Skip to content

Model-specific ChatBots

While ChatBot instances can be created from any Langchain Chat interface, we created some convenient superclasses that have varying levels of model-specific behavior.

Model-specific chatbots only differ from ChatBot in that they define static factory constructor methods, all named new. As each chat model needs to be installed separately, they must be accessed via separate imports.

1
2
3
4
import sys
sys.path.append('../src/')

import simplechatbot

I will use the keychain to manage API keys for OpenAI and Mistral.

1
keychain = simplechatbot.APIKeyChain.from_json_file('../keys.json')

Notice that we use a separate import statement to explicitly import the model-specific chatbots.

1
2
3
4
5
from simplechatbot.ollama_agent import OllamaAgent

agent = OllamaAgent.new(
    model_name = 'llama3.1', 
)

1
2
3
4
5
6
from simplechatbot.openai_agent import OpenAIAgent

agent = OpenAIAgent.new(
    model_name = 'gpt-4o-mini', 
    api_key=keychain['openai'],
)

1
2
3
4
5
6
from simplechatbot.mistral_agent import MistralAgent

agent = MistralAgent.new(
    model_name = 'mistral-large-latest', 
    api_key=keychain['mistral'],
)