Merge pull request #1853 from HydrogenSulfate/multi_optim

support for multi optimizer case

ppcls/engine/engine.py

@@ -214,16 +214,19 @@ class Engine(object):
         if self.config["Global"]["pretrained_model"] is not None:
             if self.config["Global"]["pretrained_model"].startswith("http"):
                 load_dygraph_pretrain_from_url(
-                    self.model, self.config["Global"]["pretrained_model"])
+                    [self.model, getattr(self, 'train_loss_func', None)],
+                    self.config["Global"]["pretrained_model"])
             else:
                 load_dygraph_pretrain(
-                    self.model, self.config["Global"]["pretrained_model"])
+                    [self.model, getattr(self, 'train_loss_func', None)],
+                    self.config["Global"]["pretrained_model"])
 
         # build optimizer
         if self.mode == 'train':
             self.optimizer, self.lr_sch = build_optimizer(
-                self.config["Optimizer"], self.config["Global"]["epochs"],
-                len(self.train_dataloader), [self.model])
+                self.config, self.config["Global"]["epochs"],
+                len(self.train_dataloader),
+                [self.model, self.train_loss_func])
 
         # for amp training
         if self.amp:
@@ -241,6 +244,11 @@ class Engine(object):
                 optimizers=self.optimizer,
                 level=amp_level,
                 save_dtype='float32')
+            if len(self.train_loss_func.parameters()) > 0:
+                self.train_loss_func = paddle.amp.decorate(
+                    models=self.train_loss_func,
+                    level=amp_level,
+                    save_dtype='float32')
 
         # for distributed
         world_size = dist.get_world_size()
@@ -251,7 +259,10 @@ class Engine(object):
         if self.config["Global"]["distributed"]:
             dist.init_parallel_env()
             self.model = paddle.DataParallel(self.model)
-
+            if self.mode == 'train' and len(self.train_loss_func.parameters(
+            )) > 0:
+                self.train_loss_func = paddle.DataParallel(
+                    self.train_loss_func)
         # build postprocess for infer
         if self.mode == 'infer':
             self.preprocess_func = create_operators(self.config["Infer"][
@@ -279,9 +290,9 @@ class Engine(object):
         # global iter counter
         self.global_step = 0
 
-        if self.config["Global"]["checkpoints"] is not None:
-            metric_info = init_model(self.config["Global"], self.model,
-                                     self.optimizer)
+        if self.config.Global.checkpoints is not None:
+            metric_info = init_model(self.config.Global, self.model,
+                                     self.optimizer, self.train_loss_func)
             if metric_info is not None:
                 best_metric.update(metric_info)
 
@@ -317,7 +328,8 @@ class Engine(object):
                         best_metric,
                         self.output_dir,
                         model_name=self.config["Arch"]["name"],
-                        prefix="best_model")
+                        prefix="best_model",
+                        loss=self.train_loss_func)
                 logger.info("[Eval][Epoch {}][best metric: {}]".format(
                     epoch_id, best_metric["metric"]))
                 logger.scaler(
@@ -336,7 +348,8 @@ class Engine(object):
                                      "epoch": epoch_id},
                     self.output_dir,
                     model_name=self.config["Arch"]["name"],
-                    prefix="epoch_{}".format(epoch_id))
+                    prefix="epoch_{}".format(epoch_id),
+                    loss=self.train_loss_func)
             # save the latest model
             save_load.save_model(
                 self.model,
@@ -344,7 +357,8 @@ class Engine(object):
                                  "epoch": epoch_id},
                 self.output_dir,
                 model_name=self.config["Arch"]["name"],
-                prefix="latest")
+                prefix="latest",
+                loss=self.train_loss_func)
 
         if self.vdl_writer is not None:
             self.vdl_writer.close()

ppcls/engine/train/train.py

@@ -53,16 +53,22 @@ def train_epoch(engine, epoch_id, print_batch_step):
             out = forward(engine, batch)
             loss_dict = engine.train_loss_func(out, batch[1])
 
-        # step opt and lr
+        # step opt
         if engine.amp:
             scaled = engine.scaler.scale(loss_dict["loss"])
             scaled.backward()
-            engine.scaler.minimize(engine.optimizer, scaled)
+            for i in range(len(engine.optimizer)):
+                engine.scaler.minimize(engine.optimizer[i], scaled)
         else:
             loss_dict["loss"].backward()
-            engine.optimizer.step()
-        engine.optimizer.clear_grad()
-        engine.lr_sch.step()
+            for i in range(len(engine.optimizer)):
+                engine.optimizer[i].step()
+        # clear grad
+        for i in range(len(engine.optimizer)):
+            engine.optimizer[i].clear_grad()
+        # step lr
+        for i in range(len(engine.lr_sch)):
+            engine.lr_sch[i].step()
 
         # below code just for logging
         # update metric_for_logger

ppcls/engine/train/utils.py

@@ -38,7 +38,10 @@ def update_loss(trainer, loss_dict, batch_size):
 
 
 def log_info(trainer, batch_size, epoch_id, iter_id):
-    lr_msg = "lr: {:.5f}".format(trainer.lr_sch.get_lr())
+    lr_msg = ", ".join([
+        "lr_{}: {:.8f}".format(i + 1, lr.get_lr())
+        for i, lr in enumerate(trainer.lr_sch)
+    ])
     metric_msg = ", ".join([
         "{}: {:.5f}".format(key, trainer.output_info[key].avg)
         for key in trainer.output_info
@@ -59,9 +62,10 @@ def log_info(trainer, batch_size, epoch_id, iter_id):
         len(trainer.train_dataloader), lr_msg, metric_msg, time_msg, ips_msg,
         eta_msg))
 
+    for i, lr in enumerate(trainer.lr_sch):
         logger.scaler(
-        name="lr",
-        value=trainer.lr_sch.get_lr(),
+            name="lr_{}".format(i + 1),
+            value=lr.get_lr(),
             step=trainer.global_step,
             writer=trainer.vdl_writer)
     for key in trainer.output_info:

ppcls/loss/__init__.py

@@ -47,6 +47,7 @@ class CombinedLoss(nn.Layer):
                 param.keys())
             self.loss_weight.append(param.pop("weight"))
             self.loss_func.append(eval(name)(**param))
+            self.loss_func = nn.LayerList(self.loss_func)
 
     def __call__(self, input, batch):
         loss_dict = {}

ppcls/optimizer/__init__.py

@@ -18,6 +18,7 @@ from __future__ import print_function
 
 import copy
 import paddle
+from typing import Dict, List
 
 from ppcls.utils import logger
 
@@ -44,29 +45,78 @@ def build_lr_scheduler(lr_config, epochs, step_each_epoch):
 # model_list is None in static graph
 def build_optimizer(config, epochs, step_each_epoch, model_list=None):
     config = copy.deepcopy(config)
+    optim_config = config["Optimizer"]
+    if isinstance(optim_config, dict):
+        # convert {'name': xxx, **optim_cfg} to [{name: {scope: xxx, **optim_cfg}}]
+        optim_name = optim_config.pop("name")
+        optim_config: List[Dict[str, Dict]] = [{
+            optim_name: {
+                'scope': "all",
+                **
+                optim_config
+            }
+        }]
+    optim_list = []
+    lr_list = []
+    """NOTE:
+    Currently only support optim objets below.
+    1. single optimizer config.
+    2. next level uner Arch, such as Arch.backbone, Arch.neck, Arch.head.
+    3. loss which has parameters, such as CenterLoss.
+    """
+    for optim_item in optim_config:
+        # optim_cfg = {optim_name: {scope: xxx, **optim_cfg}}
         # step1 build lr
-    lr = build_lr_scheduler(config.pop('lr'), epochs, step_each_epoch)
-    logger.debug("build lr ({}) success..".format(lr))
+        optim_name = list(optim_item.keys())[0]  # get optim_name
+        optim_scope = optim_item[optim_name].pop('scope')  # get optim_scope
+        optim_cfg = optim_item[optim_name]  # get optim_cfg
+
+        lr = build_lr_scheduler(optim_cfg.pop('lr'), epochs, step_each_epoch)
+        logger.debug("build lr ({}) for scope ({}) success..".format(
+            lr, optim_scope))
         # step2 build regularization
-    if 'regularizer' in config and config['regularizer'] is not None:
-        if 'weight_decay' in config:
+        if 'regularizer' in optim_cfg and optim_cfg['regularizer'] is not None:
+            if 'weight_decay' in optim_cfg:
                 logger.warning(
                     "ConfigError: Only one of regularizer and weight_decay can be set in Optimizer Config. \"weight_decay\" has been ignored."
                 )
-        reg_config = config.pop('regularizer')
+            reg_config = optim_cfg.pop('regularizer')
             reg_name = reg_config.pop('name') + 'Decay'
             reg = getattr(paddle.regularizer, reg_name)(**reg_config)
-        config["weight_decay"] = reg
-        logger.debug("build regularizer ({}) success..".format(reg))
+            optim_cfg["weight_decay"] = reg
+            logger.debug("build regularizer ({}) for scope ({}) success..".
+                         format(reg, optim_scope))
         # step3 build optimizer
-    optim_name = config.pop('name')
-    if 'clip_norm' in config:
-        clip_norm = config.pop('clip_norm')
+        if 'clip_norm' in optim_cfg:
+            clip_norm = optim_cfg.pop('clip_norm')
             grad_clip = paddle.nn.ClipGradByNorm(clip_norm=clip_norm)
         else:
             grad_clip = None
-    optim = getattr(optimizer, optim_name)(learning_rate=lr,
-                                           grad_clip=grad_clip,
-                                           **config)(model_list=model_list)
-    logger.debug("build optimizer ({}) success..".format(optim))
-    return optim, lr
+        optim_model = []
+        for i in range(len(model_list)):
+            if len(model_list[i].parameters()) == 0:
+                continue
+            if optim_scope == "all":
+                # optimizer for all
+                optim_model.append(model_list[i])
+            else:
+                if optim_scope.endswith("Loss"):
+                    # optimizer for loss
+                    for m in model_list[i].sublayers(True):
+                        if m.__class_name == optim_scope:
+                            optim_model.append(m)
+                else:
+                    # opmizer for module in model, such as backbone, neck, head...
+                    if hasattr(model_list[i], optim_scope):
+                        optim_model.append(getattr(model_list[i], optim_scope))
+
+        assert len(optim_model) == 1, \
+            "Invalid optim model for optim scope({}), number of optim_model={}".format(optim_scope, len(optim_model))
+        optim = getattr(optimizer, optim_name)(
+            learning_rate=lr, grad_clip=grad_clip,
+            **optim_cfg)(model_list=optim_model)
+        logger.debug("build optimizer ({}) for scope ({}) success..".format(
+            optim, optim_scope))
+        optim_list.append(optim)
+        lr_list.append(lr)
+    return optim_list, lr_list

ppcls/utils/save_load.py

@@ -18,9 +18,6 @@ from __future__ import print_function
 
 import errno
 import os
-import re
-import shutil
-import tempfile
 
 import paddle
 from ppcls.utils import logger
@@ -47,9 +44,14 @@ def _mkdir_if_not_exist(path):
 
 def load_dygraph_pretrain(model, path=None):
     if not (os.path.isdir(path) or os.path.exists(path + '.pdparams')):
-        raise ValueError("Model pretrain path {} does not "
+        raise ValueError("Model pretrain path {}.pdparams does not "
                          "exists.".format(path))
     param_state_dict = paddle.load(path + ".pdparams")
+    if isinstance(model, list):
+        for m in model:
+            if hasattr(m, 'set_dict'):
+                m.set_dict(param_state_dict)
+    else:
         model.set_dict(param_state_dict)
     return
 
@@ -85,7 +87,7 @@ def load_distillation_model(model, pretrained_model):
             pretrained_model))
 
 
-def init_model(config, net, optimizer=None):
+def init_model(config, net, optimizer=None, loss: paddle.nn.Layer=None):
     """
     load model from checkpoint or pretrained_model
     """
@@ -95,11 +97,15 @@ def init_model(config, net, optimizer=None):
             "Given dir {}.pdparams not exist.".format(checkpoints)
         assert os.path.exists(checkpoints + ".pdopt"), \
             "Given dir {}.pdopt not exist.".format(checkpoints)
-        para_dict = paddle.load(checkpoints + ".pdparams")
+        # load state dict
         opti_dict = paddle.load(checkpoints + ".pdopt")
+        para_dict = paddle.load(checkpoints + ".pdparams")
         metric_dict = paddle.load(checkpoints + ".pdstates")
-        net.set_dict(para_dict)
-        optimizer.set_state_dict(opti_dict)
+        # set state dict
+        net.set_state_dict(para_dict)
+        loss.set_state_dict(para_dict)
+        for i in range(len(optimizer)):
+            optimizer[i].set_state_dict(opti_dict)
         logger.info("Finish load checkpoints from {}".format(checkpoints))
         return metric_dict
 
@@ -120,7 +126,8 @@ def save_model(net,
                metric_info,
                model_path,
                model_name="",
-               prefix='ppcls'):
+               prefix='ppcls',
+               loss: paddle.nn.Layer=None):
     """
     save model to the target path
     """
@@ -130,7 +137,14 @@ def save_model(net,
     _mkdir_if_not_exist(model_path)
     model_path = os.path.join(model_path, prefix)
 
-    paddle.save(net.state_dict(), model_path + ".pdparams")
-    paddle.save(optimizer.state_dict(), model_path + ".pdopt")
+    params_state_dict = net.state_dict()
+    loss_state_dict = loss.state_dict()
+    keys_inter = set(params_state_dict.keys()) & set(loss_state_dict.keys())
+    assert len(keys_inter) == 0, \
+        f"keys in model and loss state_dict must be unique, but got intersection {keys_inter}"
+    params_state_dict.update(loss_state_dict)
+
+    paddle.save(params_state_dict, model_path + ".pdparams")
+    paddle.save([opt.state_dict() for opt in optimizer], model_path + ".pdopt")
     paddle.save(metric_info, model_path + ".pdstates")
     logger.info("Already save model in {}".format(model_path))