Inneke Mayachita

GPU sparse matrix error when training GAT with a large disjointed graph

4

I have a training set consists of 704 graphs with a total number of nodes is ~ 770k and a test set consists of 352 graphs with a total number...

Training DirectedSAGE node classification with multiple graphs and partly labeled nodes (such as PPI Dataset)

2

Hi, I would like to ask how can I train DirectedSAGE node classification with multiple graphs? The example of a similar setting is Protein-Protein Interaction Dataset.  (reference: [GAT](https://arxiv.org/pdf/1710.10903.pdf) paper)...

Hi all, Has anyone tried adding CRF layer on the last layer to improve performance? Thanks

Invalid argument: Matrix size-incompatible: In[0]: [0,1], In[1]: [70,70] [[{{node loss/crf_1_loss/MatMul_1}}]] [[metrics/crf_viterbi_accuracy/strided_slice_10/_229]]

3

Hi all, I tried to use CRF layer on top of Bi-LSTM-CNN NER model on this implementation https://github.com/kamalkraj/Named-Entity-Recognition-with-Bidirectional-LSTM-CNNs ``` wordEmbeddings = wordEmbeddings.reshape((wordEmbeddings.shape[0],300)) words_input = Input(shape=(None,),dtype='int32',name='words_input') words = Embedding(input_dim=wordEmbeddings.shape[0], output_dim=wordEmbeddings.shape[1], weights=[wordEmbeddings],...

I tried training the model several times and it gave me different results. I tried putting tf.set_random_seed and but still didn't work. Any idea? Thanks!

Recommended way to save graph data on TFRecords

1

Hey, I saw that in the examples you saved the data not in TFRecords. I am working on a project where we save the dataset on TFRecords. We have a...

Hi everyone, I am wondering if it is possible that the node attributes matrix (X) is 3-D instead of 2-D? For example: we want to use word embedding as the...