|
|
楼主 |
发表于 2024-12-18 12:35:32
|
显示全部楼层
"Grok 2 is incredible, with the largest models estimated to have been trained on around 1 petabyte of data. We are just getting started. By 2025, the world will generate 200 zettabytes of data — that's 200 million petabytes!
Imagine what happens when Grok 3, 4, 5, 6, and beyond tap into this massive ocean of data for training and inference. I'm looking forward that future! 🚀" -
@MichaelDell
and what do we need to process more data? $NVDA
由
翻译自 英语
“Grok 2 令人难以置信,据估计,最大的模型是在大约 1PB 的数据上进行训练的。我们才刚刚起步。到 2025 年,世界将产生 200ZB 的数据——也就是 2 亿 PB!
想象一下,当 Grok 3、4、5、6 及更高版本利用这个庞大的数据海洋进行训练和推理时会发生什么。我期待着这样的未来! 🚀 “-
@MichaelDell
我们需要处理什么来处理更多数据? $NVDA |
|