Simplify parameter explanation

- The goal here is to uniformize a bit with the rest of the library: it's not really needed to indicate default values, because they already appear in the built documentation (unless the default is some None that later gets transformed into something else, but it's not the case here). Also, I don't think it's needed to indicate that these parameters can be changed afterwards, because I don't think a lot of people will do that, and it's actually the case of the parameters of all aggregators (or it should be). Lastly, I made the description be the same between aggregator and weighting (for ease of maintainance).

Author: ValerianRey

src/torchjd/aggregation/_gradvac.py

@@ -31,11 +31,8 @@ class GradVac(GramianWeightedAggregator):
     This aggregator is stateful: it keeps :math:`\hat{\phi}` across calls. Use :meth:`reset` when
     the number of tasks or dtype changes.
 
-    :param beta: EMA decay for :math:`\hat{\phi}` (paper default ``0.5``). You may read or assign
-        the :attr:`beta` attribute between steps to tune the EMA update.
-    :param eps: Small non-negative constant added to denominators when computing cosines and the
-        vaccine weight (default ``1e-8``); set to ``0`` to omit this stabilization. You may read or
-        assign the :attr:`eps` attribute between steps to tune numerical behavior.
+    :param beta: EMA decay for :math:`\hat{\phi}`.
+    :param eps: Small non-negative constant added to denominators.
 
     .. note::
         For each task :math:`i`, the order of other tasks :math:`j` is shuffled independently
@@ -95,8 +92,8 @@ class GradVacWeighting(Weighting[PSDMatrix]):
     This weighting is stateful: it keeps :math:`\hat{\phi}` across calls. Use :meth:`reset` when
     the number of tasks or dtype changes.
 
-    :param beta: EMA decay for :math:`\hat{\phi}` (paper default ``0.5``).
-    :param eps: Small non-negative constant added to denominators (default ``1e-8``).
+    :param beta: EMA decay for :math:`\hat{\phi}`.
+    :param eps: Small non-negative constant added to denominators.
     """
 
     def __init__(self, beta: float = 0.5, eps: float = 1e-8) -> None: