browsergym-webarena

WebArena benchmark for BrowserGym

This package provides browsergym.webarena, which is an unofficial port of the WebArena benchmark for BrowserGym.

Note: the original WebArena codebase has been slightly adapted to ensure compatibility.

Server installation

You have two options to setup your webarena instance:

• option 1: follow the official webarena README
• option 2: use our unofficial setup scripts

We recommend option 2 as it allows you to easily customize the ports of each webarena domain, and offers a reset functionality that allows browsergym to trigger a full instance reset remotely.

Setup

• Install the package

pip install browsergym-webarena

• Download tokenizer resources

python -c "import nltk; nltk.download('punkt_tab')"

• Setup the URLs as environment variables. The ports for each domain here should correspond to those you used when setting up your webarena instance. Note also the WA_ prefix which is specific to browsergym.

BASE_URL=<YOUR_SERVER_URL_HERE>  # example: "http://myazuremachine.eastus.cloudapp.azure.com"

# webarena environment variables (change ports as needed)
export WA_SHOPPING="$BASE_URL:8082/"
export WA_SHOPPING_ADMIN="$BASE_URL:8083/admin"
export WA_REDDIT="$BASE_URL:8080"
export WA_GITLAB="$BASE_URL:9001"
export WA_WIKIPEDIA="$BASE_URL:8081/wikipedia_en_all_maxi_2022-05/A/User:The_other_Kiwix_guy/Landing"
export WA_MAP="$BASE_URL:443"
export WA_HOMEPAGE="$BASE_URL:80"

# if your webarena instance offers the FULL_RESET feature (optional)
export WA_FULL_RESET="$BASE_URL:7565"

# otherwise, be sure to NOT set WA_FULL_RESET, or set it to an empty string
export WA_FULL_RESET=""

• Setup an OpenAI API key

export OPENAI_API_KEY=...

NOTE: be mindful of costs, as WebArena will call GPT4 for certain evaluations (llm_fuzzy_match).