Closed paxcema closed 9 months ago
Instructions for replicating can be found here, but do note you need to pass the api_key
param at ML engine creation.
FYI @tomhuds this is important for enabling predict-time dynamic arguments in TimeGPT's current design. Users can still define any parameter they wish at creation time, though.
We should clarify with Andrey how big a task this is (looked at the code base just now, and I suspect it is not trivial) before commiting to it.
It's probably best to tackle this via the new syntax anyway.
+1 for fixing alongside the new CREATE FORECASTING MODEL refactor
fixed in #293
The
USING
parameters are failing to be included in the plan for queries like this one:The expected behavior is to find
validate_token = True
in the prediction parameters field.