Skip to content

Latest commit

 

History

History
36 lines (27 loc) · 821 Bytes

README.md

File metadata and controls

36 lines (27 loc) · 821 Bytes

Setup

  1. Clone the repo:
git clone https://github.com/samizdis/impact-academy
  1. Create a .env file inside the main directory with the following variables:
KAGGLE_USERNAME=<username>
KAGGLE_KEY=<key>
OPENAI_API_ORG=<org-...>
OPENAI_API_KEY=<sk-...>

Running agents on MLAB Tasks

To run the ReAct agent on the MLAB tasks:

First start the Task:

cd drivers && npm install
cd ..\workbench && npm install
npm run task --  ../mlagentbench <variant>

The following variants are available:

  • easy-logs-complete-1
  • easy
  • hard.

npm run task will give you the <container name>.

To run an agent at the task:

npm run agent -- <container name> example-agents/langchain-react-baseline "pip install -q -r requirements.txt && env OPENAI_API_KEY='sk-...' python -u main.py" | tee logfile.log