How to quantify mobilenet model in per_tensor mode? #377
Description
I try to quantify the mobilenet in per_tensor mode by using the following configuration. I got the result model which just puts Relu activation into Conv. So if there is any way I can quantify the model in per_tensor mode?
python version: 3.9.12
neural-compressor version: 1.14.1
system version: Windows 11
#
# Copyright (c) 2021 Intel Corporation
#
# 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.
version: 1.0
model: # mandatory. used to specify model specific information.
name: mobilenetv2
framework: onnxrt_qlinearops # mandatory. supported values are tensorflow, pytorch, pytorch_ipex, onnxrt_integer, onnxrt_qlinear or mxnet; allow new framework backend extension.
quantization: # optional. tuning constraints on model-wise for advance user to reduce tuning space.
approach: post_training_static_quant # optional. default value is post_training_static_quant.
calibration:
dataloader:
batch_size: 1
dataset:
ImagenetRaw:
data_path: /path/to/imagenet/val
image_list: /path/to/imagenet/val.txt # download from http://dl.caffe.berkeleyvision.org/caffe_ilsvrc12.tar.gz
transform:
Rescale: {}
Resize:
size: 256
CenterCrop:
size: 224
Normalize:
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
Transpose:
perm: [2, 0, 1]
Cast:
dtype: float32
model_wise: # optional. tuning constraints on model-wise for advance user to reduce tuning space.
weight:
granularity: per_tensor
scheme: asym
dtype: int8
algorithm: minmax
activation:
granularity: per_tensor
scheme: asym
dtype: int8
algorithm: minmax
evaluation: # optional. required if user doesn't provide eval_func in lpot.Quantization.
accuracy: # optional. required if user doesn't provide eval_func in lpot.Quantization.
metric:
topk: 1 # built-in metrics are topk, map, f1, allow user to register new metric.
dataloader:
batch_size: 1
dataset:
ImagenetRaw:
data_path: /path/to/imagenet/val
image_list: /path/to/imagenet/val.txt # download from http://dl.caffe.berkeleyvision.org/caffe_ilsvrc12.tar.gz
transform:
Rescale: {}
Resize:
size: 256
CenterCrop:
size: 224
Normalize:
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
Transpose:
perm: [2, 0, 1]
Cast:
dtype: float32
performance: # optional. used to benchmark performance of passing model.
warmup: 10
iteration: 1000
configs:
cores_per_instance: 4
num_of_instance: 1
dataloader:
batch_size: 1
dataset:
ImagenetRaw:
data_path: /path/to/imagenet/val
image_list: /path/to/imagenet/val.txt # download from http://dl.caffe.berkeleyvision.org/caffe_ilsvrc12.tar.gz
transform:
Rescale: {}
Resize:
size: 256
CenterCrop:
size: 224
Normalize:
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
Transpose:
perm: [2, 0, 1]
Cast:
dtype: float32
tuning:
accuracy_criterion:
relative: 0.02 # optional. default value is relative, other value is absolute. this example allows relative accuracy loss: 1%.
exit_policy:
timeout: 0 # optional. tuning timeout (seconds). default value is 0 which means early stop. combine with max_trials field to decide when to exit.
random_seed: 9527 # optional. random seed for deterministic tuning.