changlin31 / DNA

(CVPR 2020) Block-wisely Supervised Neural Architecture Search with Knowledge Distillation
234 stars 35 forks source link

# Is the get_model_complexity_info function called in any document? #16

Closed janezzzz closed 3 years ago

janezzzz commented 4 years ago

Hi, I'm a little confused with the use of flops_counter. Is this function used for calculating the number of FLOPs and parameters after training? Or to provide some constraints during the searching? Thanks!

changlin31 commented 4 years ago

Hi @janezzzz , We only call get_model_parameters_number in process_potential.py to calculate the parameters during searching. The look-up table of FLOPS in the code of process_potential.py is pre-calculated using flops_counter.py and recorded manually to prevent repeated forwarding during search.