The killer combination is to have An Ai agent to prompt chatgpt and do all the testing and anoying tasks over the course of a day, then send you a text once gpt gets it right
What’s interesting is that the raw performance (without self-prompting and reasoning) of O1 is actually worse than that of standard GPT-4. Making the model capable of self-prompting and breaking problems into smaller tasks often improves its accuracy.
The killer combination is to have An Ai agent to prompt chatgpt and do all the testing and anoying tasks over the course of a day, then send you a text once gpt gets it right
What’s interesting is that the raw performance (without self-prompting and reasoning) of O1 is actually worse than that of standard GPT-4. Making the model capable of self-prompting and breaking problems into smaller tasks often improves its accuracy.
so like a human ... if you will code without looping a code in mind you hardly code hello world ;)
hi , can u give as the code
cool
3D Tetris=Blockout.