search

DAMO-NLP-SG / VideoLLaMA2

VideoLLaMA 2: Advancing Spatial-Temporal Modeling and Audio Understanding in Video-LLMs
Apache License 2.0
758 stars 53 forks source link

Question on compute #3

Open Utkarsh4430 opened 4 months ago

Utkarsh4430 commented 4 months ago

Hey! Great work, guys! I just wanted to know how much compute did it take to pretrain the model. The number and size of GPUs and the number of hours the model was trained. Thanks!

clownrat6 commented 3 months ago

Based on our experiment settings, pretraing stage and finetuning stage take around 5\~6 hours and 16\~17 hours on 32 A100, respectively.

pritamqu commented 3 months ago

did you use 40GB or 80GB A100?