Da Zheng
The Python API is convenient for many use cases. It allows more customization and is very friendly for Jupyter Notebook users.
Few-shot link prediction proposed by [this paper](https://arxiv.org/abs/1909.01515) seems a useful technique to support in DGL-KE.
The technique described in the paper "AutoNE: Hyperparameter Optimization for Massive Network Embedding" is interesting. Similar techniques should be incorporated into DGL-KE to tune hyperparameters on large knowledge graphs effectively.
It's useful to plot the entities after training their embeddings. This helps us verify the training results.
Currently, the KGE models are trained with the logistic loss. However, it's desirable to train the models with a pair-wise ranking loss because in a lot of cases negative edges...
If we store a block matrix to SSDs, we may get this error. This bug has been temporarily fixed by converting the block matrix to a normal matrix. We need...
``` > data = 0' failed. Aborted ```
Currently, we only allow local matrices in a DAG to be resized only once. Maybe it's necessary to have every virtual local matrix to make such a decision. If this...
``` X1 = -fp.log(1 + fp.exp(-X)) X2 = X - fp.log(1 + fp.exp(X)) return fp.where(X > 0, X1, X2) ``` The code above has redundant computation. log and exp can...