How to pass an initial instruction in task_kwargs?

[xhluca]

Is it possible to do somehting like this:

import gymnasium as gym
import browsergym.core  # register the openended task as a gym environment

instruction = "Please find an apple pie recipe and summarize how to make it"

env = gym.make(
    "browsergym/openended", 
    task_kwargs={"start_url": "https://www.google.com/", "instruction": instruction}, 
    wait_for_user_message=True
)
obs, info = env.reset()
done = False
while not done:
    action = ...  # implement your agent here
    obs, reward, terminated, truncated, info = env.step(action)

where instruction = "Please find an apple pie recipe and summarize how to make it" is what I want to pass to the model first.

[xhluca]

Found it, it's

task_kwargs={"start_url": "https://www.google.com/", "goal": instruction}, 

[gasse]

Glad you found it :)