search

WayneDW / DeepLight_Deep-Lightweight-Feature-Interactions

Accelerating Inference for Recommendation Systems (WSDM'21)
https://arxiv.org/abs/2002.06987
112 stars 23 forks source link

Questions for 'data/category_emb' file #2

Closed yjhong89 closed 3 years ago

yjhong89 commented 3 years ago

Hi,

I am YJ Hong and thanks for your work.

While I am looking into this repository, it seems that there isn't any related code for generating 'data/category_emb' file, which is used in data_preprocessing.

Can you let me know what it is and how generate this file please?

Thank you.

YJHong.

WayneDW commented 3 years ago

Hi, YJHong,

As you see from the main file: emb for tiny data corresponds to *map for large data. result_dict = data_preprocess.read_data('./data/tiny_train_input.csv', './data/category_emb', criteo_num_feat_dim, feature_dim_start=0, dim=39) test_dict = data_preprocess.read_data('./data/tiny_test_input.csv', './data/category_emb', criteo_num_feat_dim, feature_dim_start=0, dim=39)

result_dict = data_preprocess.read_data('./data/large/train.csv', './data/large/criteo_feature_map', criteo_num_feat_dim, feature_dim_start=1, dim=39)

test_dict = data_preprocess.read_data('./data/large/valid.csv', './data/large/criteo_feature_map', criteo_num_feat_dim, feature_dim_start=1, dim=39)

You can move data folder to see how it is processed