This page analyzes the hyperparameter tuning results for biased matrix factorization with ALS.
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.
| embedding_size |
Integer |
LogUniform |
4 ≤ \(x\) ≤ 512 |
471 |
| regularization.user |
Float |
LogUniform |
1e-05 ≤ \(x\) ≤ 1 |
0.673 |
| regularization.item |
Float |
LogUniform |
1e-05 ≤ \(x\) ≤ 1 |
0.00536 |
| damping.user |
Float |
LogUniform |
1e-12 ≤ \(x\) ≤ 100 |
0.0353 |
| damping.item |
Float |
LogUniform |
1e-12 ≤ \(x\) ≤ 100 |
3.07e-08 |
Final Result
Searching selected the following configuration:
{
'embedding_size': 471,
'regularization': {'user': 0.6726624938362399, 'item': 0.00535745254610969},
'damping': {'user': 0.03532608682135444, 'item': 3.0727948378692085e-08},
'epochs': 6
}
With these metrics:
{
'RBP': 0.0016805195576097428,
'LogRBP': -2.5944123031243844,
'NDCG': 0.3078925720949283,
'RecipRank': 0.010075027000352418,
'RMSE': 0.7918119847363368,
'TrainTask': '6155c275-6616-4374-a9c1-d56059c7fd5b',
'TrainTime': None,
'TrainCPU': None,
'max_epochs': 30,
'done': True,
'training_iteration': 6,
'trial_id': 'ea472d4e',
'date': '2025-05-05_20-18-58',
'timestamp': 1746490738,
'time_this_iter_s': 297.6282503604889,
'time_total_s': 1991.604411125183,
'pid': 338683,
'hostname': 'CCI-ws21',
'node_ip': '10.248.127.152',
'config': {
'embedding_size': 471,
'regularization': {'user': 0.6726624938362399, 'item': 0.00535745254610969},
'damping': {'user': 0.03532608682135444, 'item': 3.0727948378692085e-08},
'epochs': 6
},
'time_since_restore': 1991.604411125183,
'iterations_since_restore': 6
}
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?
