search

gasteigerjo / ppnp

PPNP & APPNP models from "Predict then Propagate: Graph Neural Networks meet Personalized PageRank" (ICLR 2019)
https://www.daml.in.tum.de/ppnp
MIT License
317 stars 54 forks source link

when I train with tensorflow, an error about the confusion_matrix appeared #13

Closed UnBuen closed 3 years ago

UnBuen commented 3 years ago

Dear klicperajo, when I train with tensorflow, if I choose save_result when I train, it will appear name 'conf_mat' is not defined. so can you help me to deal with this problem, thank you very much. the problem is in ppnp/tensorflow/training.py line 129

gasteigerjo commented 3 years ago

It looks like you're the first person to use save_results with TensorFlow since I broke it in February 2019. 😆

Fixed in 394425a9eeffbd95c09d68d6f0e1473ac9c2fd4a.

UnBuen commented 3 years ago

Dear klicperajo, when I train with tensorflow, if I choose the fourth dataset(Ms Academic, the others is working) and train, it will appear "Cannot take a larger sample than population when 'replace=False' ". So can you help me to deal with this problem, thank you very much. the problem is in ppnp/tensorflow/training.py line 35(maybe 34). btw, if you have time, can you explain the meaning of ntrain_per_class and nstopping. idx_split_args = {'ntrain_per_class': 20, 'nstopping': 500, 'nknown': 1500, 'seed': 2413340114}😆

gasteigerjo commented 3 years ago

For MS Academic you need to use a lower number of nodes in your development/known set. See https://github.com/klicperajo/ppnp/blob/master/reproduce_results.ipynb.

ntrain_per_class is the number of nodes per class in your training set. nstopping is the number of nodes in the early stopping set. See Figure 7 in the paper for an illustration of these sets.

UnBuen commented 3 years ago

For MS Academic you need to use a lower number of nodes in your development/known set. See https://github.com/klicperajo/ppnp/blob/master/reproduce_results.ipynb.

ntrain_per_class is the number of nodes per class in your training set. nstopping is the number of nodes in the early stopping set. See Figure 7 in the paper for an illustration of these sets.

It's very helpful for me. Thank you, I'm sorry I didn't read the paper carefully.😆