artin/webarena
1
0
Fork 0
mirror of https://github.com/web-arena-x/webarena.git synced 2026-02-06 11:16:53 +00:00
Code Issues Packages Projects Releases Wiki Activity
Code repo for "WebArena: A Realistic Web Environment for Building Autonomous Agents"
1 Commit 14 Branches 2 Tags 9.9 MiB
Python 90.3%
HTML 4.1%
Jupyter Notebook 3.9%
Shell 1.7%
b454f2dcfd
Go to file
alexisxy b454f2dcfd release commit
2023-07-25 00:30:29 -10:00
.github/workflows release commit 2023-07-25 00:30:29 -10:00
agent release commit 2023-07-25 00:30:29 -10:00
browser_env release commit 2023-07-25 00:30:29 -10:00
config_files release commit 2023-07-25 00:30:29 -10:00
evaluation_harness release commit 2023-07-25 00:30:29 -10:00
llms release commit 2023-07-25 00:30:29 -10:00
media release commit 2023-07-25 00:30:29 -10:00
scripts release commit 2023-07-25 00:30:29 -10:00
tests release commit 2023-07-25 00:30:29 -10:00
.gitignore release commit 2023-07-25 00:30:29 -10:00
.pre-commit-config.yaml release commit 2023-07-25 00:30:29 -10:00
check_errors.sh release commit 2023-07-25 00:30:29 -10:00
LICENSE release commit 2023-07-25 00:30:29 -10:00
prepare.sh release commit 2023-07-25 00:30:29 -10:00
README.md release commit 2023-07-25 00:30:29 -10:00
requirements.txt release commit 2023-07-25 00:30:29 -10:00
run.py release commit 2023-07-25 00:30:29 -10:00
setup.cfg release commit 2023-07-25 00:30:29 -10:00
setup.py release commit 2023-07-25 00:30:29 -10:00

README.md

Python 3.10 pre-commit Code style: black Checked with mypy bear-ified

WebArena: A Realistic Web Environment for Building Autonomous Agents

[Website] [Paper]

Overview

WebArena is a standalone, self-hostable web environment for building autonomous agents. WebArena creates websites from four popular categories with functionality and data mimicking their real-world equivalents. To emulate human problem-solving, WebArena also embeds tools and knowledge resources as independent websites. WebArena introduces a benchmark on interpreting high-level realistic natural language command to concrete web-based interactions. We provide annotated programs designed to programmatically validate the functional correctness of each task.

Note

This README is still under constructions. Stay tuned!

Install

# Python 3.10+
conda create -n webarena python=3.10; conda activate webarena
pip install -r requirements.txt
playwright install
pip install -e .

# optional, dev only
pip install -e ".[dev]"
mypy --install-types --non-interactive browser_env
pip install pre-commit
pre-commit install

Preperation

  • Config the URLs of each website in env_config
  • python scripts/generate_test_data.py will generate individual config file for each test example in config_files
  • bash prepare.sh to obtain the auto-login cookies for all websites
  • export OPENAI_API_KEY=your_key
  • python run.py --instruction_path agent/prompts/jsons/p_cot_id_actree_2s.json --test_start_idx 0 --test_end_idx 1 --model gpt-3.5-turbo --result_dir your_result_dir to run the first example with GPT-3.5 reasoning agent. The trajectory will be saved in your_result_dir/0.html