search

GAMES-UChile / mogptk

Multi-Output Gaussian Process Toolkit
MIT License
164 stars 43 forks source link

Cannot initialize model #69

Closed luca-brambilla closed 10 months ago

luca-brambilla commented 11 months ago

Hello!

I was running the examples found in the repository and I keep finding the same error when trying to initialize a model. I noticed you recently updated the library to version 0.4.0, but I get no errors for 0.3.5.

For example, in 00_Quick_Start:

model = mogptk.CONV(data, Q=3)

output:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-81-e7b80f8f700a> in <cell line: 1>()
----> 1 model = mogptk.CONV(data, Q=3)

/usr/local/lib/python3.10/dist-packages/mogptk/models/conv.py in __init__(self, dataset, Q, inference, mean, name)
     55             kernel[q].base_variance.assign(np.random.rand(input_dims))
     56 
---> 57         super().__init__(dataset, kernel, inference, mean, name)
     58         self.Q = Q
     59 

TypeError: Model.__init__() takes from 3 to 5 positional arguments but 6 were given

Thank you!

tdewolff commented 11 months ago

Hi Luca, thank you for raising this issue. There are various bugs with 0.4.0 and I've fixed most of them in the master branch (you could downgrade, or use git clone https://github.com/GAMES-UChile/mogptk and then inside pip uninstall mogptk and then pip install -e .). I expect to release a new version in a few hours with all the fixes. Additionally, I'm working on your other issue to add seeds to all examples.

tdewolff commented 10 months ago

Please checkout v0.5.0, that should fix your issues! You can re-open or raise a new issue if you encounter other problems, I'm happy to help