Skip to content
This repository has been archived by the owner on Sep 18, 2024. It is now read-only.

fix unit test #1785

Merged
merged 64 commits into from
Nov 26, 2019
Merged
Changes from all commits
Commits
Show all changes
64 commits
Select commit Hold shift + click to select a range
400e312
fix tools/nni_gpu_tool pylint
Oct 31, 2019
d0f65c8
Merge branch 'master' of https://github.com/microsoft/nni
Oct 31, 2019
5d1a115
add instrument_layer_hook && fix quantization param calc
Nov 3, 2019
cb89e16
Merge branch 'master' of https://github.com/microsoft/nni
Nov 3, 2019
493b0f3
add QAT example
Nov 4, 2019
bd809a7
remove data
Nov 4, 2019
c41a6cc
modify framework
Cjkkkk Nov 4, 2019
1b02f6a
rm irrelevant files
Cjkkkk Nov 4, 2019
61d471e
fix pylint for QAT quantizer
Cjkkkk Nov 4, 2019
3a3f3ce
resolve conflicts
Cjkkkk Nov 4, 2019
f884f2a
API refactor
Cjkkkk Nov 4, 2019
47d639f
warning for no weight parameter
Cjkkkk Nov 4, 2019
985dc43
API refactor
Cjkkkk Nov 4, 2019
c2e3871
fix pylint for QAT_torch_quantizer.py
Cjkkkk Nov 4, 2019
71c3369
init modules_to_compress to None
Cjkkkk Nov 4, 2019
2897613
modify config
Cjkkkk Nov 5, 2019
e837314
add doc string for QAT_quantizer
Cjkkkk Nov 5, 2019
e229624
rename quant_delay to quant_start_step
Cjkkkk Nov 5, 2019
bca4b51
remove EMA
Cjkkkk Nov 5, 2019
1ddba06
add docstring to explain dequantize in dequantize method
Cjkkkk Nov 5, 2019
439cdec
fix typo
Cjkkkk Nov 5, 2019
34f3b62
fix
Cjkkkk Nov 5, 2019
8ef4c4b
change to stateless
Cjkkkk Nov 5, 2019
4a1d122
update doc
Cjkkkk Nov 5, 2019
6a29346
fix
Cjkkkk Nov 6, 2019
40f9062
remove return name in docstring
Cjkkkk Nov 6, 2019
81bb549
fix test
Cjkkkk Nov 6, 2019
2c09777
update doc
Cjkkkk Nov 6, 2019
ce7bea6
fix docstring
Cjkkkk Nov 7, 2019
8102990
fix compressor doc
Cjkkkk Nov 11, 2019
d20c820
fix wrong return statement & restore doc
Cjkkkk Nov 11, 2019
5188451
fix
Cjkkkk Nov 12, 2019
4d5a65b
fix
Cjkkkk Nov 12, 2019
9cfd04b
fix name convention
Cjkkkk Nov 13, 2019
6554b3a
fix quant_bits
Cjkkkk Nov 14, 2019
ecef611
fix quant_bits
Cjkkkk Nov 14, 2019
4b33383
fix pylint
Cjkkkk Nov 15, 2019
8b48745
fix shift error
Cjkkkk Nov 15, 2019
8026520
fix pylint
Cjkkkk Nov 18, 2019
f5d3191
fix
Cjkkkk Nov 18, 2019
a74fe20
fix
Cjkkkk Nov 18, 2019
bef64a2
fix straight through estimator
Cjkkkk Nov 18, 2019
c5ff101
fix docs
Cjkkkk Nov 18, 2019
0a2fc51
add ema for output
Cjkkkk Nov 18, 2019
32b74e4
add EMA
Cjkkkk Nov 18, 2019
34bb9af
change update_param to stateless
Cjkkkk Nov 18, 2019
38749e2
fix docs
Cjkkkk Nov 18, 2019
ed49531
fix docs
Cjkkkk Nov 18, 2019
118a1f7
Merge branch 'master' of https://github.com/microsoft/nni
Cjkkkk Nov 19, 2019
0fc4d6d
fix docstring & add quant bits length check
Cjkkkk Nov 20, 2019
f833411
Merge branch 'master' of https://github.com/microsoft/nni
Cjkkkk Nov 21, 2019
8f9d0f3
unit test for QAT
Cjkkkk Nov 21, 2019
dcf4483
unit test for QAT
Cjkkkk Nov 21, 2019
2825928
unit test for QAT
Cjkkkk Nov 21, 2019
25daf32
add modules_detection test for quantization framework
Cjkkkk Nov 21, 2019
9ecac14
add modules_detection test for quantization framework
Cjkkkk Nov 21, 2019
3437dcb
revert test
Cjkkkk Nov 21, 2019
ce2b194
fix doc string
Cjkkkk Nov 25, 2019
38109e3
fix
Cjkkkk Nov 25, 2019
b2c9e99
Merge branch 'master' of https://github.com/microsoft/nni
Cjkkkk Nov 25, 2019
00f0ff0
add unit_test
Cjkkkk Nov 25, 2019
fe29b23
Merge branch 'unit_test' of https://github.com/Cjkkkk/nni into unit_test
Cjkkkk Nov 25, 2019
c4edf5e
fix round bugs
Cjkkkk Nov 26, 2019
cc5ece3
fix
Cjkkkk Nov 26, 2019
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
2 changes: 1 addition & 1 deletion src/sdk/pynni/tests/test_compressor.py
Original file line number Diff line number Diff line change
Expand Up @@ -257,7 +257,7 @@ def test_torch_QAT_quantizer(self):
weight = torch.tensor([[-1, 2], [3, 5]]).float()
quantize_weight = quantizer.quantize_weight(weight, config_list[0], model.conv2)
assert math.isclose(model.conv2.scale, 6 / 255, abs_tol=eps)
assert model.conv2.zero_point == 42
assert model.conv2.zero_point in (42, 43)

# test ema
x = torch.tensor([[-0.2, 0], [0.1, 0.2]])
Expand Down