Bryan Whiting

This is a substantive issue that bad inputs don't yield any errors. ``` predictor = TimeSeriesPredictor( prediction_length=21, freq='h', target = 'x', eval_metric="MAE", quantile_levels = {}, ) predictor.fit( train_data=train_data, hyperparameters={ "DeepAR":...

In other cases, though, I do see a warning: Follow-up (to the https://github.com/autogluon/autogluon/discussions/2968 issue), but I pass in ` {"tabular_hyperparameters": {"lkjalkjlakdj":{}}}` to see what it does: ``` predictor.fit( train_data=train_data, hyperparameters={...

I think you've nailed it! The second case is tricky, but I support your implementation. In some markdown languages, two lines with only one line break and no spave in...

But ultimately, design it how you want!!! I'd just be an appreciative user. Thank you!

Thanks! your team rocks. keep it up!

Here's my current workaround which trains different models: ``` predictor.fit( train_data=train_data, hyperparameters={ "DeepAR": {}, "RecursiveTabular": [ {"tabular_hyperparameters": {"GBM": {}}}, {"tabular_hyperparameters": {"GBM": { "num_leaves": 5}}}, {"tabular_hyperparameters": {"GBM": { "num_leaves": 50}}} ],...

Thanks for your detailed reply! I confirmed the `space.Categorical()` approach worked. This is better than my "hacky" approach because this would let me tune both a `RecursiveTabular` as well as...

Solution should just be: ``` return smape(df, models=['model'], id_col=id_col, target_col=target_col)['model'].mean() ``` in the `auto.ipynb` notebook. If this is the right design, i'll submit a PR

+1 to this being an amazing idea!

+1 excited for this!