There is a lot of renamed files, so please ignore them.
You can look at the second commit as it's the one containing the notebook with the new experiment on MLP experiments/mlp.ipynb
It's obvious that the MLP model is following PatchCore a lot better than the LR model. However, it's still not producing good accuracy.
I'm working now on changing the model configuration to other variations to see the effect.
It was suggested online as well that I scale the MLP input data, so I'll try that as well, and update this PR (if the results were good).
closes #9
There is a lot of renamed files, so please ignore them. You can look at the second commit as it's the one containing the notebook with the new experiment on MLP
experiments/mlp.ipynbIt's obvious that the MLP model is following PatchCore a lot better than the LR model. However, it's still not producing good accuracy.
I'm working now on changing the model configuration to other variations to see the effect. It was suggested online as well that I scale the MLP input data, so I'll try that as well, and update this PR (if the results were good).