docs(autogram): Improve autogram package doc (#416)

Author: ValerianRey

src/torchjd/autogram/__init__.py

@@ -1,13 +1,15 @@
 """
-The autogram package enables the activation of Gramian-based Jacobian descent on your models (As
-described in Section 6 of `Jacobian Descent For Multi-Objective Optimization
-<https://arxiv.org/pdf/2406.16232>`_). It provides a convenient way to modify a model's backward
-pass, allowing you to seamlessly integrate multi-objective optimization in your PyTorch code.
+The autogram package provides an engine to efficiently compute the Gramian of the Jacobian of a
+tensor of outputs (generally losses) with respect to some modules' parameters. This Gramian contains
+all the inner products between pairs of gradients, and is thus a sufficient statistic for most
+weighting methods. The algorithm is formally defined in Section 6 of `Jacobian Descent For
+Multi-Objective Optimization <https://arxiv.org/pdf/2406.16232>`_).
 
-This method typically provides a memory improvement over the :doc:`autojac <../autojac/index>`
-package which typically leads to time improvement.
+Due to computing the Gramian iteratively over the layers, without ever having to store the full
+Jacobian in memory, this method is much more memory-efficient than
+:doc:`autojac <../autojac/index>`, which makes it often much faster.
 
-The exhaustive list of supported Weightings compatible with `autogram` is:
+The list of Weightings compatible with ``autogram`` is:
 
 * :class:`~torchjd.aggregation.UPGradWeighting`
 * :class:`~torchjd.aggregation.AlignedMTLWeighting`