FlexMF BPR

This page analyzes the hyperparameter tuning results for the FlexMF scorer in implicit-feedback mode with pairwise loss (Bayesian Personalized Ranking).

Parameter Search Space

/home/mde48/lenskit/lenskit-codex/.venv/lib/python3.12/site-packages/ray/tune/search/sample.py:700: RayDeprecationWarning: The `base` argument is deprecated. Please remove it as it is not actually needed in this method.

Parameter	Type	Distribution	Values	Selected
embedding_size	Integer	LogUniform	4 ≤ \(x\) ≤ 512	36
regularization	Float	LogUniform	0.0001 ≤ \(x\) ≤ 10	0.081
learning_rate	Float	LogUniform	0.001 ≤ \(x\) ≤ 0.1	0.0246
reg_method	Categorical	Uniform	L2, AdamW	AdamW
negative_count	Integer	Uniform	1 ≤ \(x\) ≤ 5	3
item_bias	Categorical	Uniform	True, False	False

Final Result

Searching selected the following configuration:

{
    'embedding_size': 36,
    'regularization': 0.08096471361858962,
    'learning_rate': 0.024616389682051425,
    'reg_method': 'AdamW',
    'negative_count': 3,
    'item_bias': False,
    'epochs': 12
}

With these metrics:

{
    'RBP': 0.09780516628896009,
    'LogRBP': 1.4694620914756937,
    'NDCG': 0.34640966733727485,
    'RecipRank': 0.3301484809465168,
    'TrainTask': 'cc735e44-b959-4543-90a5-88f0a7b9d8ca',
    'TrainTime': None,
    'TrainCPU': None,
    'max_epochs': 50,
    'done': False,
    'training_iteration': 12,
    'trial_id': '1aefad6b',
    'date': '2025-05-05_03-12-39',
    'timestamp': 1746429159,
    'time_this_iter_s': 1.6611011028289795,
    'time_total_s': 20.38316321372986,
    'pid': 3107768,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {
        'embedding_size': 36,
        'regularization': 0.08096471361858962,
        'learning_rate': 0.024616389682051425,
        'reg_method': 'AdamW',
        'negative_count': 3,
        'item_bias': False,
        'epochs': 12
    },
    'time_since_restore': 20.38316321372986,
    'iterations_since_restore': 12
}