search

microsoft / constrained-graph-variational-autoencoder

Sample code for Constrained Graph Variational Autoencoders
MIT License
229 stars 57 forks source link

Blas xGEMMBatched launch failed #3

Open jooewood opened 2 years ago

jooewood commented 2 years ago

Hi, after I run python CGVAE.py --dataset qm9, the program failed, the error is 

InternalError (see above for traceback): Blas xGEMMBatched launch failed : a.shape=[64,10,10], b.shape=[64,10,100], m=10, n=100, k=10, batch_size=64
         [[Node: graph_model/gru_scope_encoder0_1/MatMul_1 = BatchMatMul[T=DT_FLOAT, adj_x=false, adj_y=false, _device="/job:localhost/replica:0/task:0/device:GPU:0"](graph_model/gru_scope_encoder11_1/strided_slice_2, graph_model/gru_scope_encoder0_1/Reshape)]]
         [[Node: clip_by_norm_197/_195 = _Recv[client_terminated=false, recv_device="/job:localhost/replica:0/task:0/device:CPU:0", send_device="/job:localhost/replica:0/task:0/device:GPU:0", send_device_incarnation=1, tensor_name="edge_24896_clip_by_norm_197", tensor_type=DT_FLOAT, _device="/job:localhost/replica:0/task:0/device:CPU:0"]()]]

Any suggestions? Thanks!

leuchine commented 2 years ago

Hi, It seems like a compatibility issue to me. Please note that we used Tensorflow 1.3 at that time, which is a rather old version of Tensorflow now. Could you please use conda to install Tensorflow 1.3 to have a look? Thanks!

Best Regards, Qi