apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 16:48:03 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

Add paddle quantize model support for ORT, TRT and MKLDNN deploy backend (#257)

Browse Source 
* add quantize model support for trt and paddle

* fix bugs

* fix

* update paddle2onnx version

* update version

* add quantize test

Co-authored-by: Jason <jiangjiajun@baidu.com>
This commit is contained in:
yeliang2258
2022-10-09 20:00:05 +08:00
committed by GitHub
parent ff5e798b7f
commit 2a68a23baf
10 changed files with 187 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

96
tests/eval_example/test_quantize_diff.py Executable file
View File

@@ -0,0 +1,96 @@
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import fastdeploy as fd
import cv2
import os
import pickle
import numpy as np
model_url = "https://bj.bcebos.com/fastdeploy/tests/yolov6_quant.tgz"
fd.download_and_decompress(model_url, ".")
def test_quant_mkldnn():
model_path = "./yolov6_quant"
model_file = os.path.join(model_path, "model.pdmodel")
params_file = os.path.join(model_path, "model.pdiparams")
input_file = os.path.join(model_path, "input.npy")
output_file = os.path.join(model_path, "mkldnn_output.npy")
option = fd.RuntimeOption()
option.use_paddle_backend()
option.use_cpu()
option.set_model_path(model_file, params_file)
runtime = fd.Runtime(option)
input_name = runtime.get_input_info(0).name
data = np.load(input_file)
outs = runtime.infer({input_name: data})
expected = np.load(output_file)
diff = np.fabs(outs[0] - expected)
thres = 1e-05
assert diff.max() < thres, "The diff is %f, which is bigger than %f" % (
diff.max(), thres)
def test_quant_ort():
model_path = "./yolov6_quant"
model_file = os.path.join(model_path, "model.pdmodel")
params_file = os.path.join(model_path, "model.pdiparams")
input_file = os.path.join(model_path, "input.npy")
output_file = os.path.join(model_path, "ort_output.npy")
option = fd.RuntimeOption()
option.use_ort_backend()
option.use_cpu()
option.set_ort_graph_opt_level(1)
option.set_model_path(model_file, params_file)
runtime = fd.Runtime(option)
input_name = runtime.get_input_info(0).name
data = np.load(input_file)
outs = runtime.infer({input_name: data})
expected = np.load(output_file)
diff = np.fabs(outs[0] - expected)
thres = 1e-05
assert diff.max() < thres, "The diff is %f, which is bigger than %f" % (
diff.max(), thres)
def test_quant_trt():
model_path = "./yolov6_quant"
model_file = os.path.join(model_path, "model.pdmodel")
params_file = os.path.join(model_path, "model.pdiparams")
input_file = os.path.join(model_path, "input.npy")
output_file = os.path.join(model_path, "trt_output.npy")
option = fd.RuntimeOption()
option.use_trt_backend()
option.use_gpu()
option.set_model_path(model_file, params_file)
runtime = fd.Runtime(option)
input_name = runtime.get_input_info(0).name
data = np.load(input_file)
outs = runtime.infer({input_name: data})
expected = np.load(output_file)
diff = np.fabs(outs[0] - expected)
thres = 1e-05
assert diff.max() < thres, "The diff is %f, which is bigger than %f" % (
diff.max(), thres)