Hi, I'm trying to reproduce the results of your Neurips 2019 paper. After generating data for kcolor with: sh scripts/generate_kcolor.sh, I tried to run python code/train_search.py -c configs/search/kcolor/5.yaml but it failed because of missing configuration keys.
First: KeyError: 'mlp_arch', which I copied from configs/search/sr/10.yaml. Then, the same error, with gnn_iter and gnn_async keys, which I copied from sr/10.yaml. Then, with data_shuffle and eval_multi keys, which I copied from configs/search/sr/10_2.yaml.
After adding all those keys, the code seems to run (it fails on another part, please see #3 ).
My fork has the updated configs/search/kcolor/5.yaml, I can create a PR if you want.
Hi, I'm trying to reproduce the results of your Neurips 2019 paper. After generating data for kcolor with:
sh scripts/generate_kcolor.sh, I tried to runpython code/train_search.py -c configs/search/kcolor/5.yamlbut it failed because of missing configuration keys.First: KeyError: 'mlp_arch', which I copied from
configs/search/sr/10.yaml. Then, the same error, with gnn_iter and gnn_async keys, which I copied fromsr/10.yaml. Then, with data_shuffle and eval_multi keys, which I copied fromconfigs/search/sr/10_2.yaml.After adding all those keys, the code seems to run (it fails on another part, please see #3 ).
My fork has the updated
configs/search/kcolor/5.yaml, I can create a PR if you want.[EDIT: added the pointer to the other issue]