delete pact in eval,infer,export_model (#1584)

slim/quantization/README.md

@@ -9,6 +9,7 @@
 
 - [检测模型的常规训练方法](https://github.com/PaddlePaddle/PaddleDetection)
 - [PaddleSlim使用文档](https://paddlepaddle.github.io/PaddleSlim/)
+- [自定义量化PACT](https://github.com/PaddlePaddle/PaddleSlim/tree/develop/demo/quant/pact_quant_aware)
 
 已发布量化模型见[压缩模型库](../README.md)
 
@@ -76,11 +77,24 @@ python slim/quantization/train.py --not_quant_pattern yolo_output \
 - **LeaningRate.base_lr:** 根据多卡的总`batch_size`调整`base_lr`，两者大小正相关，可以简单的按比例进行调整。
 - **LearningRate.schedulers.PiecewiseDecay.milestones：** 请根据batch size的变化对其调整。
 
-
 通过`python slim/quantization/train.py --help`查看可配置参数。
 通过`python ./tools/configure.py help ${option_name}`查看如何通过命令行覆盖配置文件中的参数。
 
+### PACT自定义量化
+
+```
+python slim/quantization/train.py \
+    --eval \
+    -c ./configs/yolov3_mobilenet_v3.yml \
+    -o max_iters=30000 \
+    save_dir=./output/mobilenetv3 \
+    LearningRate.base_lr=0.0001 \
+    LearningRate.schedulers="[!PiecewiseDecay {gamma: 0.1, milestones: [10000]}]" \
+    pretrain_weights=https://paddlemodels.bj.bcebos.com/object_detection/yolov3_mobilenet_v3.pdparams \
+    --use_pact=True
+```
 
+- 在量化训练时，将`--use_pact=True`，即可选择PACT自定义量化
 
 ### 训练时的模型结构
 [PaddleSlim 量化API](https://paddlepaddle.github.io/PaddleSlim/api/quantization_api/)文档中介绍了``paddleslim.quant.quant_aware``和``paddleslim.quant.convert``两个接口。
@@ -144,6 +158,7 @@ python slim/quantization/eval.py --not_quant_pattern yolo_output  -c ./configs/y
  python slim/quantization/export_model.py --not_quant_pattern yolo_output  -c ./configs/yolov3_mobilenet_v1.yml --output_dir ${save path} \
 -o weights=./output/mobilenetv1/yolov3_mobilenet_v1/best_model
 ```
+
 ## 预测
 
 ### python预测
@@ -158,7 +173,6 @@ python slim/quantization/infer.py --not_quant_pattern yolo_output \
 -o weights=./output/mobilenetv1/yolov3_mobilenet_v1/best_model
 ```
 
-
 ### PaddleLite预测
 导出模型步骤中导出的FP32模型可使用PaddleLite进行加载预测，可参见教程[Paddle-Lite如何加载运行量化模型](https://github.com/PaddlePaddle/Paddle-Lite/wiki/model_quantization)
 

slim/quantization/pact.py

+# 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 sys
+import paddle
+import paddle.fluid as fluid
+from paddleslim.quant import quant_aware, convert
+import numpy as np
+
+from paddle.fluid.layer_helper import LayerHelper
+
+
+def pact(x, name=None):
+    helper = LayerHelper("pact", **locals())
+    dtype = 'float32'
+    init_thres = 20
+    u_param_attr = fluid.ParamAttr(
+        name=x.name + '_pact',
+        initializer=fluid.initializer.ConstantInitializer(value=init_thres),
+        regularizer=fluid.regularizer.L2Decay(0.0001),
+        learning_rate=1)
+    u_param = helper.create_parameter(attr=u_param_attr, shape=[1], dtype=dtype)
+    x = fluid.layers.elementwise_sub(
+        x, fluid.layers.relu(fluid.layers.elementwise_sub(x, u_param)))
+    x = fluid.layers.elementwise_add(
+        x, fluid.layers.relu(fluid.layers.elementwise_sub(-u_param, x)))
+
+    return x
+
+
+def get_optimizer():
+    return fluid.optimizer.MomentumOptimizer(0.0001, 0.9)

slim/quantization/train.py

@@ -39,6 +39,7 @@ from ppdet.utils.cli import ArgsParser
 from ppdet.utils.check import check_gpu, check_version, check_config
 import ppdet.utils.checkpoint as checkpoint
 from paddleslim.quant import quant_aware, convert
+from pact import pact, get_optimizer
 import logging
 FORMAT = '%(asctime)s-%(levelname)s: %(message)s'
 logging.basicConfig(level=logging.INFO, format=FORMAT)
@@ -51,14 +52,6 @@ def save_checkpoint(exe, prog, path, train_prog):
     logger.info('Save model to {}.'.format(path))
     fluid.io.save_persistables(exe, path, main_program=prog)
 
-    v = train_prog.global_block().var('@LR_DECAY_COUNTER@')
-    fluid.io.save_vars(exe, dirname=path, vars=[v])
-
-
-def load_global_step(exe, prog, path):
-    v = prog.global_block().var('@LR_DECAY_COUNTER@')
-    fluid.io.load_vars(exe, path, prog, [v])
-
 
 def main():
     if FLAGS.eval is False:
@@ -105,9 +98,10 @@ def main():
     with fluid.program_guard(train_prog, startup_prog):
         with fluid.unique_name.guard():
             model = create(main_arch)
-
             inputs_def = cfg['TrainReader']['inputs_def']
             feed_vars, train_loader = model.build_inputs(**inputs_def)
+            if FLAGS.use_pact:
+                feed_vars['image'].stop_gradient = False
             train_fetches = model.train(feed_vars)
             loss = train_fetches['loss']
             lr = lr_builder()
@@ -181,17 +175,30 @@ def main():
 
     fuse_bn = getattr(model.backbone, 'norm_type', None) == 'affine_channel'
 
-    if not FLAGS.resume_checkpoint:
-        if cfg.pretrain_weights and fuse_bn and not ignore_params:
-            checkpoint.load_and_fusebn(exe, train_prog, cfg.pretrain_weights)
-        elif cfg.pretrain_weights:
-            checkpoint.load_params(
-                exe,
-                train_prog,
-                cfg.pretrain_weights,
-                ignore_params=ignore_params)
+    if cfg.pretrain_weights and fuse_bn and not ignore_params:
+        checkpoint.load_and_fusebn(exe, train_prog, cfg.pretrain_weights)
+    elif cfg.pretrain_weights:
+        checkpoint.load_params(
+            exe, train_prog, cfg.pretrain_weights, ignore_params=ignore_params)
+
+    if FLAGS.use_pact:
+        act_preprocess_func = pact
+        optimizer_func = get_optimizer
+        executor = exe
+    else:
+        act_preprocess_func = None
+        optimizer_func = None
+        executor = None
     # insert quantize op in train_prog, return type is CompiledProgram
-    train_prog_quant = quant_aware(train_prog, place, config, for_test=False)
+    train_prog_quant = quant_aware(
+        train_prog,
+        place,
+        config,
+        scope=None,
+        act_preprocess_func=act_preprocess_func,
+        optimizer_func=optimizer_func,
+        executor=executor,
+        for_test=False)
 
     compiled_train_prog = train_prog_quant.with_data_parallel(
         loss_name=loss.name,
@@ -200,14 +207,18 @@ def main():
 
     if FLAGS.eval:
         # insert quantize op in eval_prog
-        eval_prog = quant_aware(eval_prog, place, config, for_test=True)
+        eval_prog = quant_aware(
+            eval_prog,
+            place,
+            config,
+            scope=None,
+            act_preprocess_func=act_preprocess_func,
+            optimizer_func=optimizer_func,
+            executor=executor,
+            for_test=True)
         compiled_eval_prog = fluid.CompiledProgram(eval_prog)
 
     start_iter = 0
-    if FLAGS.resume_checkpoint:
-        checkpoint.load_checkpoint(exe, eval_prog, FLAGS.resume_checkpoint)
-        load_global_step(exe, train_prog, FLAGS.resume_checkpoint)
-        start_iter = checkpoint.global_step()
 
     train_reader = create_reader(cfg.TrainReader,
                                  (cfg.max_iters - start_iter) * devices_num)
@@ -253,8 +264,6 @@ def main():
         if (it > 0 and it % cfg.snapshot_iter == 0 or it == cfg.max_iters - 1) \
            and (not FLAGS.dist or trainer_id == 0):
             save_name = str(it) if it != cfg.max_iters - 1 else "model_final"
-            save_checkpoint(exe, eval_prog,
-                            os.path.join(save_dir, save_name), train_prog)
 
             if FLAGS.eval:
                 # evaluation
@@ -288,12 +297,6 @@ def main():
 
 if __name__ == '__main__':
     parser = ArgsParser()
-    parser.add_argument(
-        "-r",
-        "--resume_checkpoint",
-        default=None,
-        type=str,
-        help="Checkpoint path for resuming training.")
     parser.add_argument(
         "--loss_scale",
         default=8.,
@@ -315,5 +318,7 @@ if __name__ == '__main__':
         type=str,
         help="Layers which name_scope contains string in not_quant_pattern will not be quantized"
     )
+    parser.add_argument(
+        "--use_pact", nargs='+', type=bool, help="Whether to use PACT or not.")
     FLAGS = parser.parse_args()
     main()