05db715004e2aa699935fc3ec4ba5d0dd101a365,python/mead/tasks.py,EncoderDecoderTask,_create_backend,#EncoderDecoderTask#,414

Before Change

        return reader

    def _create_backend(self):
        backend = super(EncoderDecoderTask, self)._create_backend()
        if backend.name == "pytorch":
            import baseline.pytorch.seq2seq as seq2seq
            self.config_params["preproc"]["show_ex"] = baseline.pytorch.show_examples_pytorch
            self.config_params["preproc"]["trim"] = True

After Change

        return reader

    def _create_backend(self):
        backend = Backend()
        backend.name = self.config_params.get("backend", "tensorflow")
        if backend.name == "pytorch":
            import baseline.pytorch.embeddings as embeddings
            import baseline.pytorch.seq2seq as seq2seq
            self.config_params["preproc"]["show_ex"] = baseline.pytorch.show_examples_pytorch
            self.config_params["preproc"]["trim"] = True
        else:
            // TODO: why not support DyNet trimming?
            self.config_params["preproc"]["trim"] = False
            if backend.name == "dynet":
                import _dynet
                import _dynet
                dy_params = _dynet.DynetParams()
                dy_params.from_args()
                dy_params.set_requested_gpus(1)
                if "autobatchsz" in self.config_params["train"]:
                    dy_params.set_autobatch(True)
                    batched = False
                else:
                    batched = True
                dy_params.init()
                backend.params = {"pc": _dynet.ParameterCollection(), "batched": batched}
                import baseline.dy.embeddings as embeddings
                import baseline.dy.seq2seq as seq2seq
                self.config_params["preproc"]["show_ex"] = baseline.dy.show_examples_dynet
            else:
                import baseline.tf.embeddings as embeddings
                import baseline.tf.seq2seq as seq2seq
                self.config_params["preproc"]["show_ex"] = baseline.tf.create_show_examples_tf(self.primary_key)
                from mead.tf.exporters import Seq2SeqTensorFlowExporter
                backend.exporter = Seq2SeqTensorFlowExporter

        backend.embeddings = embeddings
        backend.task = seq2seq
        return backend

    def initialize(self, embeddings):

Instances