Update README.md

benchmark/README.md

@@ -1,6 +1,6 @@
 We include source code and wrapper scripts for [FigTree](https://github.com/vmorariu/figtree) and [ASKIT](http://padas.ices.utexas.edu/libaskit/) for comparision, and provide basic instructions on running the libraries on example datasets below. For details, please refer to the corresponding documentation of the libraries.
 
-# FigTree
+## FigTree
 To build FigTree from source
 ```sh
 ~/rehashing/benchmark/figtree $ make
@@ -17,13 +17,14 @@ The smaller epsilon is, the more accurate the results will be, at the expense of
 
 FigTree contains an optimizer that chooses between four evaluation strategies: ```FIGTREE_EVAL_DIRECT```, ```FIGTREE_EVAL_DIRECT_TREE```, ```FIGTREE_EVAL_IFGT```, and ```FIGTREE_EVAL_IFGT_TREE```. FigTree only works with Gaussian kernels.
 
-# ASKIT
-##### Notes
+## ASKIT
+#### Notes
 The compiler flags are currently set for Intel compilers. A non-exhaustive dependency list of ASKIT include: Intel MKL, MPI, OpenMP. Input datasets to ASKIT must be space-seperated and contain no headers. To measure performance on a single core, set number of threads to 1 by ```export OMP_NUM_THREADS=1```.
-##### Build
+#### Build
 To use ASKIT, we first need to build the RKDT (Randomized KD-trees) library in ```askit_release/rkdtsrc``` (see ```askit_release/rkdtsrc/README``` for instructions). If everything builds succesfully, you should be able to see ```libcmd.a  libknn.a  librrot.a``` under ```askit_release/rkdtsrc/build```, and ```test_find_knn.exe``` under ```askit_release/rkdtsrc/parallelIO```.
+
 After building RKDT, we can proceed to build the ASKIT library in ```askit_release/treecode``` (see ```askit_release/treecode/README``` for instructions). If everything builds succesfully, you should be able to see ```libaskit.a``` under ```askit_release/treecode/build```, and ```test_askit_main.exe``` under ```askit_release/treecode/src```.
-##### Example Usage
+#### Example Usage
 Here we provide an example running ASKIT on the covtype dataset.
 First, we generate input files to ASKIT using the helper script: ```python gen_askit_inputs.py [dataset name] [dataset] [exact file]```. The exact file should contain query density and index of the query point in the dataset in each line.
 ```sh