Adjust dataset card

README.md

@@ -117,7 +117,8 @@ configs:
 ---
 
 # Dataset Card for FIM-Estimation
-## Dataset Summary
+## Dataset Description
+### Dataset Summary
 In a FIM-Estimation task, the input consists of the following components:
 
 1. A dataset of the form $\mathcal{D}_ {train} = \{(\mathbf{\lambda}_ i, x_ i)\}_ {i=1}^{\left|\mathcal{D}\right|}$, where $\vlambda_ i$ is a point in the parameter space and $x_ i$ is a sample.
@@ -128,7 +129,7 @@ The task is to estimate Fisher Information Metric (FIM) as a matrix-valued funct
 
 In this dataset we present multiple such tasks (parameterised by task name and random seed). E.g. in order to solve the IsNNN400 task with seed=4 one would download config `isnnn_400.seed04`, split `train`, estimate the FIM, and compare the results with `isnnn_400.seed04.gt_fim`.
 
-## Additional information
+### Additional information
 We (TODO:plan to) present datasets corresponding to 6 statistical manifolds: `XXZ300_Z`, `FIL24`, etc.
 The data is stored as follows:
 `data/xxz300_z` contains the data for `XXZ300_Z` datasets. Within that directory, there are subdirectories of the form `seed_??` each containing a dataset describing the same statistical manifold, but generated using a different seed (e.g. `seed_05`). We have `seed_05/d_train.parquet`, which is the dataset as described in #1 above. `seed_05/d_test.parquet` is a hold-out dataset which should not be used in estimation of $\mathbf{\lambda}$ (not even for hyperparameter tuning).