eric-haibin-lin commented on a change in pull request #14885: [Fit-API] Adress
PR comments
URL: https://github.com/apache/incubator-mxnet/pull/14885#discussion_r282141026
##########
File path: python/mxnet/gluon/contrib/estimator/event_handler.py
##########
@@ -263,108 +268,171 @@ def epoch_end(self, estimator, *args, **kwargs):
class CheckpointHandler(BatchEnd, EpochEnd):
- """Save the model after every epoch.
+ """Save the model after user define period
- :py:class:`CheckpointHandler` save the network parameters every epoch
+ :py:class:`CheckpointHandler` saves the network architecture after first
batch if the model
+ can be hybridized(), saves model parameters and trainer states after user
defined period,
+ default saves every epoch.
Parameters
----------
- filepath : str
- file name to save the parameters, it can contain directories,
- for example: ./saved_model/resnet.params
+ model_dir : str
+ file directory to save all the model related files including model
architecture,
+ model parameters, and trainer states.
+ model_prefix : str default 'model'
+ prefix to add for all checkpoint file names
monitor: EvalMetric
- the metrics to monitor
+ the metrics to monitor and determine if model has improved
verbose: int, default 0
verbosity mode
- save_best_only: bool
- if True, only save the parameters if monitored value improved
+ save_best: bool
+ if True, save the model parameters and trainer states with the best
monitored value
mode: str, default 'auto'
one of {auto, min, max}, if `save_best_only=True`, the comparison to
make
- and determine if the monitored value has improved
- period: int, default 1
- intervals between saving the network
+ and determine if the monitored value has improved. if 'auto' mode,
checkpoint
+ handler will try to use min or max based on the monitored metric name
+ epoch_period: int, default 1
+ epoch intervals between saving the network
+ batch_period: int, default None
+ batch intervals between saving the network,
+ by default don't save any checkpoint based on number of batches
Review comment:
by default, checkpoints are not saved based on the number of batches
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
With regards,
Apache Git Services
