Skip to main content

FIFO Algorithm

The FIFO (First-In-First-Out) algorithm is a simple but effective way to manage conversation context. It keeps the most recent messages while respecting token limits.

Features​

  • Simple: First-in, first-out message selection
  • Predictable: Always keeps most recent messages
  • Efficient: No complex computations
  • Configurable: Adjustable message window size
  • Type Safety: Full type hints support

Installation​

FIFO Algorithm is included with the base package:

pip install memexllm

Basic Usage​

from memexllm.algorithms import FIFOAlgorithm

# Create algorithm instance
algorithm = FIFOAlgorithm()

# Optional: limit context window size
algorithm = FIFOAlgorithm(max_messages=50)

Use with OpenAI​

from openai import OpenAI
from memexllm.storage import SQLiteStorage
from memexllm.algorithms import FIFOAlgorithm
from memexllm.integrations.openai import with_history
from memexllm.history import HistoryManager
# Create client with FIFO algorithm
client = OpenAI()
history_manager = HistoryManager(
    storage=SQLiteStorage("chat.db"),
    algorithm=FIFOAlgorithm(max_messages=50)
)
client = with_history(history_manager=history_manager)(client)

# Use as normal
response = client.chat.completions.create(
    model="gpt-3.5-turbo",
    messages=[{"role": "user", "content": "Hello!"}]
)

When to Use​

✅ Good For:

  • Most chat applications
  • Simple chatbots
  • When order matters
  • General conversation
  • Learning MemexLLM

❌ Not Good For:

  • Complex context needs
  • Semantic relevance
  • Token optimization
  • Time-based selection

Best Practices​

  1. Message Window Size:

    • Consider model token limits
    • Account for message lengths
    • Include buffer for system messages

  2. System Messages:

    • Always included in context
    • Count towards message limit
    • Keep them concise

  3. Performance:

    • Efficient by default
    • No optimization needed
    • Minimal memory usage

Next Steps​