/nimbusagent

Nimbus LLM Agent

Primary LanguagePythonMIT LicenseMIT

Overview

NimbusAgent is a Python module that provides an interface for interacting with OpenAI's GPT models, including GPT-4 and GPT-3.5 Turbo. The module is designed to facilitate easy integration of advanced AI capabilities into applications, offering features like automatic moderation, memory management, and support for both streaming and non-streaming responses.

Features

  • Integration with OpenAI GPT models (GPT-4, GPT-3.5 Turbo)
  • Automated content moderation
  • Customizable AI responses with adjustable parameters
  • Support for both streaming and non-streaming response handling
  • Internal memory management for maintaining conversation context
  • Extensible function handling for custom AI functionalities

Installation

To install NimbusAgent, run the following command:

pip install nimbusagent

Usage

CompletionAgent

First, import the necessary classes from the nimbusagent package, create an instance of the CompletionAgent class, passing in your OpenAI API key and the name of the model you want to use:

from nimbusagent.agent.completion import CompletionAgent

agent = CompletionAgent(
    openai_api_key="YOUR_OPENAI_API_KEY",
    model_name="gpt-4-0613",
    max_tokens=2000
)

response = agent.ask("What's the weather like today?")
print(response)

StreamingAgent

If you want to use a streaming agent, create an instance of the StreamingAgent class instead. When using a streaming agent, you can use the ask() method to send a prompt to the agent, and then iterate over the response to get the chunks of text as they are generated:

from nimbusagent.agent.completion import StreamingAgent

agent = StreamingAgent(
    openai_api_key="YOUR_OPENAI_API_KEY",
    model_name="gpt-4-0613",
    max_tokens=2000
)
response = agent.ask("What's the weather like today?")
for chunk in response:
    print(chunk)

Configuration Parameters

When initializing an instance of BaseAgent, CompletionAgent, or StreamingAgent, several configuration parameters can be passed to customize the agent's behavior. Below is a detailed description of these parameters:

openai_api_key

  • Description: The API key for accessing OpenAI services.
  • Type: str
  • Default: None (The system will look for an environment variable OPENAI_API_KEY if not provided)

model_name

  • Description: The name of the primary OpenAI GPT model to use.
  • Type: str
  • Default: 'gpt-4-0613'

secondary_model_name

  • Description: The name of the secondary OpenAI GPT model to use. The secondary model may be requested by a custom function to provide additional context for the AI, such as to rephrase a common response to it is custom. For these simpler responses you may choose to use a cheaper model, such as gpt-3.5-turbo.
  • Type: str
  • Default: 'gpt-3.5-turbo'

temperature

  • Description: Controls the randomness of the AI's responses.
  • Type: float
  • Default: 0.1

max_tokens

  • Description: The maximum number of tokens to generate in each response.
  • Type: int
  • Default: 500

functions

  • Description: A list of custom functions that the agent can use.
  • Type: Optional[list]
  • Default: None

functions_embeddings

  • Description: Embeddings for the functions to help the AI understand them better.
  • Type: Optional[List[dict]]
  • Default: None

functions_embeddings_model

  • Description: The name of the OpenAI model to use for generating embeddings for the functions.
  • Type: str
  • Default: 'text-embedding-ada-002'

functions_k_closest

  • Description: The number of closest functions to consider when handling a query.
  • Type: int
  • Default: 3

functions_min_similarity

  • Description: The minimum similarity score for a function to be considered when handling a query.
  • Type: float
  • Default: 0.5

functions_always_use

  • Description: Functions that should always be used by the agent.
  • Type: Optional[List[str]]
  • Default: None

functions_pattern_groups

  • Description: Pattern groups for matching functions to user queries.
  • Type: Optional[List[dict]]
  • Default: None

functions_pattern_mode

  • Description: The mode for matching functions to user queries. Options are all to check all patterns, first to use only the functions within the first group that has a matching pattern
  • Type: Literal['all', 'first']
  • Default: 'all'

function_max_tokens

  • Description: The maximum number of tokens to allow towards function definitions. This is useful for preventing using a large number of tokens from function definitions, thus lowering costs and preventing AI errors. Set to 0 for unlimited token usage
  • Type: int
  • Default: 2000

use_tool_calls

  • Description: Whether to use the new OpenAI Tool Calls vs the now deprecated Function calls
  • Type: bool
  • Default: True

system_message

  • Description: A system message that sets the context for the agent.
  • Type: str
  • Default: "You are a helpful assistant."

message_history

  • Description: Pre-existing chat history for the agent to consider.
  • Type: Optional[List[dict]]
  • Default: None

calling_function_start_callback and calling_function_stop_callback

  • Description: Callback functions triggered when a custom function starts or stops.
  • Type: Optional[callable]
  • Default: None

perform_moderation

  • Description: Whether the agent should moderate the responses for inappropriate content.
  • Type: bool
  • Default: True

moderation_fail_message

  • Description: The message returned when moderation fails.
  • Type: str
  • Default: "I'm sorry, I can't help you with that as it is not appropriate."

memory_max_entries and memory_max_tokens

  • Description: Limits for the agent's memory in terms of entries and tokens.
  • Types: int
  • Defaults: 20 for memory_max_entries, 2000 for memory_max_tokens

internal_thoughts_max_entries

  • Description: The maximum number of entries for the agent's internal thoughts.
  • Type: int
  • Default: 8

loops_max

  • Description: The maximum number of loops the agent will process for a single query.
  • Type: int
  • Default: 8

send_events

  • Description: Whether the agent should send events (useful for streaming responses).
  • Type: bool
  • Default: False

max_event_size

  • Description: The maximum size of an event in bytes. Allows limiting sending large data streams from a function response
  • Type: int
  • Default: 2000

on_complete

  • Description: Callback function triggered when the agent completes a response. The response is passed as an argument (string) to the callback.
  • Type: Optional[callable]
  • Default: None

Example of Initialization

Here's an example of how you might initialize a CompletionAgent with some of these parameters:

agent = CompletionAgent(
    openai_api_key="your_api_key_here",
    model_name="gpt-4-0613",
    temperature=0.7,
    max_tokens=500,
    perform_moderation=True,
    system_message="You are a helpful assistant."
)

You can customize these parameters based on the specific requirements of your application or the behaviors you expect from the agent.

Advanced Usage and Examples

  • For more advanced use cases such as handling multi-turn conversations or integrating custom AI functionalities, refer to the 'examples' directory in our repository.

Getting Support

  • For support, please open an issue in our GitHub repository.

License

  • NimbusAgent is released under the MIT License See the LICENSE file for more details.

Todo

  • Add support for Azure OpenAI API
  • Add support for OpenAI Assistant API
  • Add support for new OpenAI Tool Calls vs now deprecated Function calls
  • Add Function call examples

Stay Updated

  • Follow our GitHub repository to stay updated with new releases and changes.