search

haotian-liu / LLaVA

[NeurIPS'23 Oral] Visual Instruction Tuning (LLaVA) built towards GPT-4V level capabilities and beyond.
https://llava.hliu.cc
Apache License 2.0
20.42k stars 2.26k forks source link

[Question] When loading training image data, is it loaded into memory all at once, or is it loaded from disk to memory according to batch_size? #1741

Open yangzhipeng1108 opened 1 month ago

yangzhipeng1108 commented 1 month ago

Question

When loading training image data, is it loaded into memory all at once, or is it loaded from disk to memory according to batch_size?