尝试将我的 keras 模型导出到 tensorflow 服务时出现错误

Question

from tensorflow.python.saved_model import builder as saved_model_builder
export_path=r'tensorflowservingmodel'
export_version=1
builder = saved_model_builder.SavedModelBuilder(export_path)
signature = predict_signature_def(
    inputs={'input': new_model.inputs[0]},
    outputs={'output': new_model.outputs[0]})


with K.get_session() as sess:

    builder.add_meta_graph_and_variables(
        sess=sess,
        tags=[tag_constants.SERVING],
        clear_devices = True,
        signature_def_map={
            signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY: signature}
    )
builder.save()
print("tensorflow serving.......")

AssertionError Traceback (most recent call last) in 2 export_path=r'tensorflowservingmodel' 3 export_version=1 ----> 4 builder = saved_model_builder.SavedModelBuilder(export_path) 5 signature = predict_signature_def( 6 inputs={'input': new_model.inputs[0]}, ~\Anaconda\lib\site-packages\tensorflow\python\saved_model\builder_impl.py in init(self, export_dir) 93 raise AssertionError( 94 "Export directory already exists. Please specify a different export " ---> 95 "directory: %s" % export_dir) 96 97 file_io.recursive_create_dir(self._export_dir) AssertionError: Export directory already exists. Please specify a different export directory: tensorflowservingmodel

Answer 1

删除这个目录'tensorflowservingmodel'，导出模型前确保它不存在。

rm -r tensorflowservingmodel

Answer 2

我发现了问题我已经创建了一个名为 tensorflowservingmodel 的文件，这就是为什么程序说目录已经 exist.l 必须删除文件目录并重新运行代码。

尝试将我的 keras 模型导出到 tensorflow 服务时出现错误

I am getting an error when trying to export my keras model to tensorflow serving

python

keras

tensorflow

tensorflow-serving