
Resolves YES if I create the market "will it rain tomorrow?" with the description "asking for bard" and then embed that market in the comments here.
🏅 Top traders
| # | Trader | Total profit |
|---|---|---|
| 1 | Ṁ45 | |
| 2 | Ṁ21 | |
| 3 | Ṁ14 | |
| 4 | Ṁ13 | |
| 5 | Ṁ7 |
@LukeHanks Sure. Resolves YES if I create the market "will it rain tomorrow?" with the description "asking for bard" and then embed that market in the comments here. That work?
@LukeHanks You can take my bets as signal. You will notice I am not buying YES on the GPT-4 market because it is terrible at this.
@BTE Thanks for the response. I'm more interested in how you will and won't help Bard along. Will you edit the code at all yourself? Will you give it feedback to improve the output? Will your prompt or feedback include any code? Is there a limit to how many attempts you will give it? Will you follow the same procedure and restrictions when you test GPT-4?
Sorry I'm asking a lot of questions. I should have realized how many assumptions I was making when I bet in this market.
@LukeHanks All great questions!! None of them got it zero shot, so I have been copying the error message back into the prompt and given Bard a few shots now and it has made negative progress. I will make small changes to fix errors in syntax code and details like my API key and the content of the question. The only prompt I used for zero shot was "lets write a python script to create a market on manifold.markets". Claude got really close zero shot and succeed on a few shots. GPT-4 is hopeless. Doesn't even know Manifold exists. The others seem to have real time access to the internet.