hub / github.com/alibaba/nann / InlineFunctionBody

Function InlineFunctionBody

tensorflow/tensorflow/core/common_runtime/function.cc:1865–2148 · view source on GitHub ↗

Function inlining must preserve function execution semantics with regards to side-effects visibility. Tensorflow in Eager mode has an automatic control dependencies tracking mechanism, which enforces well-defined execution order of all side-effects. Any other frontend (e.g. Swift) must produce graphs following the same rules, to ensure that function inlining works correctly. IMPORTANT: Currently

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 6

RearrangeFunctionArgumentsFunction · 0.85

MakePrunedGraphCopyAndInlineMethod · 0.85

InlineFunctionCallsFunction · 0.85

ExpandInlineFunctionsFunction · 0.85

RewriteFunctionCallNodeFunction · 0.85

ConstructFunctionGraphMethod · 0.85

Calls 15

ValidateInliningFunction · 0.85

InternalFunction · 0.85

AddNoOpFunction · 0.85

AddIdentityFunction · 0.85

PropagateDebugInfoToNodeFunction · 0.85

AddPrefixAndSuffixToNodeFunction · 0.85

set_requested_deviceMethod · 0.80

op_nodesMethod · 0.80

set_deviceMethod · 0.80

signatureMethod · 0.80

IsSourceMethod · 0.80

edgesMethod · 0.80

Tested by

no test coverage detected