diff --git a/doc/src/SUMMARY.md b/doc/src/SUMMARY.md
index 9fbf540..ca5f987 100644
--- a/doc/src/SUMMARY.md
+++ b/doc/src/SUMMARY.md
@@ -1,6 +1,7 @@
 # Summary
 
 [Introduction](README.md)
+[Installation](intall.md)
 
 - [Schlandals Modelization](modelization/README.md)
     - [The Language](modelization/language.md)
diff --git a/doc/src/install.md b/doc/src/install.md
new file mode 100644
index 0000000..1af293b
--- /dev/null
+++ b/doc/src/install.md
@@ -0,0 +1,37 @@
+# Installation
+
+In order to use Schlandals, you must have the [Rust toolchain installed](https://www.rust-lang.org/tools/install).
+Once this is done, you can install Schlandals either using Cargo or from sources.
+
+## Installing using Cargo
+
+Run the following command
+```bash
+cargo install schlandals
+```
+
+it will install locally the executable. In Unix-based system you should find the executable in `~/.cargo/bin/schlandals`.
+
+## Installing from sources
+
+```bash
+git clone git@github.com:aia-uclouvain/schlandals.git && cd schlandals && cargo build --release
+```
+
+it will compile the solver, from sources, and place the executable inside `schlandals/target/release/schlandals`.
+
+## Optional features
+
+### Learning with Torch
+
+It is possible to use the learning module of Schlandals with torch tensors. The torch bindings are provided by the [tch-rs crate](https://github.com/LaurentMazare/tch-rs).
+It means that for using this feature, you must libtorch installed on your system. We refer to the [documentation of tch-rs](https://github.com/LaurentMazare/tch-rs?tab=readme-ov-file#getting-started) for the set-up of torch and tch-rs.
+
+Once torch is installed, and the appropriate variables set, you can run either
+```bash
+cargo install schlandals --features tensor
+```
+or
+```bash
+cargo build --release --features tensor
+```
diff --git a/doc/src/modelization/bn.md b/doc/src/modelization/bn.md
index d31d9e6..469d9d7 100644
--- a/doc/src/modelization/bn.md
+++ b/doc/src/modelization/bn.md
@@ -38,7 +38,7 @@ The encoding (in DIMACS-style format) is show below for the query \\( P(D = d_0)
 Notice that the probailistic variables start at 1 and are in contiguous blocks.
 The 
 
-```bash
+```
 p cnf 26 19
 c Probabilistic variables 1 2
 c p distribution 0.2 0.8
diff --git a/doc/src/modelization/pg.md b/doc/src/modelization/pg.md
index 8eb2dae..a690a96 100644
--- a/doc/src/modelization/pg.md
+++ b/doc/src/modelization/pg.md
@@ -25,7 +25,7 @@ The query is encoded by imposing the fact that the source \\( s \\) is reachable
 
 The encoding (in DIMACS-style format) is show below for the query \\( P(A \text{ connected to } E) \\).
 
-```bash
+```
 p cnf 26 19
 c Edge from A to B with variables 1 2
 c p distribution 0.4 0.6