OpenAI (ChatGPT, Whisper, DALL-E)

With OpenAI (ChatGPT, Whisper, DALL-E) modules in Boost.space Integrator, you can create chat or prompt completions, edits, and moderations, generate images, and call APIs.

To get started with OpenAI, you need an OpenAI account.

[Note] Note

Boost.space Integrator modules support GPT-3.5, and GPT-4 models provided the user has access. If you do not have access, please join the waitlist to get access when capacity is available.

GPT-4 is currently in a limited beta version and is only accessible to those who have been granted access.

Connect OpenAI to Boost.space Integrator

To connect OpenAI to Boost.space Integrator, you must obtain an API Key and Organization ID from your account.

  1. Log in to your OpenAI (ChatGPT, Whisper, DALL-E) account.

  2. Click on your profile icon in the top right corner > View API Keys.

    OpenAI1.png
  3. Click Create new secret key, give the key a name (optional), and click Create secret key.

  4. Copy the secret key, store it in a safe place as you will not be able to view it again, and click Done.

  5. Go to your account settings page and copy your Organization ID to your clipboard.

    openai-2.png
  6. Log in to your Boost.space Integrator account, add an OpenAI (ChatGPT, Whisper, DALL-E) module to a scenario, and click Create a connection.

  7. Optional: In the Connection name field, enter a name for the connection.

  8. In the API Key field, enter the secret key copied in step 4.

  9. In the Organization ID field, enter the organization ID copied in step 5 and click Save.

You have successfully established the connection. You can now edit your scenario and add more OpenAI (ChatGPT, Whisper, DALL-E) modules. If your connection needs reauthorization at any point, follow the connection renewal steps here.

Actions

Analyze Image(s) – Vision

Accepts an array of images as an input and provides an analysis result for each of the images following the instructions specified in the prompt.

[Note] Note

If you do not have access to the GPT-4 models, please join the waitlist to get access when capacity is available.

GPT-4 is currently in a limited beta version and is only accessible to those who have been granted access.

Connection

Establish a connection to your OpenAI account.

Prompt

Enter instructions for how to analyze the image(s).

Images

Add or map the images you want to analyze.

You can add images by entering an Image URL or Image file data.

Image URL: Enter the URL address to a public resource of the image. For example, https://getmyimage.com/myimage.png.

Image file: Map or select the binary data of the image. You can retrieve the binary data of an image using the HTTP: Get a file module, or another app such as Dropbox.

Model

Select or map the model you want to use.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Max Token

Enter the maximum number of tokens to use for the completion. The default value is 300.

Temperature

Specify the sampling temperature to use.

Higher temperatures generate more diverse and creative responses. For example, 0.8. Lower temperatures generate more focused and well-defined responses. For example, 0.2. The default value is 1.

The value must be lower than or equal to 2.

Top P

Specify the Top P value to use nucleus sampling. This will consider the results of the tokens with probability mass. The default value is 1.

The value must be lower than or equal to 1.

Number

Enter the number of responses to generate. If more than 1 response is generated, the results can be found in the module’s output within Choices. The default value is 1.

Create a Completion (GPT-3, GPT-3.5, GPT-4)

Creates a completion for the provided prompt or chat.

See the OpenAI model endpoint compatibility section for the list of supported models.

Connection

Establish a connection to your OpenAI account.

Select Method

Select a method to create a completion.

Model

Select or map the model you want to use.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Messages

Add the messages for which you want to create the completion by selecting the Role and entering the Message Content.

For more information about chat completion, refer to the OpenAI documentation.

Max Token

The maximum number of tokens to generate in the completion. The default value is 16.

Temperature

Specify the sampling temperature to use.

A higher value means the model will take more risks. Try 0.9 for more creative applications and 0 (argmax sampling) for ones with a well-defined answer. Defaults to 1.

Top P

An alternative to sampling with temperature is called nucleus sampling, where the model considers the results of the tokens with top_p probability mass. So 0.1 means only the tokens comprising the top 10% probability mass are considered. Defaults to 1.

Number

The number of completions to generate for each prompt. Defaults to 1.

Frequency Penalty

Positive values penalize new tokens based on their existing frequency in the text so far, decreasing the model’s likelihood to repeat the same line verbatim. This value must be a number between -2 and 2.

Presence Penalty

Positive values penalize new tokens based on whether they appear in the text so far, increasing the model’s likelihood to talk about new topics. This value must be a number between -2 and 2.

Token Probability

Add the token probability by selecting the Token and Probability. The Probability must be a number between -100 and 100.

Response Format

Choose the format for the response.

Seed

This feature is in Beta. If specified, our system will make a best effort to sample deterministically, such that repeated requests with the same seed and parameters should return the same result. Refer to the system_fingerprint response parameter to monitor changes in the backend.

For more information, refer to the OpenAI documentation.

Stop Sequences

Add up to 4 sequences where the API will stop generating further tokens.

Other Input Parameters

Add any additional input parameters by selecting the Parameter Name , the Input Type, and entering the Parameter Value.

For more information, refer to the OpenAI documentation.

Create an Edit

Creates a new edit for the provided input, instruction, and parameters.

Connection

Establish a connection to your OpenAI account.

Model

Select or map the model for which you want to create the edit.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Instruction

Enter instructions that tells the model about the process of editing the input.

Input

Enter the text to use as a starting point for creating the edit.

Temperature

Specify the sampling temperature to use.

A higher value means the model will take more risks. Try 0.9 for more creative applications and 0 (argmax sampling) for ones with a well-defined answer. The default value is 1.

Top p

Specify an alternative to sampling with temperature, called nucleus sampling, where the model considers the results of the tokens with top_p probability mass.

A value of 0.1 means only the tokens comprising the top 10% probability mass are considered.

The default value is 1.

N

Specify the number of edits to generate for the input and instruction. The default value is 1.

Create a Moderation

Classifies if text violates OpenAI’s Content Policy.

Connection

Establish a connection to your OpenAI account.

Input

Enter the text for which you want to create the moderation. The module will classify the content against OpenAI’s Content Policy.

Generate an Image

Generates an image with DALL E given a prompt.

Connection

Establish a connection to your OpenAI account.

Prompt

Enter details of the images you want to generate. The maximum number of characters allowed is 1000.

Size

Select or map the size of the images to generate. It must be one of 256x256512x512, or 1024x1024.

Response Format

Select or map the format in which the generated images are returned.

If you select Base64-encoded PNG, the raw PNG image can be obtained by using the IML function, toBinary(<module_no>.data[].b64_json; base64) in the subsequent modules.

N

Enter the number of images to generate. Enter a number between 1 and 10.

Create a Translation (Whisper)

Creates a translation of an audio into English.

Connection

Establish a connection to your OpenAI account.

File Name

Enter the name of the file you want to translate.

File Data

Enter the data of the file you want to translate.

Model

Select or map the model you want to use. Refer to the OpenAI Audio API documentation for information on available models.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Prompt

Enter text to guide the model’s style or continue a previous audio segment. (Optional) The prompt should be in English.

Temperature

Enter a sampling temperature between 0 and 1. Higher values, such as 0.8, will make the output more random. Lower values, such as 0.2, will make it more focused and deterministic.

Create a Transcription (Whisper)

Creates a transcription of an audio to text.

Connection

Establish a connection to your OpenAI account.

File Name

Enter the name of the file you want to transcribe.

File Data

Enter the data of the file you want to transcribe.

Model

Select or map the model you want to use. Refer to the OpenAI Audio API documentation for information on available models.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Prompt

Enter text to guide the model’s style or continue a previous audio segment. (Optional) The prompt should be in the same language as the audio.

Temperature

Enter a sampling temperature between 0 and 1. Higher values, such as 0.8, will make the output more random. Lower values, such as 0.2, will make it more focused and deterministic.

Language

Enter the two letter ISO code of the input audio’s language.

 

Transform a Text to a Structured Data

Identifies specified information in a prompt’s raw text and returns it as structured data.

Connection

Establish a connection to your OpenAI account.

Model

Select or map the model you want to use to transform text into structured data.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Text to Parse

Enter the text containing the data that you want to transform.

Prompt

Enter a short description explaining what type of data should be extracted from the text entered above.

Structured Data Definition

Enter parameters or map the way in which the structured data should be returned.

Parameter Name

Enter a name for the parameter. Adhere to the following rules:

1. The first character must be either a letter or an underscore.

2. Do not include any commas or spaces.

3. Do not use any special symbols, other than an underscore.

Description

Enter a short description of the parameter.

For example, if would like OpenAI to search for coordinates in the text, you can enter Latitude and Longitude of the location.

Data Type

Select or map the data type format in which the parameter is returned.

If there will be multiple occurrences of similar data types, select an Array option. For example, if there are coordinates of multiple locations to be returned, select Array (text) .

Value Examples

Enter or map examples of possible values to be returned. The more examples you provide, the more accurate the response will be.

Is Parameter Required?

Select or map whether the parameter is required.

When selecting yes, OpenAI will be forced to generate a value in the output, even if no value is detected in the text.

Generate an Audio

Generates an audio file based on the text input and settings specified.

Connection

Establish a connection to your OpenAI account.

Input

Enter text to generate into audio.

The text must be between 1 and 4096 characters long.

Model

Select or map the model you want to use.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

Voice

Select or map the voice to use in the audio. For voice samples, see the OpenAI Voice Options guide.

Output Filename

Enter a name for the generated audio file. Do not include the file extension.

Response Format

Select or map the file format for the generated audio file.

Speed

Enter a value for the speed of the audio.

This must be a number between 0.25 and 4.

Message an Assistant

Sends a message to the selected assistant and gets a response. Optionally, saves the message to the specified thread (conversation).

Connection

Establish a connection to your OpenAI account.

Message

Enter the message to send to the Assistant.

Assistant

Select or map the assistant you would like to use.

Thread ID

Enter the Thread ID where the message will be stored.

To find your Thread ID, go to the OpenAI Playground, open your assistant, and the Thread ID will be visible.

If Thread ID is left empty, we will create a new thread. You can find the new thread’s Thread ID value in the module’s response.

Model

Select or map the model you want to use.

[Note] Note

Ensure you have access to the model you want to work with. Only models available for the account are listed for selection.

File IDs

Select or map a File ID to be used in the message. To select a file, you must first upload the files to OpenAI. This is useful in cases where there are tools like retrievaland code_interpreter that can access and use files enabled in the assistant or in the Toolsparameter.

You can use the OpenAI: Upload a File module to upload files.

Tools

Specify tools in order to override the tools the assistant can use when generating the response.

Instructions

Enter instructions in order to override the default system message of the assistant when generating the response.

Upload a File

Uploads a file that will be available for further usage across the OpenAI platform.

Connection

Establish a connection to your OpenAI account.

File Name

Enter the name of the file to be uploaded, including the file extension. For example, myFile.png.

Supported file types depend on the option selected in the Purposefield below. The Fine Tune purpose supports only .jsonl files.

See the Assistants Tools guide to learn more about the supported file types.

File Data

Enter the file data to be uploaded. You can retrieve file data using the HTTP: Get a file module.

Purpose

Select or map the purpose. Select Assistants for Assistants and Messages and Fine Tune for Fine-tuning.

Make an API Call

Performs an arbitrary authorized API call.

[Note] Note

For the list of available endpoints, see OpenAI ChatGPT API Documentation.

Connection

Establish a connection to your OpenAI account.

URL

Enter a path relative to https://api.openai.com. For example, /v1/models.

Method

GET

to retrieve information for an entry.

POST

to create a new entry.

PUT

to update/replace an existing entry.

PATCH

to make a partial entry update.

DELETE

to delete an entry.

Headers

Enter the desired request headers. You don’t have to add authorization headers; we already did that for you.

Query String

Enter the request query string.

Body

Enter the body content for your API call.

Examples of Use – List Models

The following API call returns all pages from your OpenAI account.

URL:/v1/models

Method:GET

openai-5.png

The search matches can be found in the module’s Output under Bundle > Body > data

In our example, 69 models were returned:

openai-4.png