Skip to content

[POC][PatternLang]Remove constants from partitioned functions #5663

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
masahi merged 3 commits into from
May 26, 2020
Merged

[POC][PatternLang]Remove constants from partitioned functions #5663

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
6c7943d
remove constants from partitioned functions
May 23, 2020
5478492
remove print statements
May 23, 2020
a65e0f9
Merge remote-tracking branch 'upstream/master' into mbrookhart/remove…
May 26, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions src/relay/ir/dataflow_matcher.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -557,7 +557,7 @@ class PatternGrouper {
auto matches = node_map[node->ref_];
for (auto match : matches) {
if (fuzzy_matches.count(match) == 0 && match.as<OpNode>() == nullptr &&
match.as<FunctionNode>() == nullptr && match.as<ConstantNode>() == nullptr) {
match.as<FunctionNode>() == nullptr) {
inputs[match] = Var(
"FunctionVar_" + std::to_string(graph_number_) + "_" + std::to_string(var_number),
NullValue<Type>());
Expand All @@ -577,8 +577,8 @@ class PatternGrouper {
auto extractor = MatchExtractor(inputs);
auto body = extractor.Mutate(expr);

// Verify the pattern still holds
CHECK(DFPatternMatcher(body).Match(pattern_, body));
// Verify the pattern still holds, no longer valid if we're not embedding constants in the
// graph, keep here for future debug CHECK(DFPatternMatcher(body).Match(pattern_, body));
group.function = Function(params, body, NullValue<Type>(), Array<TypeVar>());
group.name = extractor.GetName();
// Check to make sure we aren't overlapping with another group
Expand Down
25 changes: 15 additions & 10 deletions tests/python/relay/test_dataflow_pattern.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -878,55 +878,60 @@ def nested_diamond(inp, weight):
)
assert tvm.ir.structural_equal(partitioned, reference)

def get_BN(x, var, mean, beta, gamma, eps = 1e-5):
return gamma * (x - mean)/relay.op.sqrt(var + relay.const(eps)) + beta
def get_BN(x, var, mean, beta, gamma, eps):
return gamma * (x - mean)/relay.op.sqrt(var + eps) + beta

def test_partition_batchnorm():
x = relay.var('x')
var = relay.var('var')
mean = relay.var('mean')
beta = relay.var('beta')
gamma = relay.var('gamma')
BN = get_BN(x, var, mean, beta, gamma)
eps = relay.const(1e-5)
BN = get_BN(x, var, mean, beta, gamma, eps)


xf = relay.var('xf')
varf = relay.var('varf')
meanf = relay.var('meanf')
betaf = relay.var('betaf')
gammaf = relay.var('gammaf')
epsf = relay.var('epsf')
# Put the arguments in toplogological order for the reference
f = relay.Function([gammaf, xf, meanf, varf, betaf], get_BN(xf, varf, meanf, betaf, gammaf)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")
f = relay.Function([gammaf, xf, meanf, varf, epsf, betaf], get_BN(xf, varf, meanf, betaf, gammaf, epsf)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")

partitioned = BatchnormCallback().pattern.partition(BN)
assert tvm.ir.structural_equal(partitioned, f(gamma, x, mean, var, beta))
assert tvm.ir.structural_equal(partitioned, f(gamma, x, mean, var, eps, beta))

def test_partition_double_batchnorm():
x = relay.var('x')
var = relay.var('var')
mean = relay.var('mean')
beta = relay.var('beta')
gamma = relay.var('gamma')
eps = relay.const(1e-5)

BN = gamma * (x - mean)/relay.op.sqrt(var + relay.const(1e-5)) + beta
BN2 = gamma * (BN - mean)/relay.op.sqrt(var + relay.const(1e-5)) + beta
BN = gamma * (x - mean)/relay.op.sqrt(var + eps) + beta
BN2 = gamma * (BN - mean)/relay.op.sqrt(var + eps) + beta

xf = relay.var('xf')
varf = relay.var('varf')
meanf = relay.var('meanf')
betaf = relay.var('betaf')
gammaf = relay.var('gammaf')
f1 = relay.Function([gammaf, xf, meanf, varf, betaf], get_BN(xf, varf, meanf, betaf, gammaf)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")
epsf = relay.var('epsf')
f1 = relay.Function([gammaf, xf, meanf, varf, epsf, betaf], get_BN(xf, varf, meanf, betaf, gammaf, epsf)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")
# The partitioner doesn't replace duplicates, so we use two copies of the function
xf2 = relay.var('xf2')
varf2 = relay.var('varf2')
meanf2 = relay.var('meanf2')
betaf2 = relay.var('betaf2')
gammaf2 = relay.var('gammaf2')
f2 = relay.Function([gammaf2, xf2, meanf2, varf2, betaf2], get_BN(xf2, varf2, meanf2, betaf2, gammaf2)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")
epsf2 = relay.var('epsf2')
f2 = relay.Function([gammaf2, xf2, meanf2, varf2, epsf2, betaf2], get_BN(xf2, varf2, meanf2, betaf2, gammaf2, epsf2)).with_attr("PartitionedFromPattern","subtract_multiply_add_sqrt_divide_add_")

partitioned = BatchnormCallback().pattern.partition(BN2)
reference = f2(gamma, f1(gamma, x, mean, var, beta), mean, var, beta)
reference = f2(gamma, f1(gamma, x, mean, var, eps, beta), mean, var, eps, beta)
assert tvm.ir.structural_equal(partitioned, reference)

def test_partition_check():
Expand Down