update

README.md

@@ -16,7 +16,8 @@
 
 <p align="center">
 <a href="https://webarena.dev/">Website</a> •
-<a href="https://arxiv.org/abs/2307.13854">Paper</a>
+<a href="https://arxiv.org/abs/2307.13854">Paper</a> •
+<a href="https://docs.google.com/spreadsheets/d/1M801lEpBbKSNwP-vDBkC_pF7LdyGU1f_ufZb_NWNBZQ/edit?usp=sharing">Leaderboard</a>
 </p>
 
 ![Overview](media/overview.png)
@@ -110,7 +111,7 @@ python run.py \
 This script will run the first example with GPT-3.5 reasoning agent. The trajectory will be saved in `<your_result_dir>/0.html`
 
 ## Develop Your Prompt-based Agent
-1. Define the prompts. We provide two baseline agents whose correrponding prompts are listed [here](./agent/prompts/raw). Each prompt is a dictionary with the following keys:
+1. Define the prompts. We provide two baseline agents whose corresponding prompts are listed [here](./agent/prompts/raw). Each prompt is a dictionary with the following keys:
 ```python
 prompt = {
   "intro": <The overall guideline which includes the task description, available action, hint and others>,

browser_env/helper_functions.py

@@ -154,7 +154,7 @@ def render(
         if render_screenshot:
             # image observation
             img_obs = observation["image"]
-            image = Image.fromarray(img_obs)
+            image = Image.fromarray(img_obs)  # type:ignore
             byte_io = io.BytesIO()
             image.save(byte_io, format="PNG")
             byte_io.seek(0)

environment_docker/README.md

@@ -22,12 +22,13 @@ Name: webarena
 ID: ami-06290d70feea35450
 ```
 
+1. Create a security group that allows all inbound traffic.
 
-1. Create an instance (recommended type: t3a.xlarge, 1000GB EBS root volume) from the webarena AMI, and allow all inbound traffic in the security group, remember to select SSH key-pair.
+2. Create an instance (recommended type: t3a.xlarge, 1000GB EBS root volume) from the webarena AMI. Use the security group just created and remember to select SSH key-pair.
 
-2. Create an Elastic IP and bind to the instance to associate the instance with a static IP and hostname. Take note of the hostname, usually in the form of "ec2-xx-xx-xx-xx.us-east-2.compute.amazonaws.com". This will be used as "<your-server-hostname>" in the following commands.
+3. Create an Elastic IP and bind to the instance to associate the instance with a static IP and hostname. Take note of the hostname, usually in the form of "ec2-xx-xx-xx-xx.us-east-2.compute.amazonaws.com". This will be used as "<your-server-hostname>" in the following commands.
 
-3. Log into the server, start all dockers by:
+4. Log into the server, start all dockers by:
 ```bash
 docker start gitlab
 docker start shopping
@@ -40,7 +41,7 @@ docker compose start
 
 :clock1: wait ~1 min to wait all services to start
 
-4. Run
+5. Run
 ```bash
 docker exec shopping /var/www/magento2/bin/magento setup:store-config:set --base-url="http://<your-server-hostname>:7770" # no trailing /
 docker exec shopping mysql -u magentouser -pMyPassword magentodb -e  'UPDATE core_config_data SET value="http://<your-server-hostname>:7770/" WHERE path = "web/secure/base_url";'
@@ -57,6 +58,9 @@ docker exec gitlab sed -i "s|^external_url.*|external_url 'http://<your-server-h
 docker exec gitlab gitlab-ctl reconfigure
 ```
 
+You should be able to access your environment websites now, and stop reading.
+However, if you are unable to use AWS AMI, read below to set up on your own machine.
+
 ## Shopping Website (OneStopShop)
 
 Download the image tar from the following mirrors:
@@ -159,7 +163,7 @@ perl -pi -e "s|<your-server-hostname>|${YOUR_ACTUAL_HOSTNAME}|g" webarena-homepa
 
 Then run
 ```
-cd webarena_homepage
+cd webarena-homepage
 flask run --host=0.0.0.0 --port=4399
 ```
 The homepage will be available at `http://<your-server-hostname>:4399`.