I've always wanted to make one of these graphs. also, first time working on an eval. do people like run this multiple times? this is kind of expensive :'(. also interesting that with this specific dataset, they don't do better with more thinking probably because they ran out of time. how do you set the time limits?
18
im sad if gpt 6.2 isn't great. im really tired of anthropic telling us what we can and can't do with our models. does gemini have the new mandate?
1
2
226
has anyone experimented with using different models for hypothesis generation? my theory is a different models will have different biases for testing new ideas / theories
1
1
48
fortune's weave has an incredible amount of chores. not my style. luckily with emulation, you can just cheat out of the chores, but then the game becomes unbalanced. hmm
1
100
i had no idea how bad claude slopped its text until i read some carbon output. dear god anthropic make it stop please
2
121
if anyone has trying to migrate from switch -> emulator for fortune's weave, this is my agent's output for doing save edits
2
1
352
man what's up with everyone saying "of course jensen would cheat" do people have absolutely no respect for marriage and commitment nowadays? people are so disgusting! no wonder why people are so scared to get married nowadays. the idea of covenant just doesn't exist?
5
156
github.com/zortos293/Lichii/… humans aren't allowed here anymore
37




