A sandbox that talks back
I gave a 2B model a real Linux box to run code in. It kept lying about code it never ran. What fixed it wasn’t a better prompt, but tool results that tell the model exactly what went wrong and what to do next.
I gave a 2B model a real Linux box to run code in. It kept lying about code it never ran. What fixed it wasn’t a better prompt, but tool results that tell the model exactly what went wrong and what to do next.