Simply set the debate topic and list of personas, run the app and watch the AI personas debate.
- To install, run the shell/install.sh script (Run this script any time you update dependencies).
-
Install the app by running shell/install.sh in a command prompt.
-
Set required environment variables:
LANGCHAIN_API_KEY=[REQUIRED] SERPAPI_API_KEY=[REQUIRED] APP_DIALOGUE_MODEL_API_KEY=[REQUIRED] # Comma seperated list of personas e.g: "AI enthusiast, AI skeptic" or "Atheist, Christian" APP_DIALOGUE_PERSONAS="Atheist, Christian" APP_DIALOGUE_TOPIC="Was the universe created by an intelligent designer?" APP_DIALOGUE_MODEL_PROVIDER="[OPTIONAL, default=openai]" APP_DIALOGUE_MODEL="[OPTIONAL, default=gpt-4]"
For a full ist of environment variables, see Environment.
-
Run the app by running shell/run.sh in a command prompt.
- If you want to do some development or run tests, run the shell/install.dev.sh script (only once).
- ❌ ⚠ Do not update
requirements.txt
directly. Rather updaterequirements.in
and run shell/install.sh - ✅ Update the CHANGELOG.md file, with your changes.
- To run the tests, run the shell/run.tests.sh script.
-
Based on the fabulous work being done by langchain.
-
The topic should match the personas for example:
Topic: "The impact of automation and artificial intelligence on employment." Personas: "AI enthusiast, AI skeptic"
Topic: "Was the universe created by an intelligent designer?" Personas: "Atheist, Christian"
For example to use anthropic
instead of openai
:
-
Only openai is installed by default. So install anthropic:
pip install -qU langchain-anthropic
. -
In your environment set the provider/model (
anthropic
) related properties.APP_DIALOGUE_MODEL_API_KEY=[REQUIRED] APP_DIALOGUE_MODEL_PROVIDER=anthropic APP_DIALOGUE_MODEL="claude-3-opus-20240229" # or any other model
-
Go through the steps in the quick start section above.