search

han-cai / PathLevel-EAS

Path-Level Network Transformation for Efficient Architecture Search, in ICML 2018.
113 stars 21 forks source link

CUDA OOM with 10 GPUs #7

Open Faldict opened 6 years ago

Faldict commented 6 years ago

I have run run_exp.py with 10 GPUs, but it occurs out of memory errors:

RuntimeError: CUDA error: out of memory

As mentioned in your paper, you use only 5 GPUs for the experiments. So what's the problem?