newsence
來源篩選

@JasonBotterill: After watching Jerry’s post-OpenAI interviews it seems he genuinely believed you could just scale RL...

Twitter

After watching Jerry’s post-OpenAI interviews it seems he genuinely believed you could just scale RL to AGI a year ago, but something changed and now he believes static weights will ever be enough and no one is investing serious amounts of compute in trying continual learning

newsence

Loading

Fetching article data