mirror of
https://github.com/saymrwulf/pytorch.git
synced 2026-05-14 20:57:59 +00:00
27c7158166
Summary: There is a module called `2to3` which you can target for future specifically to remove these, the directory of `caffe2` has the most redundant imports: ```2to3 -f future -w caffe2``` Pull Request resolved: https://github.com/pytorch/pytorch/pull/45033 Reviewed By: seemethere Differential Revision: D23808648 Pulled By: bugra fbshipit-source-id: 38971900f0fe43ab44a9168e57f2307580d36a38
284 lines
9.5 KiB
Python
284 lines
9.5 KiB
Python
|
|
|
|
import collections
|
|
|
|
import caffe2.python.hypothesis_test_util as hu
|
|
import hypothesis.strategies as st
|
|
import numpy as np
|
|
from caffe2.python import core, dyndep, workspace
|
|
from caffe2.quantization.server.dnnlowp_test_utils import check_quantized_results_close
|
|
from hypothesis import given
|
|
|
|
|
|
dyndep.InitOpsLibrary("//caffe2/caffe2/quantization/server:dnnlowp_ops")
|
|
workspace.GlobalInit(["caffe2", "--caffe2_omp_num_threads=11"])
|
|
|
|
|
|
class DNNLowPOpSumOpTest(hu.HypothesisTestCase):
|
|
# correctness test with no quantization error in inputs
|
|
@given(
|
|
N=st.integers(32, 256),
|
|
M=st.integers(1, 3),
|
|
is_empty=st.booleans(),
|
|
**hu.gcs_cpu_only
|
|
)
|
|
def test_dnnlowp_elementwise_sum_int(self, N, M, is_empty, gc, dc):
|
|
if is_empty:
|
|
N = 0
|
|
# All inputs have scale 1, so exactly represented after quantization
|
|
inputs = M * [None]
|
|
X_names = M * [None]
|
|
X_q_names = M * [None]
|
|
|
|
for i in range(M):
|
|
X = np.random.randint(-128, 127, N, np.int8).astype(np.float32)
|
|
if N != 0:
|
|
X[0] = -128
|
|
X[-1] = 127
|
|
inputs[i] = X
|
|
X_names[i] = chr(ord("A") + i)
|
|
X_q_names[i] = X_names[i] + "_q"
|
|
|
|
Output = collections.namedtuple("Output", ["Y", "op_type", "engine"])
|
|
outputs = []
|
|
|
|
op_engine_list = [("Sum", ""), ("Sum", "DNNLOWP"), ("Int8Sum", "DNNLOWP")]
|
|
|
|
for op_type, engine in op_engine_list:
|
|
net = core.Net("test_net")
|
|
|
|
if engine == "DNNLOWP":
|
|
for i in range(M):
|
|
quantize = core.CreateOperator(
|
|
"Quantize",
|
|
X_names[i],
|
|
X_q_names[i],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([quantize])
|
|
|
|
sum_ = core.CreateOperator(
|
|
op_type,
|
|
X_q_names if engine == "DNNLOWP" else X_names,
|
|
["Y_q" if engine == "DNNLOWP" else "Y"],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([sum_])
|
|
|
|
if engine == "DNNLOWP":
|
|
dequantize = core.CreateOperator(
|
|
"Dequantize", ["Y_q"], ["Y"], engine=engine, device_option=gc
|
|
)
|
|
net.Proto().op.extend([dequantize])
|
|
|
|
for i in range(M):
|
|
self.ws.create_blob(X_names[i]).feed(X, device_option=gc)
|
|
|
|
self.ws.run(net)
|
|
outputs.append(
|
|
Output(Y=self.ws.blobs["Y"].fetch(), op_type=op_type, engine=engine)
|
|
)
|
|
|
|
check_quantized_results_close(outputs)
|
|
|
|
# correctness test with no quantization error in inputs
|
|
@given(N=st.integers(32, 256), M=st.integers(1, 3), **hu.gcs_cpu_only)
|
|
def test_dnnlowp_elementwise_sum_int_inplace(self, N, M, gc, dc):
|
|
# All inputs have scale 1, so exactly represented after quantization
|
|
inputs = M * [None]
|
|
X_names = M * [None]
|
|
X_q_names = M * [None]
|
|
|
|
for i in range(M):
|
|
X = np.random.randint(-128, 127, N, np.int8).astype(np.float32)
|
|
X[0] = -128
|
|
X[-1] = 127
|
|
inputs[i] = X
|
|
X_names[i] = chr(ord("A") + i)
|
|
X_q_names[i] = X_names[i] + "_q"
|
|
|
|
Output = collections.namedtuple("Output", ["Y", "op_type", "engine"])
|
|
outputs = []
|
|
|
|
op_engine_list = [("Sum", ""), ("Sum", "DNNLOWP"), ("Int8Sum", "DNNLOWP")]
|
|
|
|
for op_type, engine in op_engine_list:
|
|
net = core.Net("test_net")
|
|
|
|
if engine == "DNNLOWP":
|
|
for i in range(M):
|
|
quantize = core.CreateOperator(
|
|
"Quantize",
|
|
X_names[i],
|
|
X_q_names[i],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([quantize])
|
|
|
|
sum_ = core.CreateOperator(
|
|
op_type,
|
|
X_q_names if engine == "DNNLOWP" else X_names,
|
|
[X_q_names[0] if engine == "DNNLOWP" else X_names[0]],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([sum_])
|
|
|
|
if engine == "DNNLOWP":
|
|
dequantize = core.CreateOperator(
|
|
"Dequantize",
|
|
[X_q_names[0]],
|
|
[X_names[0]],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([dequantize])
|
|
|
|
for i in range(M):
|
|
self.ws.create_blob(X_names[i]).feed(X, device_option=gc)
|
|
|
|
self.ws.run(net)
|
|
outputs.append(
|
|
Output(
|
|
Y=self.ws.blobs[X_names[0]].fetch(), op_type=op_type, engine=engine
|
|
)
|
|
)
|
|
|
|
check_quantized_results_close(outputs)
|
|
|
|
# correctness test with no quantization error in inputs
|
|
@given(N=st.integers(32, 256), M=st.integers(1, 3), **hu.gcs_cpu_only)
|
|
def test_dnnlowp_elementwise_sum_relu_int(self, N, M, gc, dc):
|
|
# All inputs have scale 1, so exactly represented after quantization
|
|
inputs = M * [None]
|
|
X_names = M * [None]
|
|
X_q_names = M * [None]
|
|
|
|
for i in range(M):
|
|
X = np.random.randint(-128, 127, N, np.int8).astype(np.float32)
|
|
X[0] = -128
|
|
X[-1] = 127
|
|
inputs[i] = X
|
|
X_names[i] = chr(ord("A") + i)
|
|
X_q_names[i] = X_names[i] + "_q"
|
|
|
|
Output = collections.namedtuple("Output", ["Y", "op_type", "engine"])
|
|
outputs = []
|
|
|
|
op_engine_list = [
|
|
("SumRelu", ""),
|
|
("SumRelu", "DNNLOWP"),
|
|
("Int8SumRelu", "DNNLOWP"),
|
|
]
|
|
|
|
for op_type, engine in op_engine_list:
|
|
net = core.Net("test_net")
|
|
|
|
if engine == "DNNLOWP":
|
|
for i in range(M):
|
|
quantize = core.CreateOperator(
|
|
"Quantize",
|
|
X_names[i],
|
|
X_q_names[i],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([quantize])
|
|
|
|
sum_relu = core.CreateOperator(
|
|
op_type,
|
|
X_q_names if engine == "DNNLOWP" else X_names,
|
|
["Y_q" if engine == "DNNLOWP" else "Y"],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([sum_relu])
|
|
|
|
if engine == "DNNLOWP":
|
|
dequantize = core.CreateOperator(
|
|
"Dequantize", ["Y_q"], ["Y"], engine=engine, device_option=gc
|
|
)
|
|
net.Proto().op.extend([dequantize])
|
|
|
|
for i in range(M):
|
|
self.ws.create_blob(X_names[i]).feed(X, device_option=gc)
|
|
|
|
self.ws.run(net)
|
|
outputs.append(
|
|
Output(Y=self.ws.blobs["Y"].fetch(), op_type=op_type, engine=engine)
|
|
)
|
|
|
|
check_quantized_results_close(outputs)
|
|
|
|
# correctness test with no quantization error in inputs
|
|
@given(N=st.integers(32, 256), M=st.integers(1, 3), **hu.gcs_cpu_only)
|
|
def test_dnnlowp_elementwise_sum_relu_int_inplace(self, N, M, gc, dc):
|
|
# All inputs have scale 1, so exactly represented after quantization
|
|
inputs = M * [None]
|
|
X_names = M * [None]
|
|
X_q_names = M * [None]
|
|
|
|
for i in range(M):
|
|
X = np.random.randint(-128, 127, N, np.int8).astype(np.float32)
|
|
X[0] = -128
|
|
X[-1] = 127
|
|
inputs[i] = X
|
|
X_names[i] = chr(ord("A") + i)
|
|
X_q_names[i] = X_names[i] + "_q"
|
|
|
|
Output = collections.namedtuple("Output", ["Y", "op_type", "engine"])
|
|
outputs = []
|
|
|
|
op_engine_list = [
|
|
("SumRelu", ""),
|
|
("SumRelu", "DNNLOWP"),
|
|
("Int8SumRelu", "DNNLOWP"),
|
|
]
|
|
|
|
for op_type, engine in op_engine_list:
|
|
net = core.Net("test_net")
|
|
|
|
if engine == "DNNLOWP":
|
|
for i in range(M):
|
|
quantize = core.CreateOperator(
|
|
"Quantize",
|
|
X_names[i],
|
|
X_q_names[i],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([quantize])
|
|
|
|
sum_relu = core.CreateOperator(
|
|
op_type,
|
|
X_q_names if engine == "DNNLOWP" else X_names,
|
|
[X_q_names[0] if engine == "DNNLOWP" else X_names[0]],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([sum_relu])
|
|
|
|
if engine == "DNNLOWP":
|
|
dequantize = core.CreateOperator(
|
|
"Dequantize",
|
|
[X_q_names[0]],
|
|
[X_names[0]],
|
|
engine=engine,
|
|
device_option=gc,
|
|
)
|
|
net.Proto().op.extend([dequantize])
|
|
|
|
for i in range(M):
|
|
self.ws.create_blob(X_names[i]).feed(X, device_option=gc)
|
|
|
|
self.ws.run(net)
|
|
outputs.append(
|
|
Output(
|
|
Y=self.ws.blobs[X_names[0]].fetch(), op_type=op_type, engine=engine
|
|
)
|
|
)
|
|
|
|
check_quantized_results_close(outputs)
|