Hyper-parameter Search Results (sequential)
Dataset informations: MovieLens-1m, Amazon-Books, Yelp2022
Notes: The hyper-parameter search range in the table is for reference only. You can adjust the search range according to the actual situation of the dataset, such as reducing the range appropriately on large data to reduce time consumption. The orange bold text in the table represents the recommended value within the hyper-parameter search range. And the symbol "\" indicates that the model appears out of memory when running on a GPU with 12G memory or the running time is too long to complete the hyper-parameter search.
| Model | MovieLens-1m | Amazon-Books | Yelp2022 |
|---|---|---|---|
| BERT4Rec | learning_rate in [0.0003, 0.0005, 0.001, 0.003] ft_ratio in [0, 0.1, 0.5] |
learning_rate in [0.0003, 0.0005, 0.001, 0.003] ft_ratio in [0, 0.1, 0.5] |
learning_rate in [0.0003, 0.0005, 0.001, 0.003] ft_ratio in [0, 0.1, 0.5] |
| Caser | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | \ |
| CORE | learning_rate in [0.0005, 0.001, 0.003] temperature in [0.05, 0.07, 0.1, 0.2] |
learning_rate in [0.0005, 0.001, 0.003] temperature in [0.05, 0.07, 0.1, 0.2] |
learning_rate in [0.005,0.001,0.0005,0.0001] |
| FDSA | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |
| FOSSIL | learning_rate in [0.0005, 0.001, 0.003] reg_weight in [1e-5, 1e-4] order_len in [1,2,3] alpha in [0.2,0.4,0.6] |
learning_rate in [0.0005, 0.001, 0.003] reg_weight in [1e-5, 1e-4] order_len in [1,2,3] alpha in [0.2,0.4,0.6] |
learning_rate in [0.01,0.001] reg_weight in [0,0.0001] alpha in [0.2,0.5] order_len in [1,2] |
| FPMC | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.01,0.005,0.001,0.0005,0.0001] |
| GCSAN | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |
| GRU4Rec | learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
| GRU4RecF | learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005] num_layers in [1,2] |
| HGN | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |
| HRM | learning_rate in [0.0005, 0.001, 0.003] high_order in [1, 2, 3] |
learning_rate in [0.0005, 0.001, 0.003] high_order in [1, 2, 3] |
learning_rate in [0.001,0.0005,0.0001] high_order in [1,2,4] |
| LightSANs | learning_rate in [0.0005, 0.001, 0.003] k_interests in [3, 5, 7, 9] |
learning_rate in [0.0005, 0.001, 0.003] k_interests in [3, 5, 7, 9] |
learning_rate in [0.005,0.001,0.0005,0.0001] |
| NARM | learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005,0.0001] num_layers in [1,2] |
learning_rate in [0.005,0.001,0.0005,0.0001] n_layers in [1,2] |
| NextItNet | learning_rate in [0.0005, 0.001, 0.003] kernel_size in [2, 3, 4] |
learning_rate in [0.0005, 0.001, 0.003] kernel_size in [2, 3, 4] |
\ |
| NPE | learning_rate in [0.0005, 0.001, 0.003] dropout_prob in [0.1, 0.3, 0.5] |
learning_rate in [0.0005, 0.001, 0.003] dropout_prob in [0.1, 0.3, 0.5] |
learning_rate in [0.001,0.0005,0.0001] dropout_prob in [0.2,0.3,0.5] |
| RepeatNet | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | \ | \ |
| S3Rec | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] pretrain_epochs in [100] |
learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] pretrain_epochs in [100] |
\ |
| SASRec | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] n_layers in [1,2] |
| SASRecF | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005] n_layers in [1,2] |
| SHAN | learning_rate in [0.0005, 0.001, 0.003] short_item_length in [1, 2, 3] |
learning_rate in [0.0005, 0.001, 0.003] short_item_length in [1, 2, 3] |
learning_rate in [0.005,0.001,0.0005] short_item_length in [1,2,4] |
| SINE | learning_rate in [0.0005, 0.001, 0.003] interest_size in [2, 3, 4] tau_ratio in [0.05, 0.07, 0.1, 0.2] |
learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |
| SRGNN | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] step in [1, 2] |
| STAMP | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |
| TransRec | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.0003, 0.0005, 0.001, 0.003, 0.005] | learning_rate in [0.005,0.001,0.0005,0.0001] |