Skip to content

Replicating tensorflow bert model in R #544

Open
@ozturkib

Description

@ozturkib

Stackoverflow question is here.
I am just replicating this code based on Basic Text Classification. Until the below line, it seems alright. However, the following line:

vectorize_layer %>% adapt(train_text)

throws me an error as below. Any idea, how should I approach it in order to solve it ? My tensorflow version is TensorFlow v2.8.2, is it the main issue for this error ? It is not! Because I have also recreate new environment with v2.9.1, nothing is changed.

I assume that the train_text should not be like that as well, should it ?

> train_text
<MapDataset element_spec=TensorSpec(shape=(None,), dtype=tf.string, name=None)>
The thrown error:

Error in py_call_impl(callable, dots$args, dots$keywords) : 
RuntimeError: in user code:

File "C:\ANACON~2\envs\R-TENS~1\lib\site-packages\keras\engine\base_preprocessing_layer.py", line 118, in adapt_step *
self.update_state(data)
File "C:\ANACON~2\envs\R-TENS~1\lib\site-packages\keras\layers\preprocessing\text_vectorization.py", line 431, in update_state **
self._lookup_layer.update_state(self._preprocess(data))
File "C:\ANACON~2\envs\R-TENS~1\lib\site-packages\keras\layers\preprocessing\text_vectorization.py", line 512, in _preprocess
inputs = self._standardize(inputs)
File "C:\Users\xxxxx\AppData\Local\R\win-library\4.2\reticulate\python\rpytools\call.py", line 21, in python_function
raise RuntimeError(res[kErrorKey])

RuntimeError: Evaluation error: attempt to apply non-function.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions