Zijie Yan(@zijie_y):Neon’s scientific traces are long and vary widely in length. Training on them shouldn’t mean GPUs chewing through padding or waiting for each other. Here’s how we pack sequences, balance the work, and distribute long contexts across GPUs at @periodiclabs 👇

Zijie Yan

@zijie_y

working on training infra at @periodiclabs prev led Megatron MoE at @nvidia

加入 April 2023

152 正在關注 782 粉絲

Zijie Yan@zijie_y

2026.09.17 16:07

Neon’s scientific traces are long and vary widely in length. Training on them shouldn’t mean GPUs chewing through padding or waiting for each other. Here’s how we pack sequences, balance the work, and distribute long contexts across GPUs at @periodiclabs 👇