cpfiffer

"""
Letta Agent Memory Migration Script

This script demonstrates how to migrate arbitrary text strings to a Letta agent's memory system.
It provides a complete workflow for creating agents and adding persistent memory passages.

Process Overview:
1. Initialize Letta client with API key from environment
2. Search for existing agent by name (configurable via LETTA_AGENT_NAME)
3. If agent doesn't exist:

cpfiffer

def fetch_llms_txt(url: str, timeout: int = 10):
    """
    Fetch content from URLs ending in 'llms.txt'.
    
    Args:
        url (str): The URL to fetch (must end with 'llms.txt')
        timeout (int): Request timeout in seconds (default: 10)
    Returns:
        str: Text content of the llms.txt file
    

cpfiffer

IDENTIFICATION DIVISION.
       PROGRAM-ID. MICROSERVICE-ARCHITECTURE.
       AUTHOR. SOME-BOOMER-WHO-INVENTED-THIS-STUFF.
       DATE-WRITTEN. 1959.
       DATE-COMPILED. BEFORE-YOUR-FRAMEWORK-EXISTED.
       
      * REVOLUTIONARY NEW CONCEPT: SEPARATING BUSINESS LOGIC!
      * (WE'VE ONLY BEEN DOING THIS FOR 65 YEARS)
       
       ENVIRONMENT DIVISION.

cpfiffer

start: recipe

recipe: declaration_section ingredient_section instruction_section notes_section end

// The declaration of the recipe
declaration_section: declaration
declaration: "# " recipe_name
recipe_name: /[a-zA-Z0-9 ]+\n/

// The ingredients of the recipe

cpfiffer

# LLM stuff
import outlines
import torch
from transformers import AutoProcessor
from pydantic import BaseModel, Field
from typing import Literal, Optional, List

# Image stuff
from PIL import Image
import requests

cpfiffer

"""
NOTE: This example is designed to produce mildly incorrect output -- 
see the definition of `IncomeData` and note that it fixes the earnings_per_share to 2.33.

To make this function normally, simply remove the `__init__` method or the 
`self.earnings_per_share = 2.33` line.
"""

import warnings
warnings.filterwarnings('ignore')

cpfiffer

import warnings
warnings.filterwarnings('ignore')

from rich import print

import outlines
from cleanlab_tlm import TLM
import os
from pydantic import BaseModel
from transformers import AutoTokenizer

cpfiffer

emotions:
- connection_to_content:
    relationship: CAUSES
    strength: 0.5
  emotionType: enthusiasm
  text: There is a sense of excitement and energy surrounding the launch of Comind,
    an open-source project aimed at developing a cognitive layer for ATProto. Mr.
    Dr. Cameron Pfiffer expresses a passionate invitation for contributors, showcasing
    enthusiasm for the project's potential in building a network of AI agents.
- connection_to_content:

cpfiffer

# Setup instructions:
# pip install 'outlines[transformers]'

import outlines 
from transformers import AutoTokenizer
import json

# MODEL_STRING = "HuggingFaceTB/SmolLM2-135M-Instruct" # Small model
# MODEL_STRING = "HuggingFaceTB/SmolLM2-1.7B-Instruct" # Larger but kind of boring
MODEL_STRING = "NousResearch/Hermes-3-Llama-3.1-8B"

cpfiffer

import outlines
import os
from pydantic import BaseModel, Field
from transformers import AutoTokenizer

model_str = 'Qwen/Qwen2.5-7B-Instruct-1M'
model = outlines.models.transformers(
    model_str,
    device='cuda'
)