donghande / autodebias Goto Github PK

This is the official pytorch implementation of AutoDebias, an automatic debiasing method for recommendation.

License: MIT License

Python 96.64% C++ 3.36%

autodebias's Introduction

AutoDebias

This is the official pytorch implementation of AutoDebias, a debiasing method for recommendation system. AutoDebias is proposed in the paper:

AutoDebias: Learning to Debias for Recommendation

by Jiawei Chen, Hande Dong, Yang Qiu, Xiangnan He, Xin Xin, Liang Chen, Guli Lin and Keping Yang

Published at SIGIR 2021.

Introduction

AutoDebias is an automatic debiasing method for recommendation system based on meta learning, exploiting a small amout of uniform data to learn de-biasing parameters and using these parameters to guide the learning of the recommendation model.

Environment Requirement

The code runs well under python 3.8.5. The required packages are as follows:

pytorch == 1.4.0
numpy == 1.19.1
scipy == 1.5.2
pandas == 1.1.3
cppimport == 20.8.4.2

Datasets

We use two public datasets (Yahoo!R3 and Coat) and a synthetic dataset (Simulation).

user.txt: biased data collected by normal policy of recommendation platform. For Yahoo!R3 and Coat, each line is user ID, item ID, rating of the user to the item. For Simulation, each line is user ID, item ID, position of the item, binary rating of the user to the item.
random.txt: unbiased data collected by stochastic policy where items are assigned to users randomly. Each line in the file is user ID, item ID, rating of the user to the item.

Run the Code

Explicit feedback

For dataset Yahoo!R3:

python train_explicit.py --dataset yahooR3

For dataset Coat:

python train_explicit.py --dataset coat

Implicit feedback

For dataset Yahoo!R3:

python train_implicit.py --dataset yahooR3

For dataset Coat:

python train_implicit.py --dataset coat

Feedback on list recommendation

For dataset Simulation:

python train_list.py --dataset simulation

Contact

Please contact [email protected] or [email protected] if you have any questions about the code and paper.

autodebias's People

Contributors

Stargazers

Watchers

autodebias's Issues

ModuleNotFoundError: No module named 'utils.ex'

Hi. Thank you for sharing the code.

I download your code and directly run it. But it shows:
ModuleNotFoundError: No module named 'utils.ex'

If I delete import utils.ex as ex and add exp= cppimport.imp_from_filepath('utils/ex.cpp'), it still shows the following error:
#include <forward_list>
^~~~~~~~~~~~~~
1 error generated.
error: command 'gcc' failed with exit status 1

Can you help me how to run your code if possible. Thank you very much!

High Performance of Biased Methods

Hello, I am very happy to see the release of Autodebias.

When I seek to reproduce the MF_biased and MF_combine approaches, they exhibits much better performance than the IPS/DR/CausE approaches. The MF_combine approach, in particular, reaches an AUC of 0.735-0.737, which is competitive to AutoDebias.

I'm very curious as to why the IPS/DR/CausE approach would fail in this implementation, and why the biased/combine approaches would perform better than debiased approaches. This issue is important to me because this strange result makes me question the validity of the fundamental causal approaches (IPS/DR) in practice.

Thanks!

No module named 'utils.ex'

生成了so文件依然显示无法找到utils.ex 请大佬指点一下,谢谢!

Replace the dataset

If I want to change the data set, how should I operate it?

The setting of threshold(=4) in the code conflicts with the paper(=3)

IN the paper the threshold said to be 3 but in fact it's 4 when run python train_implicit.py --dataset yahooR3

see threshold value assignment in the code

For threshold = 4, the performance is:
(pytorch) C:\Users\Administrator\Desktop\auto\AutoDebias>python train_implicit.py --dataset yahooR3
Epoch: 0 / 500, Validation: MSE:0.441 NLL:-0.527 AUC:0.550
Epoch: 1 / 500, Validation: MSE:0.305 NLL:-0.452 AUC:0.617
Epoch: 2 / 500, Validation: MSE:0.292 NLL:-0.429 AUC:0.654
Epoch: 3 / 500, Validation: MSE:0.292 NLL:-0.420 AUC:0.673
Epoch: 4 / 500, Validation: MSE:0.293 NLL:-0.417 AUC:0.685
Epoch: 5 / 500, Validation: MSE:0.293 NLL:-0.415 AUC:0.692
Epoch: 6 / 500, Validation: MSE:0.293 NLL:-0.415 AUC:0.693
Epoch: 7 / 500, Validation: MSE:0.293 NLL:-0.415 AUC:0.697
Epoch: 8 / 500, Validation: MSE:0.292 NLL:-0.416 AUC:0.697
Epoch: 9 / 500, Validation: MSE:0.291 NLL:-0.417 AUC:0.697
Epoch: 10 / 500, Validation: MSE:0.291 NLL:-0.419 AUC:0.700
Epoch: 11 / 500, Validation: MSE:0.290 NLL:-0.420 AUC:0.698
Epoch: 12 / 500, Validation: MSE:0.289 NLL:-0.422 AUC:0.699
Epoch: 13 / 500, Validation: MSE:0.289 NLL:-0.423 AUC:0.697
Epoch: 14 / 500, Validation: MSE:0.289 NLL:-0.424 AUC:0.698
Epoch: 15 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.698
Epoch: 16 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 17 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.699
Epoch: 18 / 500, Validation: MSE:0.289 NLL:-0.426 AUC:0.698
Epoch: 19 / 500, Validation: MSE:0.289 NLL:-0.427 AUC:0.699
Epoch: 20 / 500, Validation: MSE:0.288 NLL:-0.427 AUC:0.704
Epoch: 21 / 500, Validation: MSE:0.288 NLL:-0.427 AUC:0.700
Epoch: 22 / 500, Validation: MSE:0.288 NLL:-0.427 AUC:0.701
Epoch: 23 / 500, Validation: MSE:0.288 NLL:-0.427 AUC:0.702
Epoch: 24 / 500, Validation: MSE:0.288 NLL:-0.427 AUC:0.701
Epoch: 25 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 26 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 27 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 28 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 29 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 30 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 31 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 32 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.699
Epoch: 33 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 34 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 35 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 36 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.703
Epoch: 37 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 38 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 39 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 40 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 41 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 42 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 43 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 44 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.703
Epoch: 45 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 46 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 47 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 48 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.699
Epoch: 49 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 50 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.700
Epoch: 51 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.703
Epoch: 52 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 53 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.703
Epoch: 54 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 55 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 56 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 57 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.702
Epoch: 58 / 500, Validation: MSE:0.288 NLL:-0.426 AUC:0.701
Epoch: 59 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 60 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.699
Epoch: 61 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 62 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 63 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 64 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 65 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 66 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 67 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 68 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 69 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 70 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.700
Epoch: 71 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.700
Epoch: 72 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 73 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 74 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.698
Epoch: 75 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 76 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 77 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 78 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 79 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 80 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 81 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 82 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 83 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 84 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.702
Epoch: 85 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.699
Epoch: 86 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 87 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 88 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 89 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 90 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 91 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.701
Epoch: 92 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 93 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 94 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 95 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 96 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.701
Epoch: 97 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.704
Epoch: 98 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 99 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.703
Epoch: 100 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.704
Epoch: 101 / 500, Validation: MSE:0.288 NLL:-0.425 AUC:0.703
Epoch: 102 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.703
Epoch: 103 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.704
Epoch: 104 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.705
Epoch: 105 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.706
Epoch: 106 / 500, Validation: MSE:0.287 NLL:-0.425 AUC:0.704
Epoch: 107 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.706
Epoch: 108 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.706
Epoch: 109 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.707
Epoch: 110 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.709
Epoch: 111 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.708
Epoch: 112 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.706
Epoch: 113 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.708
Epoch: 114 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.708
Epoch: 115 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.711
Epoch: 116 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.709
Epoch: 117 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.710
Epoch: 118 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.713
Epoch: 119 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.710
Epoch: 120 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.713
Epoch: 121 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.713
Epoch: 122 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.714
Epoch: 123 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.712
Epoch: 124 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.714
Epoch: 125 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.714
Epoch: 126 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.715
Epoch: 127 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.717
Epoch: 128 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.718
Epoch: 129 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.717
Epoch: 130 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.715
Epoch: 131 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.717
Epoch: 132 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.716
Epoch: 133 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.716
Epoch: 134 / 500, Validation: MSE:0.287 NLL:-0.424 AUC:0.712
Epoch: 135 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.714
Epoch: 136 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.714
Epoch: 137 / 500, Validation: MSE:0.286 NLL:-0.424 AUC:0.715
Epoch: 138 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.715
Epoch: 139 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 140 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.712
Epoch: 141 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.715
Epoch: 142 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 143 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 144 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 145 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.712
Epoch: 146 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.711
Epoch: 147 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.710
Epoch: 148 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.711
Epoch: 149 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.712
Epoch: 150 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.714
Epoch: 151 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.712
Epoch: 152 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.711
Epoch: 153 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.714
Epoch: 154 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.714
Epoch: 155 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 156 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.712
Epoch: 157 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.714
Epoch: 158 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.714
Epoch: 159 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.711
Epoch: 160 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.713
Epoch: 161 / 500, Validation: MSE:0.286 NLL:-0.423 AUC:0.711
Epoch: 162 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.712
Epoch: 163 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.712
Epoch: 164 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.708
Epoch: 165 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.711
Epoch: 166 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.711
Epoch: 167 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.710
Epoch: 168 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.710
Epoch: 169 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.714
Epoch: 170 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.712
Epoch: 171 / 500, Validation: MSE:0.285 NLL:-0.423 AUC:0.711
Epoch: 172 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.711
Epoch: 173 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.712
Epoch: 174 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.710
Epoch: 175 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.712
Epoch: 176 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.711
Epoch: 177 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.712
Epoch: 178 / 500, Validation: MSE:0.285 NLL:-0.422 AUC:0.710
Epoch: 179 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.714
Epoch: 180 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.711
Epoch: 181 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.712
Epoch: 182 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.711
Epoch: 183 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.711
Epoch: 184 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.709
Epoch: 185 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.710
Epoch: 186 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.714
Epoch: 187 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.711
Epoch: 188 / 500, Validation: MSE:0.284 NLL:-0.422 AUC:0.716
Loading 128th epoch

The performance of validation set: MSE:0.287 NLL:-0.424 AUC:0.718
The performance of testing set: MSE:0.307 NLL:-0.429 AUC:0.722 Precision:0.263 Recall:0.743 NDCG:0.587

So the performance is:
NLL -0.429 AUC 0.722 NDCG@5 0.587 (for threshold of 3) , roughly the same with that in the paper: -0.419 0.741 0.645
but thus less competitive compared with other methods.

But when I change this line and set threshold to 3, the performance is:

(pytorch) C:\Users\Administrator\Desktop\auto\AutoDebias>python train_implicit.py --dataset yahooR3
Epoch: 0 / 500, Validation: MSE:0.711 NLL:-0.596 AUC:0.529
Epoch: 1 / 500, Validation: MSE:0.712 NLL:-0.565 AUC:0.569
Epoch: 2 / 500, Validation: MSE:0.725 NLL:-0.559 AUC:0.584
Epoch: 3 / 500, Validation: MSE:0.717 NLL:-0.561 AUC:0.604
Epoch: 4 / 500, Validation: MSE:0.702 NLL:-0.568 AUC:0.613
Epoch: 5 / 500, Validation: MSE:0.693 NLL:-0.577 AUC:0.622
Epoch: 6 / 500, Validation: MSE:0.693 NLL:-0.583 AUC:0.627
Epoch: 7 / 500, Validation: MSE:0.692 NLL:-0.580 AUC:0.635
Epoch: 8 / 500, Validation: MSE:0.691 NLL:-0.575 AUC:0.639
Epoch: 9 / 500, Validation: MSE:0.692 NLL:-0.573 AUC:0.639
Epoch: 10 / 500, Validation: MSE:0.691 NLL:-0.574 AUC:0.643
Epoch: 11 / 500, Validation: MSE:0.690 NLL:-0.575 AUC:0.644
Epoch: 12 / 500, Validation: MSE:0.689 NLL:-0.576 AUC:0.648
Epoch: 13 / 500, Validation: MSE:0.690 NLL:-0.576 AUC:0.645
Epoch: 14 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.648
Epoch: 15 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.649
Epoch: 16 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.649
Epoch: 17 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.647
Epoch: 18 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.651
Epoch: 19 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.651
Epoch: 20 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.649
Epoch: 21 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.647
Epoch: 22 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.651
Epoch: 23 / 500, Validation: MSE:0.689 NLL:-0.575 AUC:0.649
Epoch: 24 / 500, Validation: MSE:0.688 NLL:-0.575 AUC:0.650
Epoch: 25 / 500, Validation: MSE:0.689 NLL:-0.574 AUC:0.650
Epoch: 26 / 500, Validation: MSE:0.688 NLL:-0.575 AUC:0.649
Epoch: 27 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 28 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 29 / 500, Validation: MSE:0.688 NLL:-0.575 AUC:0.649
Epoch: 30 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 31 / 500, Validation: MSE:0.689 NLL:-0.574 AUC:0.649
Epoch: 32 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 33 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 34 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.649
Epoch: 35 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 36 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.649
Epoch: 37 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 38 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 39 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 40 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.652
Epoch: 41 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 42 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 43 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 44 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.652
Epoch: 45 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 46 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.652
Epoch: 47 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 48 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 49 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 50 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.650
Epoch: 51 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.652
Epoch: 52 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.651
Epoch: 53 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.652
Epoch: 54 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 55 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.649
Epoch: 56 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 57 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 58 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 59 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 60 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 61 / 500, Validation: MSE:0.688 NLL:-0.574 AUC:0.649
Epoch: 62 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 63 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 64 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 65 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.652
Epoch: 66 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.653
Epoch: 67 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 68 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.652
Epoch: 69 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 70 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 71 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.652
Epoch: 72 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 73 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 74 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 75 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 76 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.650
Epoch: 77 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.650
Epoch: 78 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 79 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.649
Epoch: 80 / 500, Validation: MSE:0.688 NLL:-0.573 AUC:0.651
Epoch: 81 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.651
Epoch: 82 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.649
Epoch: 83 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.651
Epoch: 84 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.652
Epoch: 85 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.649
Epoch: 86 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.650
Epoch: 87 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.649
Epoch: 88 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.650
Epoch: 89 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.651
Epoch: 90 / 500, Validation: MSE:0.687 NLL:-0.573 AUC:0.652
Epoch: 91 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.650
Epoch: 92 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.653
Epoch: 93 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.652
Epoch: 94 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.652
Epoch: 95 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.653
Epoch: 96 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.652
Epoch: 97 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.653
Epoch: 98 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.653
Epoch: 99 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.653
Epoch: 100 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.652
Epoch: 101 / 500, Validation: MSE:0.687 NLL:-0.572 AUC:0.652
Epoch: 102 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.653
Epoch: 103 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.652
Epoch: 104 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.656
Epoch: 105 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.654
Epoch: 106 / 500, Validation: MSE:0.686 NLL:-0.572 AUC:0.654
Epoch: 107 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.655
Epoch: 108 / 500, Validation: MSE:0.686 NLL:-0.571 AUC:0.654
Epoch: 109 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.655
Epoch: 110 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.655
Epoch: 111 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.655
Epoch: 112 / 500, Validation: MSE:0.685 NLL:-0.571 AUC:0.656
Epoch: 113 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.656
Epoch: 114 / 500, Validation: MSE:0.685 NLL:-0.572 AUC:0.656
Epoch: 115 / 500, Validation: MSE:0.685 NLL:-0.571 AUC:0.658
Epoch: 116 / 500, Validation: MSE:0.684 NLL:-0.572 AUC:0.657
Epoch: 117 / 500, Validation: MSE:0.685 NLL:-0.571 AUC:0.658
Epoch: 118 / 500, Validation: MSE:0.684 NLL:-0.571 AUC:0.658
Epoch: 119 / 500, Validation: MSE:0.684 NLL:-0.571 AUC:0.660
Epoch: 120 / 500, Validation: MSE:0.684 NLL:-0.571 AUC:0.658
Epoch: 121 / 500, Validation: MSE:0.684 NLL:-0.571 AUC:0.661
Epoch: 122 / 500, Validation: MSE:0.684 NLL:-0.571 AUC:0.660
Epoch: 123 / 500, Validation: MSE:0.683 NLL:-0.571 AUC:0.661
Epoch: 124 / 500, Validation: MSE:0.683 NLL:-0.571 AUC:0.662
Epoch: 125 / 500, Validation: MSE:0.683 NLL:-0.571 AUC:0.661
Epoch: 126 / 500, Validation: MSE:0.684 NLL:-0.569 AUC:0.663
Epoch: 127 / 500, Validation: MSE:0.682 NLL:-0.571 AUC:0.664
Epoch: 128 / 500, Validation: MSE:0.683 NLL:-0.570 AUC:0.665
Epoch: 129 / 500, Validation: MSE:0.683 NLL:-0.570 AUC:0.666
Epoch: 130 / 500, Validation: MSE:0.682 NLL:-0.571 AUC:0.665
Epoch: 131 / 500, Validation: MSE:0.682 NLL:-0.570 AUC:0.668
Epoch: 132 / 500, Validation: MSE:0.682 NLL:-0.570 AUC:0.666
Epoch: 133 / 500, Validation: MSE:0.682 NLL:-0.570 AUC:0.668
Epoch: 134 / 500, Validation: MSE:0.681 NLL:-0.571 AUC:0.668
Epoch: 135 / 500, Validation: MSE:0.682 NLL:-0.569 AUC:0.669
Epoch: 136 / 500, Validation: MSE:0.681 NLL:-0.570 AUC:0.669
Epoch: 137 / 500, Validation: MSE:0.681 NLL:-0.570 AUC:0.668
Epoch: 138 / 500, Validation: MSE:0.681 NLL:-0.570 AUC:0.669
Epoch: 139 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.670
Epoch: 140 / 500, Validation: MSE:0.681 NLL:-0.570 AUC:0.669
Epoch: 141 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.671
Epoch: 142 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.670
Epoch: 143 / 500, Validation: MSE:0.680 NLL:-0.570 AUC:0.671
Epoch: 144 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.671
Epoch: 145 / 500, Validation: MSE:0.680 NLL:-0.570 AUC:0.670
Epoch: 146 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.669
Epoch: 147 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.670
Epoch: 148 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.669
Epoch: 149 / 500, Validation: MSE:0.680 NLL:-0.570 AUC:0.671
Epoch: 150 / 500, Validation: MSE:0.681 NLL:-0.569 AUC:0.672
Epoch: 151 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.670
Epoch: 152 / 500, Validation: MSE:0.680 NLL:-0.570 AUC:0.671
Epoch: 153 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.672
Epoch: 154 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.674
Epoch: 155 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.673
Epoch: 156 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.674
Epoch: 157 / 500, Validation: MSE:0.680 NLL:-0.568 AUC:0.675
Epoch: 158 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.674
Epoch: 159 / 500, Validation: MSE:0.680 NLL:-0.569 AUC:0.673
Epoch: 160 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.673
Epoch: 161 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.675
Epoch: 162 / 500, Validation: MSE:0.679 NLL:-0.568 AUC:0.674
Epoch: 163 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.676
Epoch: 164 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.675
Epoch: 165 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.675
Epoch: 166 / 500, Validation: MSE:0.679 NLL:-0.568 AUC:0.675
Epoch: 167 / 500, Validation: MSE:0.679 NLL:-0.569 AUC:0.675
Epoch: 168 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.675
Epoch: 169 / 500, Validation: MSE:0.679 NLL:-0.568 AUC:0.676
Epoch: 170 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.675
Epoch: 171 / 500, Validation: MSE:0.678 NLL:-0.569 AUC:0.675
Epoch: 172 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.676
Epoch: 173 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.677
Epoch: 174 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.675
Epoch: 175 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.676
Epoch: 176 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.676
Epoch: 177 / 500, Validation: MSE:0.677 NLL:-0.569 AUC:0.677
Epoch: 178 / 500, Validation: MSE:0.678 NLL:-0.568 AUC:0.677
Epoch: 179 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.678
Epoch: 180 / 500, Validation: MSE:0.677 NLL:-0.569 AUC:0.676
Epoch: 181 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.678
Epoch: 182 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.676
Epoch: 183 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.679
Epoch: 184 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.679
Epoch: 185 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.678
Epoch: 186 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.680
Epoch: 187 / 500, Validation: MSE:0.677 NLL:-0.568 AUC:0.680
Epoch: 188 / 500, Validation: MSE:0.677 NLL:-0.567 AUC:0.681
Epoch: 189 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.681
Epoch: 190 / 500, Validation: MSE:0.676 NLL:-0.568 AUC:0.680
Epoch: 191 / 500, Validation: MSE:0.676 NLL:-0.568 AUC:0.679
Epoch: 192 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.682
Epoch: 193 / 500, Validation: MSE:0.676 NLL:-0.568 AUC:0.682
Epoch: 194 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.684
Epoch: 195 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.685
Epoch: 196 / 500, Validation: MSE:0.676 NLL:-0.568 AUC:0.684
Epoch: 197 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.684
Epoch: 198 / 500, Validation: MSE:0.675 NLL:-0.568 AUC:0.684
Epoch: 199 / 500, Validation: MSE:0.676 NLL:-0.567 AUC:0.684
Epoch: 200 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.685
Epoch: 201 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 202 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.687
Epoch: 203 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 204 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 205 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 206 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 207 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.688
Epoch: 208 / 500, Validation: MSE:0.675 NLL:-0.567 AUC:0.686
Epoch: 209 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.687
Epoch: 210 / 500, Validation: MSE:0.675 NLL:-0.566 AUC:0.688
Epoch: 211 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.687
Epoch: 212 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.688
Epoch: 213 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.688
Epoch: 214 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.688
Epoch: 215 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.688
Epoch: 216 / 500, Validation: MSE:0.674 NLL:-0.567 AUC:0.688
Epoch: 217 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.688
Epoch: 218 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.689
Epoch: 219 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.688
Epoch: 220 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.688
Epoch: 221 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.687
Epoch: 222 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.689
Epoch: 223 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.690
Epoch: 224 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.687
Epoch: 225 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.687
Epoch: 226 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.688
Epoch: 227 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.690
Epoch: 228 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.688
Epoch: 229 / 500, Validation: MSE:0.674 NLL:-0.566 AUC:0.689
Epoch: 230 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.692
Epoch: 231 / 500, Validation: MSE:0.673 NLL:-0.567 AUC:0.688
Epoch: 232 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.689
Epoch: 233 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.689
Epoch: 234 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.689
Epoch: 235 / 500, Validation: MSE:0.673 NLL:-0.565 AUC:0.688
Epoch: 236 / 500, Validation: MSE:0.672 NLL:-0.567 AUC:0.688
Epoch: 237 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.690
Epoch: 238 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.690
Epoch: 239 / 500, Validation: MSE:0.672 NLL:-0.567 AUC:0.689
Epoch: 240 / 500, Validation: MSE:0.673 NLL:-0.565 AUC:0.689
Epoch: 241 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.689
Epoch: 242 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.688
Epoch: 243 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 244 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.688
Epoch: 245 / 500, Validation: MSE:0.673 NLL:-0.566 AUC:0.688
Epoch: 246 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.689
Epoch: 247 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 248 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.689
Epoch: 249 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 250 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 251 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.688
Epoch: 252 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.687
Epoch: 253 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.690
Epoch: 254 / 500, Validation: MSE:0.673 NLL:-0.565 AUC:0.688
Epoch: 255 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 256 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.687
Epoch: 257 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.688
Epoch: 258 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.687
Epoch: 259 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.687
Epoch: 260 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.689
Epoch: 261 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.689
Epoch: 262 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.687
Epoch: 263 / 500, Validation: MSE:0.672 NLL:-0.566 AUC:0.688
Epoch: 264 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.688
Epoch: 265 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.687
Epoch: 266 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.689
Epoch: 267 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.687
Epoch: 268 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.686
Epoch: 269 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.688
Epoch: 270 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.687
Epoch: 271 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.687
Epoch: 272 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.687
Epoch: 273 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.687
Epoch: 274 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.687
Epoch: 275 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.687
Epoch: 276 / 500, Validation: MSE:0.672 NLL:-0.565 AUC:0.688
Epoch: 277 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.687
Epoch: 278 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.688
Epoch: 279 / 500, Validation: MSE:0.672 NLL:-0.564 AUC:0.686
Epoch: 280 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.686
Epoch: 281 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.688
Epoch: 282 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.687
Epoch: 283 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.688
Epoch: 284 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.686
Epoch: 285 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.687
Epoch: 286 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.687
Epoch: 287 / 500, Validation: MSE:0.671 NLL:-0.566 AUC:0.688
Epoch: 288 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.688
Epoch: 289 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.689
Epoch: 290 / 500, Validation: MSE:0.671 NLL:-0.565 AUC:0.687
Loading 230th epoch

The performance of validation set: MSE:0.673 NLL:-0.567 AUC:0.692
The performance of testing set: MSE:0.678 NLL:-0.568 AUC:0.684 Precision:0.414 Recall:0.748 NDCG:0.676

So the performance is NLL:-0.568 AUC:0.684 NDCG:0.676
And so this is much different with that in the paper: -0.419 0.741 0.645.

implementation of baseline methods

Hi,

Could you please provide the code of baseline methods? That would help a lot.

Thanks!

metrics中auc的实现问题

hello，
def auc(vector_predict, vector_true, device = 'cuda'): pos_indexes = torch.where(vector_true == 1)[0].to(device) pos_whe=(vector_true == 1).to(device) sort_indexes = torch.argsort(vector_predict).to(device) rank=torch.zeros((len(vector_predict))).to(device) rank[sort_indexes] = torch.FloatTensor(list(range(len(vector_predict)))).to(device) rank = rank * pos_whe auc = (torch.sum(rank) - len(pos_indexes) * (len(pos_indexes) - 1) / 2) / (len(pos_indexes) * (len(vector_predict) - len(pos_indexes))) return auc.item()
vector_predict是所有用户对物品的预测，但是计算auc时应该要分用户进行计算，这样混着计算没什么用吧？

Recommend Projects

React

A declarative, efficient, and flexible JavaScript library for building user interfaces.
Vue.js

🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
Typescript

TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
TensorFlow

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
D3

Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

javascript

JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
web

Some thing interesting about web. New door for the world.
server

A server is a program made to process requests and deliver data to clients.
Machine learning

Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

We are working to build community through open source technology. NB: members must have two-factor auth.
Microsoft

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.

donghande / autodebias Goto Github PK

autodebias's Introduction

AutoDebias

Introduction

Environment Requirement

Datasets

Run the Code

Contact

autodebias's People

Contributors

Stargazers

Watchers

Forkers

autodebias's Issues

Recommend Projects

Recommend Topics

Recommend Org