Closed venzino-han closed 1 year ago
The training time of a machine learning model can be influenced by several factors, for example, the computational resources available (such as GPU type and the number of tasks running on one GPU) and the choice of hyperparameters. For your convenience, we have included the average training time (hours) with standard deviation for all models and datasets using the NVIDIA RTX A5000 GPU as a reference.
Model | Assist2009 | Algebra2005 | Bridge2006 | NIPS34 | Statics2011 | Assist2015 | POJ |
---|---|---|---|---|---|---|---|
DKT | 0.03±0.02 | 0.10±0.07 | 0.68±0.46 | 0.10±0.13 | 0.06±0.04 | 0.30±0.27 | 0.51±0.46 |
DKT+ | 0.01±0.01 | 0.04±0.03 | 0.11±0.13 | 0.02±0.02 | 0.01±0.00 | 0.04±0.06 | 0.17±0.18 |
DKT-F | - | 0.29±0.15 | 1.33±0.60 | 0.12±0.07 | 0.06±0.04 | - | 0.28±0.21 |
KQN | 0.02±0.01 | 0.10±0.06 | 0.11±0.06 | 0.12±0.09 | 0.02±0.01 | 0.23±0.19 | 0.23±0.31 |
DKVMN | 0.20±0.11 | 0.51±0.43 | 0.35±0.23 | 0.21±0.07 | 0.06±0.03 | 1.67±1.31 | 1.79±2.09 |
ATKT | 0.08±0.05 | 0.10±0.06 | 0.20±0.07 | 0.21±0.08 | 0.03±0.01 | 0.37±0.39 | 0.42±0.42 |
GKT | 7.07±5.52 | 10.92±7.08 | 27.22±15.67 | 13.62±9.67 | 5.67±2.91 | 16.39±11.90 | 31.40±21.48 |
SAKT | 0.04±0.02 | 0.12±0.09 | 0.07±0.09 | 0.10±0.09 | 0.01±0.01 | 0.18±0.17 | 0.23±0.28 |
SAINT | 0.03±0.02 | 0.06±0.07 | 0.16±0.17 | 0.39±0.44 | 0.04±0.03 | 0.30±0.24 | 0.10±0.07 |
AKT | 0.08±0.07 | 0.45±0.50 | 0.33±0.31 | 0.17±0.11 | 0.03±0.02 | 0.39±0.23 | 0.32±0.21 |
@meta-tabchen thanks! It helps a lot.
How long does it takes for GKT training?
Sharing other datasets training time of GKT might be helpful!
Thank you!