&`iddlZddlZddlZddlmZddlmZmZddlZ ddlm Z ddl m Z ddl mZerddlmZed Gd d e ZdS) N)Path) TYPE_CHECKINGOptional)keras)FrameworkCheckpoint) PublicAPI) Preprocessorbeta) stabilityceZdZdZdZedddejdedddfd Z eddd e dedddfd Z eddd e dedddfd Z de jjfdZdS)TensorflowCheckpointzKA :py:class:`~ray.train.Checkpoint` with TensorFlow-specific functionality._model_filenameN) preprocessormodelrr returnc*tj}d}|t||||}|r||||j|i|S)auCreate a :py:class:`~ray.train.Checkpoint` that stores a Keras model. The checkpoint created with this method needs to be paired with `model` when used. Args: model: The Keras model, whose weights are stored in the checkpoint. preprocessor: A fitted preprocessor to be applied before inference. Returns: A :py:class:`TensorflowCheckpoint` containing the specified model. Examples: .. testcode:: from ray.train.tensorflow import TensorflowCheckpoint import tensorflow as tf model = tf.keras.applications.resnet.ResNet101() checkpoint = TensorflowCheckpoint.from_model(model) .. testoutput:: :options: +MOCK :hide: ... # Model may or may not be downloaded z model.keras) tempfilemkdtempsaveras_posixfrom_directoryset_preprocessorupdate_metadataMODEL_FILENAME_KEY)clsrrtempdirfilename checkpoints ~/home/jaya/work/projects/VOICE-AGENT/VIET/agent-env/lib/python3.11/site-packages/ray/train/tensorflow/tensorflow_checkpoint.py from_modelzTensorflowCheckpoint.from_modelsH"$$  4**3355666''00  6  ' ' 5 5 5""C$:H#EFFF file_pathctj|r|dst dt j}tj|}t|| }tj ||| |}|r| |||j|i|S)aTCreate a :py:class:`~ray.train.Checkpoint` that stores a Keras model from H5 format. The checkpoint generated by this method contains all the information needed. Thus no `model` is needed to be supplied when using this checkpoint. Args: file_path: The path to the .h5 file to load model from. This is the same path that is used for ``model.save(path)``. preprocessor: A fitted preprocessor to be applied before inference. Returns: A :py:class:`TensorflowCheckpoint` converted from h5 format. z.h5zAPlease supply a h5 file path to `TensorflowCheckpoint.from_h5()`.)ospathisfileendswith ValueErrorrrbasenamerrshutilcopyrrrr)rr"rrrnew_checkpoint_filers rfrom_h5zTensorflowCheckpoint.from_h5Es&w~~i((  0B0B50I0I S "$$7##I.."7H55>>@@ I2333''00  6  ' ' 5 5 5""C$:H#EFFFr!dir_pathcjtj|stdt j}tj|tj||| |}|r| || |j di|S)acCreate a :py:class:`~ray.train.Checkpoint` that stores a Keras model from SavedModel format. The checkpoint generated by this method contains all the information needed. Thus no `model` is needed to be supplied when using this checkpoint. Args: dir_path: The directory containing the saved model. This is the same directory as used by ``model.save(dir_path)``. preprocessor: A fitted preprocessor to be applied before inference. Returns: A :py:class:`TensorflowCheckpoint` converted from SavedModel format. zDPlease supply a directory to `TensorflowCheckpoint.from_saved_model`.) r$r%isdirr(rrrmdirr*copytreerrrr)rr.rrrs rfrom_saved_modelz%TensorflowCheckpoint.from_saved_modelgs&w}}X&& V "$$ '***''00  6  ' ' 5 5 5""C$:C#@AAAr!cR|}|j|vrtd||j}|5}t ||}t j|cdddS#1swxYwYdS)zRetrieve the model stored in this checkpoint. Returns: The Tensorflow Keras model stored in the checkpoint. z`TensorflowCheckpoint` cannot retrieve the model if you override the checkpoint metadata. Please use `Checkpoint.update_metadata` instead.N) get_metadatarr( as_directoryrrrmodels load_model)selfmetadatamodel_filenamecheckpoint_dir model_paths r get_modelzTensorflowCheckpoint.get_models$$&&  "( 2 2X "$"9:     7Nnn==FFHHJ<**:66 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7 7sABB #B )__name__ __module__ __qualname____doc__r classmethodrModelrr strr-r4tfr?r!rr r s'UU* 26 +++{+~. + +++[+ZIM.6~.F [BHL    -5n-E     [ D7 777777r!r )r$r*rpathlibrtypingrr tensorflowrGr(ray.train._internal.framework_checkpointrray.util.annotationsrray.data.preprocessorr r rHr!rrOs ********HHHHHH******3222222 VI7I7I7I7I7.I7I7I7I7I7r!