You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I will concede that there are issues with some of the instructions and some missing files. For those that want to try the CodeSolutionAgency you can follow the instructions as outlined. I would recommend you create a virtual environment with venv for this.
Python 3.11 tends to work with most current package without issue
Do this in the clone of the repo and install the requirements
Change to CodeSolutionAgency
There is no requirements.txt, you can create this yourself or just pip install the packages needed which are
selenium
instructor
webdriver-manager
selenium_stealth
Select the PlannerAgent then just ender your instructions, as an example: "In python write the code to calculate the value of pi to 22 decimals"
There will be many iterations as each agent attempts to engage and carry out tasks.
Hope this helps for just this example
** Be aware that these agents can spin and churn even when the result seems spot on. As such, extremely chatty and can take considerable time over say straight up querying an LLM for the direct result. Agents need rule based governance over the top to help stop spinning and intreating over the problem domain endlessly
The text was updated successfully, but these errors were encountered:
I will concede that there are issues with some of the instructions and some missing files. For those that want to try the CodeSolutionAgency you can follow the instructions as outlined. I would recommend you create a virtual environment with venv for this.
Python 3.11 tends to work with most current package without issue
python3.11 -m venv venv
source venv/bin/activate && python --version
Do this in the clone of the repo and install the requirements
Change to CodeSolutionAgency
There is no requirements.txt, you can create this yourself or just pip install the packages needed which are
selenium
instructor
webdriver-manager
selenium_stealth
pip install selenium selenium_stealth webdriver-manager instructor
Then simply execute: python agency.py
Be sure you have the .env set in this directory with your OPENAI_API_KEY
You can then hit the local site: http://localhost:7860 for the UI
Select the PlannerAgent then just ender your instructions, as an example: "In python write the code to calculate the value of pi to 22 decimals"
There will be many iterations as each agent attempts to engage and carry out tasks.
Hope this helps for just this example
** Be aware that these agents can spin and churn even when the result seems spot on. As such, extremely chatty and can take considerable time over say straight up querying an LLM for the direct result. Agents need rule based governance over the top to help stop spinning and intreating over the problem domain endlessly
The text was updated successfully, but these errors were encountered: