Javier Arturo Porras Luraschi

@kevinykuo Not sure if I used the right term here, I meant `canned` as the non-custom ones, in this case, the example uses `linear_regressor`, would this estimator hit `canned/head.py`?

@jjallaire / @kevinykuo alright, I finally understand what is going on, at least for `LinearRegressor` which is what I origanlly tried: See: https://github.com/tensorflow/tensorflow/blob/9bdb72e124e50e1b12b3286b38cbb1c971552741/tensorflow/python/estimator/canned/head.py#L654-L664 ```python classifier_output = _classification_output( scores=probabilities, n_classes=self._n_classes, label_vocabulary=self._label_vocabulary)...

@jjallaire Ah, right! Alright, so here is what is going on... Back to the reference mentioned above, an estimator has 3 signatures: ```python return model_fn.EstimatorSpec( mode=model_fn.ModeKeys.PREDICT, predictions=predictions, export_outputs={ _DEFAULT_SERVING_KEY: classifier_output,...

The problem is that under `export_savedmodel`, we create a placeholder which seems to affect subsequent use of the TF session...

See https://github.com/rstudio/tfestimators/compare/bugfix/export-savedmodel-default?expand=1

Yes, you probably need to wrap the entries under a `list()` since `predict_savedmodel()` expects a list of entries to predict. For example, first train a model as specified in the...

@trotsiuk @rcitrone @hrbrmstr I also want to get a WARC parser going for R, mostly to use it with Apache Spark. I have a draft extension here https://github.com/javierluraschi/sparkwarc which is...

@rsilvery it's possible that the timeout is different for the container, it's also possible that the container has fewer resources available and the operation takes longer. I would also try...

@rsilvery my bad, verbose is enabled with: ```r options(sparklyr.verbose = TRUE) ```