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 played for several weeks with different GPT Agents. And I'm really interested in BeeBot's potential, approach and tools-focusing.
Thank you a lot for your outstanding work!
During my tries with BeeBot, I mentioned interesting behaviour. Sometimes AI decides to choose a command that does not exist in the list of available commands but is close to it. For example:
save_data(data="Important data", file_name="data.txt") instead of write_file(filename="data.txt", text_content="Important data")
python(function_call="google_search", question="How much is the fish?") instead of google_search (query="How much is the fish")
Finally, I wrote and tested a very small fix that asks AI to map a non-exited command to an exited one if such behaviour occurs.
Do you think that this approach makes sense and is it worse to make a PR for it?
Hi, guys.
This is Roman - CTO from Sinai RnD.
I played for several weeks with different GPT Agents. And I'm really interested in BeeBot's potential, approach and tools-focusing.
Thank you a lot for your outstanding work!
During my tries with BeeBot, I mentioned interesting behaviour. Sometimes AI decides to choose a command that does not exist in the list of available commands but is close to it. For example:
save_data(data="Important data", file_name="data.txt")
instead ofwrite_file(filename="data.txt", text_content="Important data")
python(function_call="google_search", question="How much is the fish?")
instead ofgoogle_search (query="How much is the fish")
Finally, I wrote and tested a very small fix that asks AI to map a non-exited command to an exited one if such behaviour occurs.
Do you think that this approach makes sense and is it worse to make a PR for it?
@erik-megarad @romanzubenko
The text was updated successfully, but these errors were encountered: