NotImplementedError While Loading Saved .nc Model

[tim-mcwilliams]

Hi all!!

I've been exploring using the budget allocation methodology and have come across an interesting error. Hoping to get some guidance or find anyone who has came across this before.

After building a model, I've saved it like so - my_mmm.save('some_cool_model.nc'). That file gets saved to my working dir for later use. Once budget allocation is starting to kick off, we would load that model - my_cool_model = MMM.load('some_cool_model.nc').

Upon loading that cool model up, I'm getting this error

NotImplementedError: At the moment, convolving with weight arrays that don't have a concrete shape at compile time is not supported.
Traceback:
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/streamlit/runtime/scriptrunner/exec_code.py", line 88, in exec_func_with_error_handling
    result = func()
             ^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 579, in code_to_exec
    exec(code, module.__dict__)
File "/opt/rstudio-connect/mnt/app/pages/4_Budget_Allocator.py", line 28, in <module>
    model_load = MMM.load(model_path)
                 ^^^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/model_builder.py", line 527, in load
    model.build_model(X, y)
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/mmm.py", line 475, in build_model
    var=self.forward_pass(x=channel_data_),
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/mmm.py", line 337, in forward_pass
    return second.apply(x=first.apply(x=x, dims="channel"), dims="channel")
                          ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/components/base.py", line 556, in apply
    return self.function(x, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/components/adstock.py", line 189, in function
    return geometric_adstock(
           ^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/transformers.py", line 245, in geometric_adstock
    return batched_convolution(x, w, axis=axis, mode=mode)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/rstudio-connect/mnt/app/python/env/lib/python3.12/site-packages/pymc_marketing/mmm/transformers.py", line 131, in batched_convolution
    raise NotImplementedError(

My env:

python = "3.12.3"
pymc_marketing = "0.9.0"
numpy = "1.26.4"
arviz = "0.17.1"
pymc = "5.15.0"
pytensor = "2.20.0"
scipy = "1.12.0"