/memoripy

An AI memory layer with short- and long-term storage, semantic clustering, and optional memory decay for context-aware applications.

Primary LanguagePythonApache License 2.0Apache-2.0

Memoripy

Memoripy is a Python library designed to manage and retrieve context-aware memory interactions using both short-term and long-term storage. It supports AI-driven applications requiring memory management, with compatibility for OpenAI and Ollama APIs. Features include contextual memory retrieval, memory decay and reinforcement, hierarchical clustering, and graph-based associations.

Features

  • Short-term and Long-term Memory: Manages memory as short-term or long-term based on usage and relevance.

  • Contextual Retrieval: Retrieves memories based on embeddings, concepts, and past interactions.

  • Concept Extraction and Embeddings: Uses OpenAI or Ollama models for concept extraction and embedding generation.

  • Graph-Based Associations: Builds a concept graph and uses spreading activation for relevance-based retrieval.

  • Hierarchical Clustering: Clusters similar memories into semantic groups to aid in contextually relevant retrieval.

  • Decay and Reinforcement: Older memories decay over time, while frequently accessed memories are reinforced.

Installation

You can install Memoripy with pip:

pip install memoripy

Usage

The following example demonstrates how to set up and use Memoripy in a Python script.

Example: example.py

This example script shows the primary functionality of Memoripy, including initialization, storing interactions, retrieving relevant memories, and generating responses.

from memoripy import MemoryManager, JSONStorage from memoripy.implemented_models import OpenAIChatModel, OllamaEmbeddingModel

def main():
    # Replace 'your-api-key' with your actual OpenAI API key
    api_key = "your-key"
    if not api_key:
        raise ValueError("Please set your OpenAI API key.")

    # Define chat and embedding models
    chat_model_name = "gpt-4o-mini"  # Specific chat model name
    embedding_model_name = "mxbai-embed-large"  # Specific embedding model name

    # Choose your storage option
    storage_option = JSONStorage("interaction_history.json")
    # Or use in-memory storage:
    # from memoripy import InMemoryStorage
    # storage_option = InMemoryStorage()

    # Initialize the MemoryManager with the selected models and storage
    memory_manager = MemoryManager(
        OpenAIChatModel(api_key, chat_model_name),
        OllamaEmbeddingModel(embedding_model_name),
        storage=storage_option
    )

    # New user prompt
    new_prompt = "My name is Khazar"

    # Load the last 5 interactions from history (for context)
    short_term, _ = memory_manager.load_history()
    last_interactions = short_term[-5:] if len(short_term) >= 5 else short_term

    # Retrieve relevant past interactions, excluding the last 5
    relevant_interactions = memory_manager.retrieve_relevant_interactions(new_prompt, exclude_last_n=5)

    # Generate a response using the last interactions and retrieved interactions
    response = memory_manager.generate_response(new_prompt, last_interactions, relevant_interactions)

    # Display the response
    print(f"Generated response:\n{response}")

    # Extract concepts for the new interaction
    combined_text = f"{new_prompt} {response}"
    concepts = memory_manager.extract_concepts(combined_text)

    # Store this new interaction along with its embedding and concepts
    new_embedding = memory_manager.get_embedding(combined_text)
    memory_manager.add_interaction(new_prompt, response, new_embedding, concepts)

if __name__ == "__main__":
    main()

Classes and Modules

  • MemoryManager: Manages memory interactions, retrieves relevant information, and generates responses based on past interactions.

  • MemoryStore: Stores and organizes interactions in short-term and long-term memory, with support for clustering and retrieval based on relevance.

  • InMemoryStorage and JSONStorage: Store memory in either in-memory data structures or JSON files.

  • BaseStorage: Abstract base class for defining storage methods.

##Core Functionalities

  1. Initialize Memory: Load previous interactions from the chosen storage and initialize memory.

  2. Add Interaction: Store a new interaction with its embedding, concepts, prompt, and output.

  3. Retrieve Relevant Interactions: Search past interactions based on a query using cosine similarity, decay factors, and spreading activation.

  4. Generate Response: Combine the current prompt and retrieved interactions to generate a contextually relevant response.

  5. Decay and Reinforcement: Increase decay on unused memories and reinforce frequently accessed memories.

Requirements

Memoripy relies on several dependencies, including:

  • openai

  • faiss-cpu

  • numpy

  • networkx

  • scikit-learn

  • langchain

  • ollama

These dependencies will be installed automatically with pip install memoripy.

License

Memoripy is licensed under the Apache 2.0 License.

Contributing

Contributions are welcome! Feel free to open issues or submit pull requests for improvements.

Contributors