MCPcopy Create free account
hub / github.com/HA-Transformer/MAT / LinearizedConvolution

Class LinearizedConvolution

fairseq/modules/linearized_convolution.py:14–87  ·  view source on GitHub ↗

An optimized version of nn.Conv1d. At training time, this module uses ConvTBC, which is an optimized version of Conv1d. At inference time, it optimizes incremental generation (i.e., one time step at a time) by replacing the convolutions with linear layers. Note that the input order

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 3

LinearizedConv1dFunction · 0.90
LinearizedConv1dFunction · 0.90
LinearizedConv1dFunction · 0.90

Calls

no outgoing calls

Tested by

no test coverage detected