This page analyzes the hyperparameter tuning results for the FlexMF scorer in implicit-feedback mode with WARP loss.
Parameter Search Space
<frozen abc>:106: UserWarning: component class Placeholder does not define a config attribute type
embedding_size_exp |
Integer |
Uniform |
3 ≤ \(x\) ≤ 10 |
5 |
regularization |
Float |
LogUniform |
0.0001 ≤ \(x\) ≤ 10 |
0.126 |
learning_rate |
Float |
LogUniform |
0.001 ≤ \(x\) ≤ 0.1 |
0.00297 |
reg_method |
Categorical |
Uniform |
L2, AdamW |
AdamW |
item_bias |
Categorical |
Uniform |
True, False |
True |
Final Result
Searching selected the following configuration:
{
'embedding_size_exp': 5,
'regularization': 0.12635528554790926,
'learning_rate': 0.002968879172151884,
'reg_method': 'AdamW',
'item_bias': True,
'epochs': 16
}
With these metrics:
{
'RBP': 0.24155438907443896,
'DCG': 11.760526561110588,
'NDCG': 0.47866059157942203,
'RecipRank': 0.43078719832874235,
'Hit10': 0.674901185770751,
'max_epochs': 50,
'epoch_train_s': 13.15165871893987,
'epoch_measure_s': 7.955529009923339,
'done': True,
'training_iteration': 16,
'trial_id': '4a1d559e',
'date': '2025-09-29_15-21-54',
'timestamp': 1759173714,
'time_this_iter_s': 21.11275315284729,
'time_total_s': 343.76303362846375,
'pid': 3534603,
'hostname': 'CCI-ws21',
'node_ip': '10.248.127.152',
'config': {
'embedding_size_exp': 5,
'regularization': 0.12635528554790926,
'learning_rate': 0.002968879172151884,
'reg_method': 'AdamW',
'item_bias': True,
'epochs': 16
},
'time_since_restore': 60.31975746154785,
'iterations_since_restore': 3
}
Parameter Analysis
Embedding Size
The embedding size is the hyperparameter that most affects the model’s fundamental logic, so let’s look at performance as a fufnction of it:
Iteration Completion
How many iterations, on average, did we complete?
How did the metric progress in the best result?
How did the metric progress in the longest results?