Skip to content

[ONNX] Support depth_to_space op for FQ2I #8966

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 16 commits into from
Sep 9, 2021
Merged

[ONNX] Support depth_to_space op for FQ2I #8966

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
ea9eb06
WIP support per-channel quantization
Aug 23, 2021
f33f94f
more WIP
Aug 23, 2021
ed047ec
More WIP
Aug 25, 2021
8e3dbb0
fix issue with per-channel bias_add
Aug 25, 2021
0327dab
Fix fake quantize tests (#4)
Aug 25, 2021
f12538e
Add Relu
Aug 25, 2021
0a0cd11
One more little one (#5)
Aug 25, 2021
7cf9729
Non-working Per-channel Dense
Aug 28, 2021
90951cc
Fix legalization for non spatial operators. (#6)
Aug 30, 2021
b09f1d6
fix axis normalization
Aug 30, 2021
4c6dc86
Per channel fq2i (#8)
Aug 31, 2021
607ba7c
respond to review comments
Sep 3, 2021
d3a8049
start dtos
anwang2009 Sep 3, 2021
865cf67
wip depth_to_space
anwang2009 Sep 7, 2021
4d5c150
dtos ident
anwang2009 Sep 7, 2021
9519b58
merge
anwang2009 Sep 8, 2021
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
1 change: 1 addition & 0 deletions python/tvm/relay/transform/fake_quantization_to_integer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,6 +87,7 @@ def identity(expr, type_map):
register_unary_identity("expand_dims")
register_unary_identity("nn.max_pool2d")
register_unary_identity("nn.batch_flatten")
register_unary_identity("nn.depth_to_space")


@register_fake_quantization_to_integer("nn.avg_pool2d")
Expand Down
13 changes: 13 additions & 0 deletions tests/python/relay/test_pass_fake_quantization_to_integer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -501,3 +501,16 @@ def test_fake_quantize_pad():
x_np = np.random.randint(-25, 25, size=[1, 383, 128], dtype="int8")

compare_fq_to_int(op, [x_np])


def test_fake_quantize_depth_to_space():
x = relay.var("x", shape=[1, 3, 224, 224], dtype="int8")

zero = relay.const(0)
x = relay.qnn.op.dequantize(x, relay.const(2.0), zero)
op = relay.op.nn.depth_to_space(x, 4)
op = relay.qnn.op.quantize(op, relay.const(2.0), zero)

x_np = np.random.randint(-128, 127, size=[1, 3, 224, 224], dtype="int8")

compare_fq_to_int(op, [x_np])