From adb9fa4ae2757104bb4e05bf7e628490ae677c2f Mon Sep 17 00:00:00 2001
From: Rounak Datta <rounakdatta12@gmail.com>
Date: Thu, 11 Apr 2024 12:50:07 +0530
Subject: [PATCH] Add additional tests for neuron module

---
 lib/neuron.ml             | 14 +++++++++++++-
 lib/neuron.mli            | 10 ++++++++++
 test/neuron_operations.ml | 25 +++++++++++++++++++++++++
 test/smolgrad_tests.ml    |  3 +++
 4 files changed, 51 insertions(+), 1 deletion(-)
 create mode 100644 test/neuron_operations.ml

diff --git a/lib/neuron.ml b/lib/neuron.ml
index a9c3e65..e4f21aa 100644
--- a/lib/neuron.ml
+++ b/lib/neuron.ml
@@ -8,13 +8,25 @@ module Neuron = struct
     is_non_linear : bool;
   }
 
+  (* this is just a getter data structure for the parameters *)
+  type out_t = {
+      weights : Variable.Variable.t list;
+      bias : Variable.Variable.t;
+  }
+
+  (* note how we explicitly define the types to avoid confusions and fights with the compiler *)
+  let parameters (neuron: t) : out_t = {
+    weights = neuron.weights;
+    bias = neuron.bias;
+  }
+
   let create number_of_inputs is_non_linear = {
     weights = List.init number_of_inputs (fun _ -> Variable.Variable.create (random_weight_initializer));
     bias = Variable.Variable.create 0.0;
     is_non_linear = is_non_linear;
   }
 
-  let weigh_inputs neuron input_vector =
+  let weigh_inputs (neuron: t) input_vector =
     (* one-to-one multiplication of inputs to their corresponding weights *)
     let weighted_sum = List.fold_left2 (fun accumulator weight_i input_i -> Variable.Variable.(accumulator + weight_i * input_i))
       (Variable.Variable.create 0.0) neuron.weights input_vector in
diff --git a/lib/neuron.mli b/lib/neuron.mli
index 7df98f2..223be7c 100644
--- a/lib/neuron.mli
+++ b/lib/neuron.mli
@@ -2,6 +2,16 @@
 module Neuron : sig
     type t
 
+    (* we've defined this public type both here in the interface file as well as the implementation file *)
+    (* this is required for the internal attributes of the type to be accessible publicly *)
+    type out_t = {
+        weights : Variable.Variable.t list;
+        bias : Variable.Variable.t;
+    }
+
+    (* getter for the parameters (aka weights and biases) of the neuron *)
+    val parameters : t -> out_t 
+
     (* Constructor; constructs a unit neuron *)
     val create : int -> bool -> t
 
diff --git a/test/neuron_operations.ml b/test/neuron_operations.ml
new file mode 100644
index 0000000..bd8ce1c
--- /dev/null
+++ b/test/neuron_operations.ml
@@ -0,0 +1,25 @@
+open Smolgrad.Neuron
+
+let test_neuron_initialization () =
+  let n = Neuron.create 5 true in
+  let n_weights = (Neuron.parameters n).weights in
+  assert (Smolgrad.Variable.Variable.data (Neuron.parameters n).bias = 0.0);
+
+  Alcotest.(check int)
+    "Neuron: Right number of parameters are set"
+    5
+    (List.length (Neuron.parameters n).weights);
+
+  Alcotest.(check (float 0.0))
+    "Neuron: Bias is initially set to 0.0"
+    0.0
+    (Smolgrad.Variable.Variable.data (Neuron.parameters n).bias);
+
+  let are_weights_in_range weights =
+    List.for_all (fun x -> (Smolgrad.Variable.Variable.data x) >= -1.0 && (Smolgrad.Variable.Variable.data x) <= 1.0) weights in
+
+  Alcotest.(check bool)
+    "Neuron: All weights are in range [-1, 1]"
+    true
+    (are_weights_in_range n_weights);
+;;
diff --git a/test/smolgrad_tests.ml b/test/smolgrad_tests.ml
index fbded76..38cbabc 100644
--- a/test/smolgrad_tests.ml
+++ b/test/smolgrad_tests.ml
@@ -4,4 +4,7 @@ let () =
   Variable_operations.test_custom_operator ();
   Variable_operations.test_graph_construction ();
   Variable_operations.test_backpropagation ();
+
+  (* unit tests for Neuron *)
+  Neuron_operations.test_neuron_initialization ();
 ;;