration);
const resp openai.createCompletion({
model: "code-davinci-002",
prompt: "You: How do I combine arrays?
JavaScript chatbot: You can use the concat() method.
You: How do you make an alert appear after 10 seconds?
JavaScript chatbot",
temperature: 0,
max_tokens: 60,
top_p: 1.0,
frequency_penalty: 0.5,
presence_penalty: 0.0,
stop: ["You:"],
});
Detailed descriptions
prompt string or array Optional Defaults to <|endoftext|>
The prompt(s) to generate completions for, encoded as a string, array of strings, array of tokens, or array of token arrays.
The document separator that the model observes during training is <|endoftext|>, thus if a prompt is not given, the model will create as if from a fresh document's beginning.
suffix string Optional Defaults to null
The suffix that comes after the completion of inserted text.
max_tokens integer Optional Defaults to 16
The maximum number of tokens to generate in the completion.
Your prompt's token count + max tokens cannot be greater than the model's context length. The standard context length for models is 2048 tokens (except for the newest models, which support 4096).
temperature number Optional Defaults to 1
What sampling temperature to use?
Higher values mean the model will take more risks. You could try 0.9 for more creative applications, or 0 (argmax sampling) for ones with a well-defined answer.
It is generally recommended to alter this or top_p but not both.
top_p number Optional Defaults to 1
Nucleus sampling is an alternative to temperature sampling where the model takes into account the outcomes of the tokens with top p probability mass. 0.1 signifies that only the tokens that make up the top 10% of the probability mass are taken into account.
Generally recommended to alter this or temperature but not both.
n integer Optional Defaults to 1
How many completions are to be generated for each prompt?
This parameter produces a lot of completions, which means it can soon exhaust your token allotment. Use with caution and make sure your stop and max tokens settings are appropriate.
stream boolean Optional Defaults to false
Whether to stream back partial progress.
In the case that it is set, tokens will be provided as data-only server-sent events as they become available, with a data: [DONE] message ending the stream.
logprobs integer Optional Defaults to null
Include the log probabilities on the logprobs most likely tokens, as well the chosen tokens.
The API will provide a list of the five most probable tokens, for instance, if logprobs is 5. There may be up to logprobs+1 items in the response because the API will always provide the logprob of the sampled token. Logprobs can have a value as high as 5.
echo boolean Optional Defaults to false
echo back the prompt in addition to the completion
stop string or array Optional Defaults to null
Up to 4 sequences where the API will stop generating further tokens. The returned text will not contain the stop sequence.
presence_penalty number Optional Defaults to 0
Number between -2.0 and 2.0. Positive values penalize new tokens based on whether they appear in the text so far, increasing the model's likelihood of talking about new topics.
frequency_penalty number Optional Defaults to 0
Number between -2.0 and 2.0. Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model's likelihood of repeating the same line verbatim.
best_of integer Optional Defaults to 1
Generates best_of completions server-side and returns the "best" (the one with the highest log probability per token). Results cannot be streamed.
When used with n, best_of controls the number of candidate completions and n specifies how many to return – best_of must be greater than n.
Note: This parameter produces a lot of completions, which means it can soon exhaust your token allotment. Use with caution and make sure your stop and max tokens settings are appropriate.
logit_bias map Optional Defaults to null
Modify the likelihood of specified tokens appearing in the completion. Accepts a json object that maps tokens (specified by their token ID in the GPT tokenizer) to an associated bias value from -100 to 100.
Use this tokenizer tool to transform text into token IDs (it supports GPT-2 and GPT-3). The bias is mathematically added to the logits produced by the model before sampling. Values between -1 and 1 should raise or reduce the chance of selection; values like -100 or 100 should result in a ban or exclusive selection of the relevant token. The precise effect will vary depending on the model.
For instance, you may give "50256": -100" to stop the generation of the "endoftext" token.
user string Optional
A unique identifier representing your end-user, which can help OpenAI to monitor and detect abuse.
ContentWeb
ContentWeb, a platform we've developed using OpenAi's Text completion API, allows users to easily create, manage, and distribute content on different platforms. It is intended for companies, marketers, and people who wish to expand their internet presence and connect with their target market.
The ability to build and administer several accounts from a single platform is one of ContentWeb's primary capabilities. Users can simply update and maintain their social media profiles because of the centralization of their content management.
The built-in SEO tools of ContentWeb are another feature that makes it simple for users to optimize their content for search engines. This covers resources for optimizing users' search engine rankings, such as sitemaps, meta tags, and keyword research tools.
Users may construct profiles with a professional appearance using ContentWeb's extensive selection of themes and designs without any prior design knowledge. Users can also alter the templates and designs to meet their own requirements.
There are a number of features that make it simple for users to share and distribute their material. This includes tools for email marketing, social media integration, and analytics that may show users how their material is being interpreted by their intended audience.