Netflix
Benchmarks
TaskDatasetSOTA resultTrendResults
Netflix (test)
39.5NDCG@100
30
Netflix
5.13Recall@20
16
Netflix
77.1AUC
14
Netflix strong generalization
36.66AOA Recall@20
14
Netflix
0.803RMSE
14
Netflix 15/12/2005
7.24Recall@20
12
Netflix
9.22HR @ 5
10
Netflix coherent users
56Recall@20
10
Netflix 80/20 (test)
21.56Recall@10
10
Netflix Randomly-perturbed (test)
-58.27Log-likelihood
9
Netflix Adversarially-perturbed Ta (test)
-60.61Log-Likelihood
9
Netflix large (test)
36.2Recall@20
9
Netflix
36.2Recall@20
8
Netflix (test)
36.2Recall@20
8
Netflix Original T (test)
-57.59Log-Likelihood
7
Netflix w/o history
2.13Performance Metric
7
Netflix strong generalization
36.58R@20
7
Netflix (test)
0.803Test RMSE
7
Netflix Prize (test)
0.537Hit Rate@10
6
Netflix (test)
3.3Training Time (s)
4
Netflix
173,459Throughput (ops/s)
4
Netflix w/ history
4.141Expected User Rating
4
Netflix high reward trajectories
106.2NDCG@10
4
Netflix
0.034Avg Insert Latency (ms)
4
Netflix
75.2Recall@1
4