SONG Ge comments

Results 57 comments of


                                            SONG Ge

Use spark backend got error:NotImplementedError: Saving the model to HDF5 format requires the model to be a Functional model or a Sequential model. It does not work for subclassed models, because such models are defined via the body of a Python method, which isn't safely serializable. Consider saving to the Tensorflow SavedModel format (by setting save_format="tf") or using `save_weights`.

> > > > > > > > > > the DeepCross model is a user defined subclassed model, in tf2, it cannot save as h5 file. > > >...

[Orca] Code Design for Distributed PyTorch DataLoader on Ray

@yushan111

spark backend distributed predicting got error

Hi! Could you please provide your code for us to reproduce, since it could help us to locate the problem soon? By the way, where did you run the program,...

spark backend distributed predicting got error

Hi, I have reviewed the error report, I think it may be related to your model code or TensorFlow, which is not a bug caused by BigDL. From the error...

spark backend distributed predicting got error

> > Hi, I have reviewed the error report, I think it may be related to your model code or TensorFlow, which is not a bug caused by BigDL. >...

spark backend distributed predicting got error

Hey, would you mind providing the dataset or where could I reach it?

spark backend distributed predicting got error

> > Hey, would you mind providing the dataset or where could I reach it? > > Sorry, the dataset is my company's. There are only four int field, you...

TypeError: 'JavaPackage' object is not callable, when spark-commit xgboost_example

## Reason: Lack of BigDL jar files in your script, `JavaPackage` happens because `python` cannot find the java class. ## Solutions: ### Use `spark-submit` 1. Download bigdl dllib jars file...

[Orca] TF2 Ray Estimator Supports Callback to Save Model

> What's the status of this PR? All done, need to set up jenkins test.

Exception happened if using orca estimator train pytorch model with xshards

> @jason-dai @dding3 Maybe a more common case is stacking multiple columns into a single tensor, passing to a single input model. In this case, according to current API, users...