search

dicarlolab / archconvnets

Architecturally optimized neural networks trained with regularized backpropagation
5 stars 5 forks source link

Saving checkpoints #10

Closed yamins81 closed 10 years ago

yamins81 commented 10 years ago

To store, with experiment_id="mytestrun"

python run_synthetic_training.py --data-path=/export/imgnet_storage_full/yamins_skdata/sythetic_batches_0 --crop=7 --save-path=/export/imgnet_storage_full/ --test-range=950-999 --train-range=0-50 --layer-def=/home/yamins/archconvnets/archconvnets/convnet/ut_model_full/layer_nofc_0.cfg --layer-params=/home/darren/archconvnets/archconvnets/convnet/ut_model_full/layer-params.cfg --data-provider=general-cropped --test-freq=10 --conserve-mem=1 --max-filesize=99999999 --img-size=128 --save-db=1 --experiment-id=mytestrun

To load for further running:

python run_synthetic_training.py --data-path=/export/imgnet_storage_full/yamins_skdata/sythetic_batches_0 --save-path=/export/imgnet_storage_full/ --test-range=950-999 --train-range=0-50 --layer-params=/home/darren/archconvnets/archconvnets/convnet/ut_model_full/layer-params.cfg --data-provider=general-cropped --test-freq=10 --conserve-mem=1 --max-filesize=99999999 --img-size=128 --save-db=1 --load-query='{"experiment_id":"mytestrun"}'

load_query can be any json-parsable query to the mongodb.

yamins81 commented 10 years ago

@ardila @daseibert can you review this so we can merge comparatively quickly. I can go over functionality with you guys if you want.

ardila commented 10 years ago

This looks good to me. I'm just going to update examples and then merge.