Prompt file update for more specific command identification

[robinicus]

Taking inspiration from the method that Langchain uses to instruct the bot on which tool to use, I've fleshed out each of the tools with a brief description to help the bot make a better decision. In testing it appears to work well and the bot seems to find its way to the goal faster.

[robinicus]

Fyi, here's what I was referring to regarding Langchain: https://python.langchain.com/en/latest/modules/agents/agents/examples/mrkl_chat.html - note their descriptions next to each tool.

[ryanmac]

Recommended change in PR #170 to switch the command and thoughts.

[amauryfischer]

@robinicus This seems a very good idea.
Did you see improvement in usage of this prompt ?

[Torantulino]

The prompt did actually use to be like this, but I reduced it to reduce the token count. I suppose it's probably a good trade-off though.

Will Benchmark!

[nponeccop]

Rebase against the current master, please.

[github-actions]

This pull request has conflicts with the base branch, please resolve those so we can evaluate the pull request.

[p-i-]

This is a mass message from the AutoGPT core team.
Our apologies for the ongoing delay in processing PRs.
This is because we are re-architecting the AutoGPT core!

For more details (and for infor on joining our Discord), please refer to:
https://github.com/Significant-Gravitas/Auto-GPT/wiki/Architecting

[waynehamadi]

@robinicus we're currently building challenges in order to measure the effect of these types of changes, and I would like to discuss with you.

Please join use on Discord through this link https://discord.gg/autogpt (if not already)

DM me on the Auto-GPT discord channel (my discord is merwanehamadi).

[Boostrix]

I've fleshed out each of the tools with a brief description to help the bot make a better decision. In testing it appears to work well and the bot seems to find its way to the goal faster.

You might also want to take a look at the following recent comments (look at the summary that @valayDave come up with):

• GPT-3 often uses commands incorrectly #2987 (comment)
• GPT-3 often uses commands incorrectly #2987 (comment)
• GPT-3 often uses commands incorrectly #2987 (comment)

[Boostrix]

See:

#3954

[Pwuts]

Closing in favor of #4683 and marking with fridge for future inspiration in case we need better command descriptions