Args: model_path (str): a model name (model-xxxx) or a ``checkpoint`` file. prefix (str): during restore, add a ``prefix/`` for every variable in this checkpoint. ignore (tuple[str]): tensor names that should be ignored during loading, e.g. learning-rate
(self, model_path, prefix=None, ignore=())
| 92 | Restore a tensorflow checkpoint saved by :class:`tf.train.Saver` or :class:`ModelSaver`. |
| 93 | """ |
| 94 | def __init__(self, model_path, prefix=None, ignore=()): |
| 95 | """ |
| 96 | Args: |
| 97 | model_path (str): a model name (model-xxxx) or a ``checkpoint`` file. |
| 98 | prefix (str): during restore, add a ``prefix/`` for every variable in this checkpoint. |
| 99 | ignore (tuple[str]): tensor names that should be ignored during loading, e.g. learning-rate |
| 100 | """ |
| 101 | if model_path.endswith('.npy') or model_path.endswith('.npz'): |
| 102 | logger.warn("SaverRestore expect a TF checkpoint, but got a model path '{}'.".format(model_path) + |
| 103 | " To load from a dict, use 'DictRestore'.") |
| 104 | model_path = get_checkpoint_path(model_path) |
| 105 | self.path = model_path # attribute used by AutoResumeTrainConfig! |
| 106 | self.prefix = prefix |
| 107 | self.ignore = [i if i.endswith(':0') else i + ':0' for i in ignore] |
| 108 | |
| 109 | def _setup_graph(self): |
| 110 | dic = self._get_restore_dict() |
nothing calls this directly
no test coverage detected