Continuation of issue #2474 as discussed here

Add default parameters for all projection heads

It's helpful to know what the default parameters were in the papers to get started. We should add the default projection head parameters which were used for pre-training on Imagenet to all projection and prediction heads in lightly/models/modules/heads.py.

The default RubrixLogHTTPMiddleware record mapper for token classification expect a structured including a text field for inputs. This could make prediction model inputs a bit cumbersome. Default mapper could accepts also flat strings as inputs:

def token_classification_mapper(inputs, outputs):
    i

When computing the isosurface, when 4 simplex edges crosses the isosurface, we add 2 triangles to the isosurface triangulation.

Right now we add these arbitrarily (in these 3 lines) but we should add them so that the curvature is minimized.

The current state o

We should swap most of our metrics for torchmetrics this would make our code more robust.

Only ECE is special I think and we should keep our own implementation.

In simulation mode, when I try to run a single simulation using the nn-2-layer classifier and sbert feature extraction method, it gives me this error

ValueError: Error: method 'nn2layer' is not implemented for entry point asreview.models.classifiers.

However, using this combination of classification and feature extraction method in a simulation-batch does not give me the same error. Using th

Some query strategies require the probabilities for all labels of a sample, currently only probabilities for successfully predicted labels are returned.