q_implant_validator.py

Go to the documentation of this file.

#!/usr/bin/env python3
 
# Copyright (c) 2023 Samsung Electronics Co., Ltd. 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 h5py as h5
import numpy as np
import json
 
 
def validate(h5_path, qparam_dir, qparam_json):
    valid = True
    with open(qparam_json, "r") as qparams:
        json_load = json.load(qparams)
    with h5.File(h5_path, "r") as model:
        for node_name in model.keys():
            # not quantized node exists (reshape, pad...)
            if not json_load.get(node_name):
                continue
 
            for tensor_name in json_load[node_name]:
                np_path = f"{qparam_dir}/{json_load[node_name][tensor_name]}"
                if tensor_name == "value":
                    expected_weights = np.load(np_path)
                    h5_weights = model[node_name]["weights"][:]
                    if np.allclose(h5_weights, expected_weights, rtol=1.e-5,
                                   atol=1.e-5) == False:
                        print("Implanted weights of " + node_name + "." + tensor_name +
                              " (" + str(h5_weights) +
                              ") do not match with expected value (" +
                              str(expected_weights) + ").")
                        valid = False
 
                if tensor_name == "scale":
                    expected_scale = np.load(np_path)
                    h5_scale = model[node_name]["scale"][:]
                    if np.allclose(h5_scale, expected_scale, rtol=1.e-5,
                                   atol=1.e-5) == False:
                        print("Implanted scale of " + node_name + "." + tensor_name +
                              " (" + str(h5_scale) +
                              ") do not match with expected value (" +
                              str(expected_scale) + ").")
                        valid = False
 
                if tensor_name == "zerop":
                    expected_zerop = np.load(np_path)
                    input_zerop = model[node_name]["zero_point"][:]
                    if np.allclose(input_zerop, expected_zerop, rtol=0, atol=1) == False:
                        print("Implanted zero point of " + tensor_name + " (" +
                              str(input_zerop) + ") do not match with expected value (" +
                              str(expected_zerop) + ").")
                        valid = False
 
    return valid