It did, actually. The model was trained with multiple rounds of reinforcement learning where human judges provided the feedback: first with full answers, and then with ranking of answers as most relevant.
So the model in production is probably frozen, but before that it went through multiple rounds of interaction with the world.
The reinforcement learning was on giving the right answer, not on interacting with the world. But there is movement in the right direction with https://ai.googleblog.com/2022/12/rt-1-robotics-transformer-...
and other RL stuff. (RT-1 isn't RL but there is other related stuff that is)
Oh, you meant interaction as a joint training with images, actions, feedback etc. That would be the next generation I guess.
I am simply thinking of interaction here as similar to learning a language in a classroom. First the teacher provides sample questions/answers, then the teacher asks the students to come up with answers themselves, and tell them which one is better. The end result here is I think ChatGPT is quite good at answering questions and can pass as a human, especially if it's augmented with a fact database, so obviously wrong answers can be pruned.
Nothing really. He addresses this in his post. But also the purpose isn't to sell the song (it's already creative commons). Purely just a commemorative token (and the only one he plans to make. It's not about ownership of the song though.
But the reason why this is such a great example for me is because the token actually contains something substantial. This song is now permanently a part of the blockchain, as opposed to just some link to an image host that might not exist in a decade.
Leverage. Call options at 2000 strike expiring on 9/16/22 cost roughly 1000 now. For each dollar GOOG goes up (or down), the deep in money options' price will probably move roughly one dollar in the same direction, as long as GOOG price doesn't go down too much. So instead of investing on 100 Google shares for 300K dollars, she can buy 3 call options (each call option is for 100 shares) for the same money and makes three times as much if GOOG goes up, and loses three times as much (or more) if it goes down: e.g., GOOG is under 2000 in 9/16, she will lose all 200K dollars.
Essentially, she's bullish on GOOG. She's also rich enough to take the risk to leverage 3 times. She's even more bullish on RBLX though.
