Official code, datasets and checkpoints for "Timer: Generative Pre-trained Transformers Are Large Time Series Models" (ICML 2024)
253
stars
19
forks
source link
请问在使用下游数据集fine-tune时，样本是从整条序列中随机采样，还是仅使用其中连续的一小段？ #8
Closed
BoffinZhang closed 1 month ago
您好，为了模拟真实场景中的少样本情况，我们在实验中是在整条训练集中采样