Runway 赌上 1000 张 A100：世界模型绕开数字孪生

核心论点 · 点时间戳可跳到原声

原话 · 已逐字校验

There’s gonna be a point where most of content will be generated, and that was maybe the initial thesis of Runway was we will need, as a result of those generative models, rethink how creative tools are made.

会有一个时间点，大部分内容都将被生成出来，这也许就是 Runway 最初的论点：由于这些生成模型，我们需要重新思考创意工具是怎么做的。

Anastasis0:00

we signed this deal to build a cluster of a thousand A100s, which at the time we were a Series B startup. That was a almost, slightly irrational decision maybe, but we really believed that if we trained a video model at a large scale, we would get, like, a great model at the end.

我们签了建一个一千张 A100 集群的协议，当时我们还是一家 Series B 创业公司。那也许是一个几乎有点不理性的决定，但我们真的相信，如果大规模训练一个视频模型，最终会得到一个很棒的模型。

Anastasis8:12

There were a lot of, a lot of chatter on Twitter about Runway. Runway’s done. like, there is no way Runway will catch up.

Twitter 上有很多很多议论，说 Runway 完了，Runway 不可能追得上了。

Anastasis34:58

It’s the end-to-end philosophy applying applied to front ends.

这是把端到端哲学应用到前端上。

Anastasis45:18

the most plentiful source of data will ultimately wins.

最丰富的数据源最终会赢。

Anastasis1:01:55

I call it the lucid dream test.

我把它叫做清醒梦测试。

Anastasis1:07:47

Like one nice thing about image and video models is you can immediately tell with your eyes like what feels good from an aesthetic standpoint.

图像和视频模型的一个好处是，你一眼就能看出什么在审美上是对的。

Anastasis1:23:12

And now my sense is increasingly people are gonna care about latency more and more. As those models become better, the ability to iterate very quickly becomes more important.

现在我的感觉是，人们会越来越在乎延迟。随着模型变好，快速迭代的能力会变得更重要。

Anastasis1:26:08

数字与实体

术语

lucid dream test清醒梦测试

分不清眼前是真实 pass-through 还是生成渲染，就说明世界模型够好了。

step distillation步数蒸馏

把扩散模型的多步采样压缩成少步，以降低服务成本、提高生成速度。

error accumulation误差累积

自回归生成中，每帧的小误差被喂回模型，随时间不断放大。

counterfactual generation反事实生成

让模型生成「没发生」的结果，比如没进球，用于模拟失败和在线 RL。

收听指南