OpenAI
Introduction
The KOG library provides a comprehensive set of tools and functionalities for interacting with AI and machine learning models, particularly focusing on tasks such as natural language processing, document analysis, and interaction with large language models (LLMs). This documentation outlines the various components of the KOG library, including class definitions, function definitions, procedural calls, constants, and utility functions.
Prerequisites for Using the KOG Library
Before you can start using the Kognitos ensure that you have the following prerequisites:
- API Access: Depending on the functionalities you intend to use, you may need API access to services like OpenAI
- Basic Understanding of AI/ML Concepts: Familiarity with concepts such as tokens, models, and natural language processing will help you utilize the library more effectively..
Class Definitions
LLMRegions
The LLMRegions
class defines the regions where the large language models are hosted.
LLMProviders
The LLMProviders
class lists the available providers for large language models.
LLMFeatures
The LLMFeatures
class outlines the features supported by the large language models.
LLMAttributes
The LLMAttributes
class specifies the attributes of the large language models.
Function Definitions
_get_llm_region_and_provider
Retrieves the region and provider for a given large language model.
_get_model_with_version
Fetches the model along with its version.
get_model_with_version
Public function to get the model with its version.
_get_api_info
Retrieves API information for a given service.
_get_args_from
Extracts arguments from a given input.
get_secret
Fetches a secret from the secret management service.
openai_client
Initializes and returns an OpenAI client.
_get_openai_max_response_tokens
Calculates the maximum number of response tokens for OpenAI.
get_gpt_response
Fetches a response from GPT models.
num_tokens_from_images
Calculates the number of tokens from images.
extract_page_number
Extracts the page number from a document.
get_document_uri_and_pages
Retrieves the URI and pages of a document.
get_document_page_images
Fetches the images of document pages.
get_gpt_image_messages
Generates GPT messages from images.
num_tokens_from_messages
Calculates the number of tokens from messages.
read_jsons_from_result
Reads JSON objects from a result set.
extract_answers
Extracts answers from a given input.
get_openai_answers
Fetches answers from OpenAI.
verify_assertion_openai
Verifies an assertion using OpenAI.
extract_noun_from_sentence_openai
Extracts nouns from a sentence using OpenAI.
describe_thing
Generates a description of a given thing.
to_ask_koncierge
Asks a question to the Koncierge service.
to_find_a_thing_in_a_string
Finds a specific thing in a string.
to_ask_koncierge_for_a_documents_thing
Asks Koncierge for a specific thing in a document.
get_raw_document_text
Retrieves the raw text from a document.
get_invoices_gpt
Fetches invoices using GPT models.
Procedural Calls or Decorators
@procedure Decorator
The @procedure
decorator is used to define procedural calls in the following functions:
to_ask_koncierge
to_find_a_thing_in_a_string
to_ask_koncierge_for_a_documents_thing
Constants and Data Structures
Token Constants
TOKEN_1K
TOKENS_4K
TOKENS_8K
TOKENS_16K
TOKENS_128K
Available LLMs
AVAILABLE_LLMS_OPENAI
DEFAULT_OPENAI_MODEL
KONCIERGE_MODEL_MAP_OPENAI
AVAILABLE_LLMS_AZURE
KONCIERGE_MODEL_MAP_AZURE
Utility Functions
get_secret
Fetches a secret from the secret management service.
openai_client
Initializes and returns an OpenAI client.
num_tokens_from_images
Calculates the number of tokens from images.
extract_page_number
Extracts the page number from a document.
get_document_uri_and_pages
Retrieves the URI and pages of a document.
get_document_page_images
Fetches the images of document pages.
get_gpt_image_messages
Generates GPT messages from images.
num_tokens_from_messages
Calculates the number of tokens from messages.
read_jsons_from_result
Reads JSON objects from a result set.
extract_answers
Extracts answers from a given input.
get_openai_answers
Fetches answers from OpenAI.
verify_assertion_openai
Verifies an assertion using OpenAI.
extract_noun_from_sentence_openai
Extracts nouns from a sentence using OpenAI.
describe_thing
Generates a description of a given thing.
get_raw_document_text
Retrieves the raw text from a document.
get_invoices_gpt
Fetches invoices using GPT models.
Glossary of Terms
- API (Application Programming Interface): A set of rules and protocols for building and interacting with software applications. The KOG library uses APIs to communicate with services like OpenAI and Azure.
- LLM (Large Language Model): A type of AI model designed to understand and generate human language. Examples include GPT-3 from OpenAI.
- Token: A unit of text that the model processes. Tokens can be as short as one character or as long as one word.
- Koncierge: A service within the Kognitos that handles specific queries and tasks.
Updated 3 months ago