07. Dictionaries for AI Config

Why Dictionaries for AI

AI systems require configuration: model names, hyperparameters, API endpoints. Dictionaries map cleanly to JSON, the format of most AI APIs.

Creating Dictionaries

config = {
    "model": "gpt-4",
    "temperature": 0.7,
    "max_tokens": 500,
    "stop_sequences": ["###", "END"]
}

Accessing Values

model = config["model"]           # Raises KeyError if missing
model = config.get("model")       # Returns None if missing
model = config.get("model", "gpt-3.5")  # Default value if missing

Modifying Dictionaries

config["temperature"] = 0.9       # Update existing
config["api_key"] = "sk-..."      # Add new key
del config["stop_sequences"]      # Delete key

Iterating Over Dictionaries

# Keys only
for key in config:
    print(key)

# Keys and values
for key, value in config.items():
    print(f"{key}: {value}")

# Values only
for value in config.values():
    print(value)

Nested Dictionaries

request = {
    "model": "gpt-4",
    "parameters": {
        "temperature": 0.7,
        "top_p": 0.9,
        "frequency_penalty": 0.0
    },
    "messages": [
        {"role": "system", "content": "You are helpful."},
        {"role": "user", "content": "Hello."}
    ]
}

# Access nested values
temp = request["parameters"]["temperature"]
first_message = request["messages"][0]["content"]

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.

Local verification checkpoint

Run the smallest example from this chapter in a local workspace and record the package version, runtime, data path, and observed output. If the result depends on model size, vector count, CPU/GPU backend, or available memory, note that constraint beside the exercise so the lesson remains reproducible.