save best and test on tiny/s0

deepspeech/training/trainer.py

@@ -18,7 +18,7 @@ import paddle
 from paddle import distributed as dist
 from tensorboardX import SummaryWriter
 
-from deepspeech.utils import checkpoint
+from deepspeech.utils.checkpoint import KBestCheckpoint
 from deepspeech.utils import mp_tools
 from deepspeech.utils.log import Log
 
@@ -139,9 +139,12 @@ class Trainer():
             "epoch": self.epoch,
             "lr": self.optimizer.get_lr()
         })
-        checkpoint.save_parameters(self.checkpoint_dir, self.iteration
+        self.checkpoint.add_checkpoint(self.checkpoint_dir, self.iteration
                                    if tag is None else tag, self.model,
                                    self.optimizer, infos)
+        # checkpoint.save_parameters(self.checkpoint_dir, self.iteration
+        #                            if tag is None else tag, self.model,
+        #                            self.optimizer, infos)
 
     def resume_or_scratch(self):
         """Resume from latest checkpoint at checkpoints in the output 
@@ -151,7 +154,7 @@ class Trainer():
         resume training.
         """
         scratch = None
-        infos = checkpoint.load_parameters(
+        infos = self.checkpoint.load_parameters(
             self.model,
             self.optimizer,
             checkpoint_dir=self.checkpoint_dir,
@@ -180,7 +183,7 @@ class Trainer():
         from_scratch = self.resume_or_scratch()
         if from_scratch:
             # save init model, i.e. 0 epoch
-            self.save(tag='init')
+            self.save(tag='init', infos=None)
 
         self.lr_scheduler.step(self.iteration)
         if self.parallel:
@@ -263,6 +266,9 @@ class Trainer():
 
         self.checkpoint_dir = checkpoint_dir
 
+        self.checkpoint = KBestCheckpoint(max_size=self.config.training.max_epoch, 
+                            last_size=self.config.training.last_epoch)
+
     @mp_tools.rank_zero_only
     def destory(self):
         """Close visualizer to avoid hanging after training"""

deepspeech/utils/checkpoint.py

@@ -23,130 +23,226 @@ from paddle.optimizer import Optimizer
 from deepspeech.utils import mp_tools
 from deepspeech.utils.log import Log
 
+import glob
+
 logger = Log(__name__).getlog()
 
 __all__ = ["load_parameters", "save_parameters"]
 
+class KBestCheckpoint(object):
+    def __init__(self,
+                 max_size: int=5,
+                 last_size: int=1):
+        self.best_records: Mapping[Path, float] = {}
+        self.last_records = []
+        self.max_size = max_size
+        self.last_size = last_size
+        self._save_all = (max_size == -1)
+    
+    def should_save_best(self, metric: float) -> bool:
+        if not self.best_full():
+            return True
+
+        # already full
+        worst_record_path = max(self.best_records, key=self.best_records.get)
+        worst_metric = self.best_records[worst_record_path]
+        return metric < worst_metric
+
+    def best_full(self):
+        return (not self._save_all) and len(self.best_records) == self.max_size
+    
+    def last_full(self):
+        return len(self.last_records) == self.last_size
+
+    def add_checkpoint(self,
+                        checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos):
+        if("val_loss" not in infos.keys()):
+            self.save_parameters(checkpoint_dir, tag_or_iteration, 
+                                model, optimizer, infos)
+            return
+
+        #save best
+        if self.should_save_best(infos["val_loss"]):
+            self.save_checkpoint_and_update(infos["val_loss"], 
+                        checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos)
+        #save last
+        self.save_last_checkpoint_and_update(checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos)
+        
+        if isinstance(tag_or_iteration, int):
+            self._save_record(checkpoint_dir, tag_or_iteration)
+    
+    def save_checkpoint_and_update(self, metric, 
+                        checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos):
+        # remove the worst
+        if self.best_full():
+            worst_record_path = max(self.best_records,
+                                    key=self.best_records.get)
+            self.best_records.pop(worst_record_path)
+            if(worst_record_path not in self.last_records):
+                print('----to remove (best)----')
+                print(worst_record_path)
+                self.del_checkpoint(checkpoint_dir, worst_record_path)
+
+        # add the new one
+        self.save_parameters(checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos)
+        self.best_records[tag_or_iteration] = metric
+ 
+    def save_last_checkpoint_and_update(self, checkpoint_dir, tag_or_iteration,
+                        model, optimizer, infos):
+        # remove the old
+        if self.last_full():
+            to_del_fn = self.last_records.pop(0)
+            if(to_del_fn not in self.best_records.keys()):
+                print('----to remove (last)----')
+                print(to_del_fn)
+                self.del_checkpoint(checkpoint_dir, to_del_fn)
+        self.last_records.append(tag_or_iteration)
+
+        self.save_parameters(checkpoint_dir, tag_or_iteration,
+                    model, optimizer, infos)
+            # with open(os.path.join(checkpoint_dir, "checkpoint"), "w") as handle:
+            #     for iteration in self.best_records
+            #     handle.write("model_checkpoint_path:{}\n".format(iteration))
+
+
+    def del_checkpoint(self, checkpoint_dir, tag_or_iteration):
+        checkpoint_path = os.path.join(checkpoint_dir,
+                                    "{}".format(tag_or_iteration))
+        for filename in glob.glob(checkpoint_path+".*"):
+            os.remove(filename)
+            print("delete file: "+filename)
+        
+
+
+    def _load_latest_checkpoint(self, checkpoint_dir: str) -> int:
+        """Get the iteration number corresponding to the latest saved checkpoint.
+        Args:
+            checkpoint_dir (str): the directory where checkpoint is saved.
+        Returns:
+            int: the latest iteration number. -1 for no checkpoint to load.
+        """
+        checkpoint_record = os.path.join(checkpoint_dir, "checkpoint_last")
+        if not os.path.isfile(checkpoint_record):
+            return -1
+
+        # Fetch the latest checkpoint index.
+        with open(checkpoint_record, "rt") as handle:
+            latest_checkpoint = handle.readlines()[-1].strip()
+            iteration = int(latest_checkpoint.split(":")[-1])
+        return iteration
+
+
+    def _save_record(self, checkpoint_dir: str, iteration: int):
+        """Save the iteration number of the latest model to be checkpoint record.
+        Args:
+            checkpoint_dir (str): the directory where checkpoint is saved.
+            iteration (int): the latest iteration number.
+        Returns:
+            None
+        """
+        checkpoint_record_last = os.path.join(checkpoint_dir, "checkpoint_last")
+        checkpoint_record_best = os.path.join(checkpoint_dir, "checkpoint_best")
+        # Update the latest checkpoint index.
+        # with open(checkpoint_record, "a+") as handle:
+        #     handle.write("model_checkpoint_path:{}\n".format(iteration))
+        with open(checkpoint_record_best, "w") as handle:
+            for i in self.best_records.keys():
+                handle.write("model_checkpoint_path:{}\n".format(i))
+        with open(checkpoint_record_last, "w") as handle:
+            for i in self.last_records:
+                handle.write("model_checkpoint_path:{}\n".format(i))
+
+
+    def load_parameters(self, model,
+                        optimizer=None,
+                        checkpoint_dir=None,
+                        checkpoint_path=None):
+        """Load a specific model checkpoint from disk. 
+        Args:
+            model (Layer): model to load parameters.
+            optimizer (Optimizer, optional): optimizer to load states if needed.
+                Defaults to None.
+            checkpoint_dir (str, optional): the directory where checkpoint is saved.
+            checkpoint_path (str, optional): if specified, load the checkpoint
+                stored in the checkpoint_path(prefix) and the argument 'checkpoint_dir' will 
+                be ignored. Defaults to None. 
+        Returns:
+            configs (dict): epoch or step, lr and other meta info should be saved.
+        """
+        configs = {}
+
+        if checkpoint_path is not None:
+            tag = os.path.basename(checkpoint_path).split(":")[-1]
+        elif checkpoint_dir is not None:
+            iteration = self._load_latest_checkpoint(checkpoint_dir)
+            if iteration == -1:
+                return configs
+            checkpoint_path = os.path.join(checkpoint_dir, "{}".format(iteration))
+        else:
+            raise ValueError(
+                "At least one of 'checkpoint_dir' and 'checkpoint_path' should be specified!"
+            )
+
+        rank = dist.get_rank()
+
+        params_path = checkpoint_path + ".pdparams"
+        model_dict = paddle.load(params_path)
+        model.set_state_dict(model_dict)
+        logger.info("Rank {}: loaded model from {}".format(rank, params_path))
 
-def _load_latest_checkpoint(checkpoint_dir: str) -> int:
-    """Get the iteration number corresponding to the latest saved checkpoint.
-    Args:
-        checkpoint_dir (str): the directory where checkpoint is saved.
-    Returns:
-        int: the latest iteration number. -1 for no checkpoint to load.
-    """
-    checkpoint_record = os.path.join(checkpoint_dir, "checkpoint")
-    if not os.path.isfile(checkpoint_record):
-        return -1
-
-    # Fetch the latest checkpoint index.
-    with open(checkpoint_record, "rt") as handle:
-        latest_checkpoint = handle.readlines()[-1].strip()
-        iteration = int(latest_checkpoint.split(":")[-1])
-    return iteration
-
-
-def _save_record(checkpoint_dir: str, iteration: int):
-    """Save the iteration number of the latest model to be checkpoint record.
-    Args:
-        checkpoint_dir (str): the directory where checkpoint is saved.
-        iteration (int): the latest iteration number.
-    Returns:
-        None
-    """
-    checkpoint_record = os.path.join(checkpoint_dir, "checkpoint")
-    # Update the latest checkpoint index.
-    with open(checkpoint_record, "a+") as handle:
-        handle.write("model_checkpoint_path:{}\n".format(iteration))
-
-
-def load_parameters(model,
-                    optimizer=None,
-                    checkpoint_dir=None,
-                    checkpoint_path=None):
-    """Load a specific model checkpoint from disk. 
-    Args:
-        model (Layer): model to load parameters.
-        optimizer (Optimizer, optional): optimizer to load states if needed.
-            Defaults to None.
-        checkpoint_dir (str, optional): the directory where checkpoint is saved.
-        checkpoint_path (str, optional): if specified, load the checkpoint
-            stored in the checkpoint_path(prefix) and the argument 'checkpoint_dir' will 
-            be ignored. Defaults to None. 
-    Returns:
-        configs (dict): epoch or step, lr and other meta info should be saved.
-    """
-    configs = {}
-
-    if checkpoint_path is not None:
-        tag = os.path.basename(checkpoint_path).split(":")[-1]
-    elif checkpoint_dir is not None:
-        iteration = _load_latest_checkpoint(checkpoint_dir)
-        if iteration == -1:
-            return configs
-        checkpoint_path = os.path.join(checkpoint_dir, "{}".format(iteration))
-    else:
-        raise ValueError(
-            "At least one of 'checkpoint_dir' and 'checkpoint_path' should be specified!"
-        )
-
-    rank = dist.get_rank()
-
-    params_path = checkpoint_path + ".pdparams"
-    model_dict = paddle.load(params_path)
-    model.set_state_dict(model_dict)
-    logger.info("Rank {}: loaded model from {}".format(rank, params_path))
-
-    optimizer_path = checkpoint_path + ".pdopt"
-    if optimizer and os.path.isfile(optimizer_path):
-        optimizer_dict = paddle.load(optimizer_path)
-        optimizer.set_state_dict(optimizer_dict)
-        logger.info("Rank {}: loaded optimizer state from {}".format(
-            rank, optimizer_path))
-
-    info_path = re.sub('.pdparams$', '.json', params_path)
-    if os.path.exists(info_path):
-        with open(info_path, 'r') as fin:
-            configs = json.load(fin)
-    return configs
-
-
-@mp_tools.rank_zero_only
-def save_parameters(checkpoint_dir: str,
-                    tag_or_iteration: Union[int, str],
-                    model: paddle.nn.Layer,
-                    optimizer: Optimizer=None,
-                    infos: dict=None):
-    """Checkpoint the latest trained model parameters.
-    Args:
-        checkpoint_dir (str): the directory where checkpoint is saved.
-        tag_or_iteration (int or str): the latest iteration(step or epoch) number.
-        model (Layer): model to be checkpointed.
-        optimizer (Optimizer, optional): optimizer to be checkpointed.
-            Defaults to None.
-        infos (dict or None): any info you want to save.
-    Returns:
-        None
-    """
-    checkpoint_path = os.path.join(checkpoint_dir,
-                                   "{}".format(tag_or_iteration))
-
-    model_dict = model.state_dict()
-    params_path = checkpoint_path + ".pdparams"
-    paddle.save(model_dict, params_path)
-    logger.info("Saved model to {}".format(params_path))
-
-    if optimizer:
-        opt_dict = optimizer.state_dict()
         optimizer_path = checkpoint_path + ".pdopt"
-        paddle.save(opt_dict, optimizer_path)
-        logger.info("Saved optimzier state to {}".format(optimizer_path))
-
-    info_path = re.sub('.pdparams$', '.json', params_path)
-    infos = {} if infos is None else infos
-    with open(info_path, 'w') as fout:
-        data = json.dumps(infos)
-        fout.write(data)
+        if optimizer and os.path.isfile(optimizer_path):
+            optimizer_dict = paddle.load(optimizer_path)
+            optimizer.set_state_dict(optimizer_dict)
+            logger.info("Rank {}: loaded optimizer state from {}".format(
+                rank, optimizer_path))
+
+        info_path = re.sub('.pdparams$', '.json', params_path)
+        if os.path.exists(info_path):
+            with open(info_path, 'r') as fin:
+                configs = json.load(fin)
+        return configs
+
+
+    @mp_tools.rank_zero_only
+    def save_parameters(self, checkpoint_dir: str,
+                        tag_or_iteration: Union[int, str],
+                        model: paddle.nn.Layer,
+                        optimizer: Optimizer=None,
+                        infos: dict=None):
+        """Checkpoint the latest trained model parameters.
+        Args:
+            checkpoint_dir (str): the directory where checkpoint is saved.
+            tag_or_iteration (int or str): the latest iteration(step or epoch) number.
+            model (Layer): model to be checkpointed.
+            optimizer (Optimizer, optional): optimizer to be checkpointed.
+                Defaults to None.
+            infos (dict or None): any info you want to save.
+        Returns:
+            None
+        """
+        checkpoint_path = os.path.join(checkpoint_dir,
+                                    "{}".format(tag_or_iteration))
+
+        model_dict = model.state_dict()
+        params_path = checkpoint_path + ".pdparams"
+        paddle.save(model_dict, params_path)
+        logger.info("Saved model to {}".format(params_path))
+
+        if optimizer:
+            opt_dict = optimizer.state_dict()
+            optimizer_path = checkpoint_path + ".pdopt"
+            paddle.save(opt_dict, optimizer_path)
+            logger.info("Saved optimzier state to {}".format(optimizer_path))
+
+        info_path = re.sub('.pdparams$', '.json', params_path)
+        infos = {} if infos is None else infos
+        with open(info_path, 'w') as fout:
+            data = json.dumps(infos)
+            fout.write(data)
 
-    if isinstance(tag_or_iteration, int):
-        _save_record(checkpoint_dir, tag_or_iteration)

examples/tiny/s0/conf/deepspeech2.yaml

@@ -43,12 +43,15 @@ model:
   share_rnn_weights: True 
 
 training:
-  n_epoch: 24
+  n_epoch: 6
   lr: 1e-5 
   lr_decay: 1.0 
   weight_decay: 1e-06
   global_grad_clip: 5.0
   log_interval: 1
+  max_epoch: 3
+  last_epoch: 2
+
 
 decoding:
   batch_size: 128