Noam Brown

speaker

1,199 appearances 1 recordings 1 series first heard Dec 2022 last heard Dec 2022

Noam Brown’s voice in public audio — every appearance, attributed to the second.

Trend

recordings per month · last 12 months

No recordings in the last 12 months.Older appearances are listed below; set an alert to hear about the next one.

newest first · ▶ plays the moment

We have to understand how the other person is likely to play rather than just assuming that they're going to play like a machine.

So the way that RL and planning is done is actually not using language.

So we're, we're coming up with this like plan for the action that we're going to play and the other person's going to play.

And then we feed that action into the dialogue model that will then send a message according to those plans.

Basically one message at a time.

So we'll feed into the dialogue model, like here are the actions that you should be discussing.

Here's the message.

Here's like the content of the message that we would like you to send.

And then it will actually generate a message that corresponds to that.

It works surprisingly well.

I mean, there are a lot of ways that this could fail.

So, for example, I mean, you could have a situation where you're basically like –

We don't tell the language model, here are the pieces of our action or the other person's action that you should be communicating.

And so let's say you're about to attack somebody.

You probably don't want to tell them that you're going to attack them.

But there's nothing in the language.

The language model is not very smart at the end of the day.

So it doesn't really have a way of knowing, well, what should I be talking about?

Should I tell this person that I'm about to attack them or not?

Um, so we have to like develop a lot of other techniques that, that deal with that.

Showing 861–880 of 1,199 · page 44 of 60