FlexMF WARP

This page analyzes the hyperparameter tuning results for the FlexMF scorer in implicit-feedback mode with WARP loss.

Parameter Search Space

Parameter	Type	Distribution	Values	Selected
embedding_size_exp	Integer	Uniform	3 ≤ \(x\) ≤ 10	5
regularization	Float	LogUniform	0.0001 ≤ \(x\) ≤ 10	0.0235
learning_rate	Float	LogUniform	0.001 ≤ \(x\) ≤ 0.1	0.00304
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': 0.023507830034360473,
    'learning_rate': 0.003042904939520177,
    'reg_method': 'AdamW',
    'item_bias': True,
    'epochs': 18
}

With these metrics:

{
    'RBP': 0.2224918914617804,
    'DCG': 11.970690046578941,
    'NDCG': 0.4435227643596746,
    'RecipRank': 0.40112016702900927,
    'Hit10': 0.6238383838383839,
    'max_epochs': 50,
    'epoch_train_s': 14.259723965078592,
    'epoch_measure_s': 6.25665595009923,
    'done': True,
    'training_iteration': 18,
    'trial_id': 'c4a81999',
    'date': '2025-10-01_20-04-09',
    'timestamp': 1759363449,
    'time_this_iter_s': 20.523654222488403,
    'time_total_s': 553.1135029792786,
    'pid': 1380443,
    'hostname': 'CCI-ws21',
    'node_ip': '10.248.127.152',
    'config': {
        'embedding_size_exp': 5,
        'regularization': 0.023507830034360473,
        'learning_rate': 0.003042904939520177,
        'reg_method': 'AdamW',
        'item_bias': True,
        'epochs': 18
    },
    'time_since_restore': 553.1135029792786,
    'iterations_since_restore': 18
}

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?