pylibcugraphops.operators.update_efeat_e2e_sum_fwd#

pylibcugraphops.operators.update_efeat_e2e_sum_fwd = <nanobind.nb_func object>#
Computes the forward pass for an operation which sums

edge features, and the node features of the source and destination nodes of each each in an edge-to-edge operation.

update_efeat_e2e_concat_fwd(
    output_edge_embedding: device array,
    input_edge_embedding: Optional[device array],
    input_src_node_embedding: Optional[device array],
    input_dst_node_embedding: Optional[device array],
    graph: pylibcugraphops.bipartite_csc_int[32|64], stream_id: int = 0
)
Parameters:
output_edge_embeddingdevice array type

Device array containing the output edge embeddings.

input_edge_embeddingdevice array type | None

Device array containing the input edge embeddings, ignored if passed as None

input_src_node_embeddingdevice array type | None

Device array containing the input source node embeddings, ignored if passed as None

input_dst_node_embeddingdevice array type | None

Device array containing the input destination node embeddings, ignored if passed as None

graphopaque CSC graph type

graph used for the operation.

stream_idint, default=0

CUDA stream pointer as a python int