From 0734d891ef1b532f060c805f8a1c703615d8daea Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Fri, 23 May 2025 23:32:41 -0400
Subject: [PATCH 01/39] changed policy and rename factors to match common names

---
 pufferlib/config/ocean/boids.ini | 11 ++++++-----
 pufferlib/ocean/boids/binding.c  |  6 +++---
 pufferlib/ocean/boids/boids.h    | 22 +++++++++++-----------
 pufferlib/ocean/boids/boids.py   | 12 ++++++------
 4 files changed, 26 insertions(+), 25 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index 2f8412d248..0e6c8d33e6 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -1,7 +1,8 @@
 [base]
 package = ocean
 env_name = puffer_boids 
-policy_name = Boids
+; policy_name = Boids
+policy_name = Policy
 rnn_name = Recurrent
 ; rnn_name = None
 
@@ -10,10 +11,10 @@ num_envs = 64
 num_boids = 64
 ; num_envs = 1
 ; num_boids = 1
-margin_turn_factor = 0.0
-centering_factor = 0.00
-avoid_factor = 1.00
-matching_factor = 1.00
+margin_turn_factor = 1.0
+cohesion_factor = 0.0
+seperation_factor = 0.0
+alignment_factor = 0.0
 
 [vec]
 num_workers = 2
diff --git a/pufferlib/ocean/boids/binding.c b/pufferlib/ocean/boids/binding.c
index a3483d6520..27b40defb4 100644
--- a/pufferlib/ocean/boids/binding.c
+++ b/pufferlib/ocean/boids/binding.c
@@ -7,9 +7,9 @@ static int my_init(Env* env, PyObject* args, PyObject* kwargs) {
     env->num_boids = unpack(kwargs, "num_boids");
     env->report_interval = unpack(kwargs, "report_interval");
     env->margin_turn_factor = unpack(kwargs, "margin_turn_factor");
-    env->centering_factor = unpack(kwargs, "centering_factor");
-    env->avoid_factor = unpack(kwargs, "avoid_factor");
-    env->matching_factor = unpack(kwargs, "matching_factor");
+    env->cohesion_factor = unpack(kwargs, "cohesion_factor");
+    env->seperation_factor = unpack(kwargs, "seperation_factor");
+    env->alignment_factor = unpack(kwargs, "alignment_factor");
     init(env);
     return 0;
 }
diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index bf2bf6331e..5828b57169 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -52,9 +52,9 @@ typedef struct {
     Boid* boids;
     unsigned int num_boids;
     float margin_turn_factor;
-    float centering_factor;
-    float avoid_factor;
-    float matching_factor;
+    float cohesion_factor;
+    float seperation_factor;
+    float alignment_factor;
     unsigned tick;
     Log log;
     Log* boid_logs;
@@ -134,8 +134,8 @@ void c_step(Boids *env) {
             current_boid->velocity.x = flclip(current_boid->velocity.x + (mouse_x - current_boid->x), -VELOCITY_CAP, VELOCITY_CAP);
             current_boid->velocity.y = flclip(current_boid->velocity.y + (mouse_y - current_boid->y), -VELOCITY_CAP, VELOCITY_CAP);
         } else {
-            current_boid->velocity.x = flclip(current_boid->velocity.x + 2*env->actions[current_indx * 2 + 0], -VELOCITY_CAP, VELOCITY_CAP);
-            current_boid->velocity.y = flclip(current_boid->velocity.y + 2*env->actions[current_indx * 2 + 1], -VELOCITY_CAP, VELOCITY_CAP);
+            current_boid->velocity.x = flclip(current_boid->velocity.x + env->actions[current_indx * 2 + 0], -VELOCITY_CAP, VELOCITY_CAP);
+            current_boid->velocity.y = flclip(current_boid->velocity.y + env->actions[current_indx * 2 + 1], -VELOCITY_CAP, VELOCITY_CAP);
         }
         current_boid->x = flclip(current_boid->x + current_boid->velocity.x, 0, WIDTH  - BOID_WIDTH);
         current_boid->y = flclip(current_boid->y + current_boid->velocity.y, 0, HEIGHT - BOID_HEIGHT);
@@ -161,8 +161,8 @@ void c_step(Boids *env) {
             }
         }
         if (protected_count > 0) {
-            //current_boid_reward -= fabsf(protected_dist_sum / protected_count) * env->avoid_factor;
-            current_boid_reward -= flclip(protected_count/5.0, 0.0f, 1.0f) * env->avoid_factor;
+            //current_boid_reward -= fabsf(protected_dist_sum / protected_count) * env->seperation_factor;
+            current_boid_reward -= flclip(protected_count/5.0, 0.0f, 1.0f) * env->seperation_factor;
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;
@@ -170,10 +170,10 @@ void c_step(Boids *env) {
             vis_vx_avg = vis_vx_sum / visual_count;
             vis_vy_avg = vis_vy_sum / visual_count;
 
-            current_boid_reward -= fabsf(vis_vx_avg - current_boid->velocity.x) * env->matching_factor;
-            current_boid_reward -= fabsf(vis_vy_avg - current_boid->velocity.y) * env->matching_factor;
-            current_boid_reward -= fabsf(vis_x_avg  - current_boid->x) * env->centering_factor;
-            current_boid_reward -= fabsf(vis_y_avg  - current_boid->y) * env->centering_factor;
+            current_boid_reward -= fabsf(vis_vx_avg - current_boid->velocity.x) * env->alignment_factor;
+            current_boid_reward -= fabsf(vis_vy_avg - current_boid->velocity.y) * env->alignment_factor;
+            current_boid_reward -= fabsf(vis_x_avg  - current_boid->x) * env->cohesion_factor;
+            current_boid_reward -= fabsf(vis_y_avg  - current_boid->y) * env->cohesion_factor;
         }
         if (current_boid->y < TOP_MARGIN || current_boid->y > HEIGHT - BOTTOM_MARGIN) {
             current_boid_reward -= env->margin_turn_factor;
diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index 329b36dea3..b386449b99 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -19,9 +19,9 @@ def __init__(
         report_interval=1,
         num_boids=1,
         margin_turn_factor=1.0,
-        centering_factor=0.0,
-        avoid_factor=0.0,
-        matching_factor=0.0
+        cohesion_factor=0.0,
+        seperation_factor=0.0,
+        alignment_factor=0.0
     ):
         ACTION_SPACE_SIZE = 2
         self.num_agents = num_envs * num_boids
@@ -59,9 +59,9 @@ def __init__(
                 num_boids=num_boids,
                 report_interval=self.report_interval,
                 margin_turn_factor=margin_turn_factor,
-                centering_factor=centering_factor,
-                avoid_factor=avoid_factor,
-                matching_factor=matching_factor,
+                cohesion_factor=cohesion_factor,
+                seperation_factor=seperation_factor,
+                alignment_factor=alignment_factor,
             ))
         
         self.c_envs = binding.vectorize(*c_envs)

From e94a4aae43e0568d1e284bea95b0b95da65ecf4a Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sat, 24 May 2025 13:35:15 -0400
Subject: [PATCH 02/39] remove unused log fields

---
 pufferlib/ocean/boids/binding.c |  3 ---
 pufferlib/ocean/boids/boids.h   | 14 +++-----------
 2 files changed, 3 insertions(+), 14 deletions(-)

diff --git a/pufferlib/ocean/boids/binding.c b/pufferlib/ocean/boids/binding.c
index 27b40defb4..89cb389759 100644
--- a/pufferlib/ocean/boids/binding.c
+++ b/pufferlib/ocean/boids/binding.c
@@ -15,10 +15,7 @@ static int my_init(Env* env, PyObject* args, PyObject* kwargs) {
 }
 
 static int my_log(PyObject* dict, Log* log) {
-    assign_to_dict(dict, "perf", log->perf);
     assign_to_dict(dict, "score", log->score);
-    assign_to_dict(dict, "episode_return", log->episode_return);
-    assign_to_dict(dict, "episode_length", log->episode_length);
     assign_to_dict(dict, "n", log->n);
     return 0;
 }
diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 5828b57169..1feb6be255 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -22,10 +22,7 @@
 #define BOID_TEXTURE_PATH "./resources/puffers_128.png"
 
 typedef struct {
-    float perf;
     float score;
-    float episode_return;
-    float episode_length;
     float n;
 } Log;
 
@@ -60,7 +57,6 @@ typedef struct {
     Log* boid_logs;
     unsigned report_interval;
     Client* client;
-
 } Boids;
 
 static inline float flmax(float a, float b) { return a > b ? a : b; }
@@ -191,13 +187,9 @@ void c_step(Boids *env) {
         env->rewards[current_indx] = current_boid_reward / 2.0f;
 
         //log updates
-        if (env->tick == env->report_interval) {
-            env->log.score          += env->rewards[current_indx];
-            env->log.n              += 1.0f;
-
-            /* clear per-boid log for next episode */
-            // env->boid_logs[boid_indx] = (Log){0};
-            env->tick = 0;
+        if (env->tick % env->report_interval == 0) {
+            env->log.score += env->rewards[current_indx];
+            env->log.n += 1.0f;
         }
     }
     //env->log.score /= env->num_boids;

From 817dc14f08bf0d625e574d4235690d6edc1feb98 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sat, 24 May 2025 13:35:29 -0400
Subject: [PATCH 03/39] remove unused variable

---
 pufferlib/ocean/boids/boids.h | 2 --
 1 file changed, 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 1feb6be255..cc8bfeced7 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -88,8 +88,6 @@ void init(Boids *env) {
 
 
 static void compute_observations(Boids *env) {
-    unsigned base_indx;
-
     int idx = 0;
     for (unsigned i=0; i<env->num_boids; i++) {
         for (unsigned j=0; j<env->num_boids; j++) {

From e232d3e2b1670c1c046e2e5f354071e44f02af0e Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sat, 24 May 2025 16:08:29 -0400
Subject: [PATCH 04/39] remove unused commented code

---
 pufferlib/ocean/boids/boids.py | 8 --------
 1 file changed, 8 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index b386449b99..96784b7bd9 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -5,7 +5,6 @@
 
 import numpy as np
 import gymnasium
-
 import pufferlib
 from pufferlib.ocean.boids import binding
 
@@ -74,12 +73,6 @@ def reset(self, seed=0):
     def step(self, actions):
         # Clip actions to valid range
         clipped_actions = (actions.astype(np.float32) - 2.0) / 4.0
-        #clipped_actions = np.clip(actions, -1.0, 1.0)
-        
-        # Copy the clipped actions to our flat actions buffer for C binding
-        # Flatten from [num_agents, num_boids, 2] to a 1D array for C
-        # TODO: Check if I even need this? its not like I'm using the actions anywhere else
-        #self.flat_actions[:] = clipped_actions.reshape(-1)
         
         # Save the original actions for the experience buffer
         # TODO: Same thing with this
@@ -94,7 +87,6 @@ def step(self, actions):
             if log_data:
                 info.append(log_data)
 
-        # print(f"OBSERVATIONS: {self.observations}")
         return (self.observations, self.rewards,
             self.terminals, self.truncations, info)
 

From 4eb410acae0de76ef6c3407115f6079c43f72c7a Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 28 May 2025 23:33:31 -0400
Subject: [PATCH 05/39] remove unused boid_logs and fix logs calculation

---
 pufferlib/ocean/boids/boids.h | 20 +++++++++-----------
 1 file changed, 9 insertions(+), 11 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index cc8bfeced7..f815036114 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -54,7 +54,6 @@ typedef struct {
     float alignment_factor;
     unsigned tick;
     Log log;
-    Log* boid_logs;
     unsigned report_interval;
     Client* client;
 } Boids;
@@ -69,12 +68,10 @@ static void respawn_boid(Boids *env, unsigned int i) {
     env->boids[i].y = rndf(BOTTOM_MARGIN, HEIGHT - TOP_MARGIN);
     env->boids[i].velocity.x = 0;
     env->boids[i].velocity.y = 0;
-    env->boid_logs[i]       = (Log){0};
 }
 
 void init(Boids *env) {
     env->boids = (Boid*)calloc(env->num_boids, sizeof(Boid));
-    env->boid_logs = (Log*)calloc(env->num_boids, sizeof(Log));
     env->log = (Log){0};
     env->tick = 0;
 
@@ -117,10 +114,12 @@ void c_step(Boids *env) {
     bool manual_control = IsKeyDown(KEY_LEFT_SHIFT);
     float mouse_x = (float)GetMouseX();
     float mouse_y = (float)GetMouseY();
+    float avg_reward = 0.0f;
 
     env->tick++;
     env->rewards[0] = 0;
     env->log.score = 0;
+    env->log.n = 0;
     for (unsigned current_indx = 0; current_indx < env->num_boids; current_indx++) {
         // apply action
         current_boid = &env->boids[current_indx];
@@ -183,14 +182,14 @@ void c_step(Boids *env) {
         // env->rewards[current_indx] = current_boid_reward / 15.0f;
         // printf("current_boid_reward: %f\n", current_boid_reward);
         env->rewards[current_indx] = current_boid_reward / 2.0f;
-
-        //log updates
-        if (env->tick % env->report_interval == 0) {
-            env->log.score += env->rewards[current_indx];
-            env->log.n += 1.0f;
-        }
+        avg_reward += env->rewards[current_indx];
+    }
+    //log updates
+    avg_reward /= env->num_boids;
+    if (env->tick % env->report_interval == 0) {
+        env->log.score = avg_reward;
+        env->log.n = 1;
     }
-    //env->log.score /= env->num_boids;
 
     compute_observations(env);
 }
@@ -210,7 +209,6 @@ void c_close_client(Client* client) {
 
 void c_close(Boids* env) {
     free(env->boids);
-    free(env->boid_logs);
     if (env->client != NULL) {
         c_close_client(env->client);
     }

From 1d424d7c3b86f0ca45a2d32f9091a6245dcdac27 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sat, 31 May 2025 16:08:37 -0400
Subject: [PATCH 06/39] fix overflow and zero report_interval

---
 pufferlib/ocean/boids/boids.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index 735f8e8c9e..14de0d5e5a 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -7,6 +7,7 @@
 
 // --- Demo Configuration ---
 #define NUM_BOIDS_DEMO 20   // Number of boids for the standalone demo
+#define REPORT_INTERVAL_DEMO 1000 // Report interval for the demo
 #define MAX_STEPS_DEMO 500 // Max steps per episode in the demo
 #define ACTION_SCALE 3.0f   // Corresponds to action space [-3.0, 3.0]
 
@@ -27,11 +28,12 @@ void demo() {
     // Initialize Boids environment struct
     Boids env = {0}; 
     env.num_boids = NUM_BOIDS_DEMO;
+    env.report_interval = REPORT_INTERVAL_DEMO;
     
     // In the Python binding, these pointers are assigned from NumPy arrays.
     // Here, we need to allocate them explicitly.
-    size_t obs_size = env.num_boids * 4; // num_boids * (x, y, vx, vy)
-    size_t act_size = env.num_boids * 2; // num_boids * (dvx, dvy)
+    size_t obs_size = env.num_boids * env.num_boids * 4; // the 4 = (x, y, vx, vy)
+    size_t act_size = env.num_boids * 2; // the 2 = (dvx, dvy)
     env.observations = (float*)calloc(obs_size, sizeof(float));
     env.actions = (float*)calloc(act_size, sizeof(float));
     env.rewards = (float*)calloc(env.num_boids, sizeof(float)); // Env-level reward

From 7376334bd58bb5ce389ffae62f8b08c752c97a4e Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sat, 31 May 2025 16:32:35 -0400
Subject: [PATCH 07/39] add above zero checks for num_boids and report_interval

---
 pufferlib/ocean/boids/boids.h | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index f815036114..c26ffa56e7 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -71,6 +71,14 @@ static void respawn_boid(Boids *env, unsigned int i) {
 }
 
 void init(Boids *env) {
+    if(env->num_boids < 1) {
+        printf("ERROR: num_boids must be bigger than 0\n");
+        exit(1);
+    }
+    if (env->report_interval < 1) {
+        printf("ERROR: report_interval must be bigger than 0\n");
+        exit(1);
+    }
     env->boids = (Boid*)calloc(env->num_boids, sizeof(Boid));
     env->log = (Log){0};
     env->tick = 0;

From ff483f6d08575504e5ad3050e3fa39979dd40111 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Mon, 2 Jun 2025 19:38:43 -0400
Subject: [PATCH 08/39] remove unused commented flat_actions

---
 pufferlib/ocean/boids/boids.py | 1 -
 1 file changed, 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index 96784b7bd9..a597b46ca3 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -49,7 +49,6 @@ def __init__(
         for env_num in range(num_envs):
             c_envs.append(binding.env_init(
                 self.observations[env_num*num_boids:(env_num+1)*num_boids],
-                #self.flat_actions[env_num*num_boids*ACTION_SPACE_SIZE:(env_num+1)*num_boids*ACTION_SPACE_SIZE],
                 self.actions[env_num*num_boids:(env_num+1)*num_boids],
                 self.rewards[env_num*num_boids:(env_num+1)*num_boids],
                 self.terminals[env_num*num_boids:(env_num+1)*num_boids],

From 26bebefe42050d58d7186a6a15677483f7a91ec4 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Mon, 2 Jun 2025 19:50:18 -0400
Subject: [PATCH 09/39] simplify seperation reward and test it

---
 pufferlib/config/ocean/boids.ini |  8 ++++----
 pufferlib/ocean/boids/boids.h    | 30 +++++++++++++++++-------------
 2 files changed, 21 insertions(+), 17 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index 0e6c8d33e6..691ef6c2a4 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -11,10 +11,10 @@ num_envs = 64
 num_boids = 64
 ; num_envs = 1
 ; num_boids = 1
-margin_turn_factor = 1.0
-cohesion_factor = 0.0
-seperation_factor = 0.0
-alignment_factor = 0.0
+margin_turn_factor = 0.0
+centering_factor = 0.00
+avoid_factor = 1.00
+matching_factor = 1.00
 
 [vec]
 num_workers = 2
diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index c26ffa56e7..1eb161532d 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -117,7 +117,7 @@ void c_step(Boids *env) {
     Boid* current_boid;
     Boid observed_boid;
     float vis_vx_sum, vis_vy_sum, vis_x_sum, vis_y_sum, vis_x_avg, vis_y_avg, vis_vx_avg, vis_vy_avg;
-    float diff_x, diff_y, dist, protected_dist_sum, current_boid_reward;
+    float diff_x, diff_y, dist, current_boid_reward;
     unsigned visual_count, protected_count;
     bool manual_control = IsKeyDown(KEY_LEFT_SHIFT);
     float mouse_x = (float)GetMouseX();
@@ -142,7 +142,7 @@ void c_step(Boids *env) {
         current_boid->y = flclip(current_boid->y + current_boid->velocity.y, 0, HEIGHT - BOID_HEIGHT);
 
         // reward calculation
-        current_boid_reward = 0.0f, protected_dist_sum = 0.0f, protected_count = 0.0f;
+        current_boid_reward = 0.0f, protected_count = 0.0f;
         visual_count = 0.0f, vis_vx_sum = 0.0f, vis_vy_sum = 0.0f, vis_x_sum = 0.0f, vis_y_sum = 0.0f;
         for (unsigned observed_indx = 0; observed_indx < env->num_boids; observed_indx++) {
             if (current_indx == observed_indx) continue;
@@ -151,7 +151,6 @@ void c_step(Boids *env) {
             diff_y = current_boid->y - observed_boid.y;
             dist = sqrtf(diff_x*diff_x + diff_y*diff_y);
             if (dist < PROTECTED_RANGE) {
-                protected_dist_sum += (PROTECTED_RANGE - dist);
                 protected_count++;
             } else if (dist < VISUAL_RANGE) {
                 vis_x_sum += observed_boid.x;
@@ -162,8 +161,7 @@ void c_step(Boids *env) {
             }
         }
         if (protected_count > 0) {
-            //current_boid_reward -= fabsf(protected_dist_sum / protected_count) * env->seperation_factor;
-            current_boid_reward -= flclip(protected_count/5.0, 0.0f, 1.0f) * env->seperation_factor;
+            current_boid_reward -= flclip(protected_count/env->num_boids * env->avoid_factor, 0.0f, 1.0f);
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;
@@ -189,15 +187,21 @@ void c_step(Boids *env) {
         // Normalization
         // env->rewards[current_indx] = current_boid_reward / 15.0f;
         // printf("current_boid_reward: %f\n", current_boid_reward);
-        env->rewards[current_indx] = current_boid_reward / 2.0f;
-        avg_reward += env->rewards[current_indx];
-    }
-    //log updates
-    avg_reward /= env->num_boids;
-    if (env->tick % env->report_interval == 0) {
-        env->log.score = avg_reward;
-        env->log.n = 1;
+        // env->rewards[current_indx] = current_boid_reward / 2.0f;
+        env->rewards[current_indx] = current_boid_reward;
+
+
+        //log updates
+        if (env->tick == env->report_interval) {
+            env->log.score          += env->rewards[current_indx];
+            env->log.n              += 1.0f;
+
+            /* clear per-boid log for next episode */
+            // env->boid_logs[boid_indx] = (Log){0};
+            env->tick = 0;
+        }
     }
+    //env->log.score /= env->num_boids;
 
     compute_observations(env);
 }

From 06878ebc84ba802fdd205ed62d2cd85b45b74572 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Mon, 2 Jun 2025 19:51:30 -0400
Subject: [PATCH 10/39] test out only avoid factor

---
 pufferlib/config/ocean/boids.ini | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index 691ef6c2a4..c9be5ee885 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -14,7 +14,7 @@ num_boids = 64
 margin_turn_factor = 0.0
 centering_factor = 0.00
 avoid_factor = 1.00
-matching_factor = 1.00
+matching_factor = 0.00
 
 [vec]
 num_workers = 2

From 35de375a2a01fa36827eb9157df7d31a0ca5eb01 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Tue, 3 Jun 2025 16:07:11 -0400
Subject: [PATCH 11/39] remove unused avg_reward and change seperation factor
 reward

---
 pufferlib/ocean/boids/boids.h | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 1eb161532d..f5c9037e62 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -122,7 +122,6 @@ void c_step(Boids *env) {
     bool manual_control = IsKeyDown(KEY_LEFT_SHIFT);
     float mouse_x = (float)GetMouseX();
     float mouse_y = (float)GetMouseY();
-    float avg_reward = 0.0f;
 
     env->tick++;
     env->rewards[0] = 0;
@@ -161,7 +160,7 @@ void c_step(Boids *env) {
             }
         }
         if (protected_count > 0) {
-            current_boid_reward -= flclip(protected_count/env->num_boids * env->avoid_factor, 0.0f, 1.0f);
+            current_boid_reward -= (float)((env->num_boids - protected_count) - protected_count) / env->num_boids * env->seperation_factor;
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;

From 9709a469a071fc11976237256fa7b7bb64065c7f Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 4 Jun 2025 14:17:11 -0400
Subject: [PATCH 12/39] fix factor names

---
 pufferlib/config/ocean/boids.ini | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index c9be5ee885..0edba1bff0 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -12,9 +12,9 @@ num_boids = 64
 ; num_envs = 1
 ; num_boids = 1
 margin_turn_factor = 0.0
-centering_factor = 0.00
-avoid_factor = 1.00
-matching_factor = 0.00
+cohesion_factor = 0.00
+seperation_factor = 1.00
+alignment_factor = 0.00
 
 [vec]
 num_workers = 2

From 1c28c7230a8a89de573563692667bcddd3460c9d Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 4 Jun 2025 15:06:38 -0400
Subject: [PATCH 13/39] remove unused commented code

---
 pufferlib/ocean/boids/boids.h | 4 ----
 1 file changed, 4 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index f5c9037e62..32ead6670c 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -194,13 +194,9 @@ void c_step(Boids *env) {
         if (env->tick == env->report_interval) {
             env->log.score          += env->rewards[current_indx];
             env->log.n              += 1.0f;
-
-            /* clear per-boid log for next episode */
-            // env->boid_logs[boid_indx] = (Log){0};
             env->tick = 0;
         }
     }
-    //env->log.score /= env->num_boids;
 
     compute_observations(env);
 }

From bf8c75f7520d781e52918cc6747f9e5f45001f4e Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 02:29:46 -0400
Subject: [PATCH 14/39] fix seperation factor reward calculation

---
 pufferlib/ocean/boids/boids.h | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 32ead6670c..476b461e7e 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -117,7 +117,7 @@ void c_step(Boids *env) {
     Boid* current_boid;
     Boid observed_boid;
     float vis_vx_sum, vis_vy_sum, vis_x_sum, vis_y_sum, vis_x_avg, vis_y_avg, vis_vx_avg, vis_vy_avg;
-    float diff_x, diff_y, dist, current_boid_reward;
+    float diff_x, diff_y, dist, current_boid_reward, protected_range_diff;
     unsigned visual_count, protected_count;
     bool manual_control = IsKeyDown(KEY_LEFT_SHIFT);
     float mouse_x = (float)GetMouseX();
@@ -160,7 +160,8 @@ void c_step(Boids *env) {
             }
         }
         if (protected_count > 0) {
-            current_boid_reward -= (float)((env->num_boids - protected_count) - protected_count) / env->num_boids * env->seperation_factor;
+            protected_range_diff = (float)(env->num_boids - protected_count) - protected_count;
+            current_boid_reward += protected_range_diff / env->num_boids * env->seperation_factor;
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;

From 23e2399bc0582bded8a47ff1ffe8a418e790ca50 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 03:02:49 -0400
Subject: [PATCH 15/39] remove unused commented params

---
 pufferlib/config/ocean/boids.ini | 47 +-------------------------------
 1 file changed, 1 insertion(+), 46 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index 0edba1bff0..0b3fe5a6d0 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -25,49 +25,4 @@ batch_size = auto
 total_timesteps = 100_000_000
 gamma = 0.95
 learning_rate = 0.025
-minibatch_size = 16384
-; minibatch_size = 1
-
-; [sweep]
-; method = protein
-; metric = episode_length
-
-; [sweep.train.total_timesteps]
-; distribution = log_normal
-; min = 1e6
-; max = 1e7
-; mean = 5e6
-; scale = 0.5
-
-; [sweep.train.gamma]
-; distribution = log_normal
-; min = 0.9
-; max = 0.999
-; mean = 0.97
-
-; [sweep.train.gae_lambda]
-; distribution = log_normal
-; min = 0.7
-; max = 0.999
-; mean = 0.95
-
-; [sweep.train.learning_rate]
-; distribution = log_normal
-; min = 0.0001
-; max = 0.001
-; mean = 0.00025
-; scale = 0.5
-
-; [sweep.train.batch_size]
-; min = 32768
-; max = 131072
-; mean = 65536
-; scale = 0.5
-
-; [sweep.train.minibatch_size]
-; min = 512
-; max = 2048
-; mean = 1024
-; scale = 0.5
-
-
+minibatch_size = 16384
\ No newline at end of file

From bb162fb424fb6030a5ff21d2762ac1409754f7b8 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 12:58:48 -0400
Subject: [PATCH 16/39] remove normalization from separation factor calculation

---
 pufferlib/ocean/boids/boids.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 476b461e7e..0b0db55b0e 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -161,7 +161,7 @@ void c_step(Boids *env) {
         }
         if (protected_count > 0) {
             protected_range_diff = (float)(env->num_boids - protected_count) - protected_count;
-            current_boid_reward += protected_range_diff / env->num_boids * env->seperation_factor;
+            current_boid_reward += protected_range_diff * env->seperation_factor;
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;

From 85d589168b6cf3e31119e86ca9da1ffcd69d189c Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 14:21:18 -0400
Subject: [PATCH 17/39] fix visual range

---
 pufferlib/ocean/boids/boids.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 0b0db55b0e..77a98f204f 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -13,7 +13,7 @@
 #define LEFT_MARGIN 50
 #define RIGHT_MARGIN 50
 #define VELOCITY_CAP 5
-#define VISUAL_RANGE 20
+#define VISUAL_RANGE 400
 #define PROTECTED_RANGE 100
 #define WIDTH 1080
 #define HEIGHT 720

From 435ac9eaf9ca9359832c1f0b4fde5953091993d4 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 20:24:34 -0400
Subject: [PATCH 18/39] remove positve margin rewards and remove commented code

---
 pufferlib/ocean/boids/boids.h | 15 ++++-----------
 1 file changed, 4 insertions(+), 11 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 77a98f204f..744440cf90 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -10,8 +10,8 @@
 
 #define TOP_MARGIN 50
 #define BOTTOM_MARGIN 50
-#define LEFT_MARGIN 50
-#define RIGHT_MARGIN 50
+#define LEFT_MARGIN 100
+#define RIGHT_MARGIN 100
 #define VELOCITY_CAP 5
 #define VISUAL_RANGE 400
 #define PROTECTED_RANGE 100
@@ -176,20 +176,13 @@ void c_step(Boids *env) {
         }
         if (current_boid->y < TOP_MARGIN || current_boid->y > HEIGHT - BOTTOM_MARGIN) {
             current_boid_reward -= env->margin_turn_factor;
-        } else {
-            current_boid_reward += env->margin_turn_factor;
         }
         if (current_boid->x < LEFT_MARGIN || current_boid->x > WIDTH  - RIGHT_MARGIN) {
             current_boid_reward -= env->margin_turn_factor;
-        } else {
-            current_boid_reward += env->margin_turn_factor;
         }
-        // Normalization
-        // env->rewards[current_indx] = current_boid_reward / 15.0f;
-        // printf("current_boid_reward: %f\n", current_boid_reward);
-        // env->rewards[current_indx] = current_boid_reward / 2.0f;
-        env->rewards[current_indx] = current_boid_reward;
 
+        // Normalization
+        env->rewards[current_indx] = current_boid_reward / 4.0f;
 
         //log updates
         if (env->tick == env->report_interval) {

From 463f60a5394f99b875c878bd62a5cff05a49b9b4 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 20:33:52 -0400
Subject: [PATCH 19/39] add factors to env run with "boids.c"

---
 pufferlib/ocean/boids/boids.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index 14de0d5e5a..aca6d35c68 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -10,6 +10,10 @@
 #define REPORT_INTERVAL_DEMO 1000 // Report interval for the demo
 #define MAX_STEPS_DEMO 500 // Max steps per episode in the demo
 #define ACTION_SCALE 3.0f   // Corresponds to action space [-3.0, 3.0]
+#define MARGIN_TURN_FACTOR 1.0
+#define COHESION_FACTOR 0.0
+#define SEPERATION_FACTOR 0.0
+#define ALIGNMENT_FACTOR 0.0
 
 // Dummy action generation: random velocity changes for each boid
 void generate_dummy_actions(Boids* env) {
@@ -29,6 +33,10 @@ void demo() {
     Boids env = {0}; 
     env.num_boids = NUM_BOIDS_DEMO;
     env.report_interval = REPORT_INTERVAL_DEMO;
+    env.margin_turn_factor = MARGIN_TURN_FACTOR;
+    env.cohesion_factor = COHESION_FACTOR;
+    env.seperation_factor = SEPERATION_FACTOR;
+    env.alignment_factor = ALIGNMENT_FACTOR;
     
     // In the Python binding, these pointers are assigned from NumPy arrays.
     // Here, we need to allocate them explicitly.

From 00af443301c3e5745710aaeb6c7566bd29585db9 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 20:44:00 -0400
Subject: [PATCH 20/39] add debug margin lines and adjust reward normalization

---
 pufferlib/ocean/boids/boids.h | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 744440cf90..7caef00d68 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -182,7 +182,7 @@ void c_step(Boids *env) {
         }
 
         // Normalization
-        env->rewards[current_indx] = current_boid_reward / 4.0f;
+        env->rewards[current_indx] = current_boid_reward / 2.0f;
 
         //log updates
         if (env->tick == env->report_interval) {
@@ -257,6 +257,11 @@ void c_render(Boids* env) {
         BeginDrawing();
         ClearBackground((Color){6, 24, 24, 255});
 
+        DrawLine(LEFT_MARGIN, 0, LEFT_MARGIN, HEIGHT, RED);
+        DrawLine(WIDTH - RIGHT_MARGIN, 0, WIDTH - RIGHT_MARGIN, HEIGHT, RED);
+        DrawLine(0, TOP_MARGIN, WIDTH, TOP_MARGIN, RED);
+        DrawLine(0, HEIGHT - BOTTOM_MARGIN, WIDTH, HEIGHT - BOTTOM_MARGIN, RED);
+
         for (unsigned boid_indx = 0; boid_indx < env->num_boids; boid_indx++) {
             DrawTexturePro(
                 env->client->boid_texture,

From fc4e72217b3123e9686c78f37c891caf36ca4d06 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 20:45:09 -0400
Subject: [PATCH 21/39] only turn on margin turn factor and adjust total
 timesteps

---
 pufferlib/config/ocean/boids.ini | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index 0b3fe5a6d0..b3732b5126 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -11,10 +11,10 @@ num_envs = 64
 num_boids = 64
 ; num_envs = 1
 ; num_boids = 1
-margin_turn_factor = 0.0
-cohesion_factor = 0.00
-seperation_factor = 1.00
-alignment_factor = 0.00
+margin_turn_factor = 1.0
+cohesion_factor = 0.0
+seperation_factor = 0.0
+alignment_factor = 0.0
 
 [vec]
 num_workers = 2
@@ -22,7 +22,8 @@ num_envs = 2
 batch_size = auto
 
 [train]
-total_timesteps = 100_000_000
+total_timesteps = 150_000_000
+; total_timesteps = 15_000_000
 gamma = 0.95
 learning_rate = 0.025
 minibatch_size = 16384
\ No newline at end of file

From a413221013fd3656a69354d69d572a7af553654b Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 21:35:23 -0400
Subject: [PATCH 22/39] change top/bottom margins

---
 pufferlib/ocean/boids/boids.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 7caef00d68..aef336cd1a 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -8,8 +8,8 @@
 
 #include "raylib.h"
 
-#define TOP_MARGIN 50
-#define BOTTOM_MARGIN 50
+#define TOP_MARGIN 100
+#define BOTTOM_MARGIN 100
 #define LEFT_MARGIN 100
 #define RIGHT_MARGIN 100
 #define VELOCITY_CAP 5

From 342d83c80ee65c80037fe6c5cf9250289758b8de Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Thu, 5 Jun 2025 21:39:33 -0400
Subject: [PATCH 23/39] account for boid width and hight in margin reward
 calculation

---
 pufferlib/ocean/boids/boids.h | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index aef336cd1a..54242bd65e 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -174,10 +174,10 @@ void c_step(Boids *env) {
             current_boid_reward -= fabsf(vis_x_avg  - current_boid->x) * env->cohesion_factor;
             current_boid_reward -= fabsf(vis_y_avg  - current_boid->y) * env->cohesion_factor;
         }
-        if (current_boid->y < TOP_MARGIN || current_boid->y > HEIGHT - BOTTOM_MARGIN) {
+        if (current_boid->y < TOP_MARGIN || current_boid->y + BOID_HEIGHT > HEIGHT - BOTTOM_MARGIN) {
             current_boid_reward -= env->margin_turn_factor;
         }
-        if (current_boid->x < LEFT_MARGIN || current_boid->x > WIDTH  - RIGHT_MARGIN) {
+        if (current_boid->x < LEFT_MARGIN || current_boid->x + BOID_WIDTH > WIDTH - RIGHT_MARGIN) {
             current_boid_reward -= env->margin_turn_factor;
         }
 

From 01a84c0bb0c9398a6dc7fb8b76516657a410a08b Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 00:11:20 -0400
Subject: [PATCH 24/39] increase max steps

---
 pufferlib/ocean/boids/boids.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index aca6d35c68..8718e70065 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -8,7 +8,7 @@
 // --- Demo Configuration ---
 #define NUM_BOIDS_DEMO 20   // Number of boids for the standalone demo
 #define REPORT_INTERVAL_DEMO 1000 // Report interval for the demo
-#define MAX_STEPS_DEMO 500 // Max steps per episode in the demo
+#define MAX_STEPS_DEMO 10000 // Max steps per episode in the demo
 #define ACTION_SCALE 3.0f   // Corresponds to action space [-3.0, 3.0]
 #define MARGIN_TURN_FACTOR 1.0
 #define COHESION_FACTOR 0.0

From 117e9b6e610278cdffa4a92669f6763202d7c71e Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 20:17:24 -0400
Subject: [PATCH 25/39] remove debug margin lines

---
 pufferlib/ocean/boids/boids.h | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 54242bd65e..6b0311d376 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -257,11 +257,6 @@ void c_render(Boids* env) {
         BeginDrawing();
         ClearBackground((Color){6, 24, 24, 255});
 
-        DrawLine(LEFT_MARGIN, 0, LEFT_MARGIN, HEIGHT, RED);
-        DrawLine(WIDTH - RIGHT_MARGIN, 0, WIDTH - RIGHT_MARGIN, HEIGHT, RED);
-        DrawLine(0, TOP_MARGIN, WIDTH, TOP_MARGIN, RED);
-        DrawLine(0, HEIGHT - BOTTOM_MARGIN, WIDTH, HEIGHT - BOTTOM_MARGIN, RED);
-
         for (unsigned boid_indx = 0; boid_indx < env->num_boids; boid_indx++) {
             DrawTexturePro(
                 env->client->boid_texture,

From 618cb0b4e6cc875f41e276fe76778442bf5cf1bc Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 20:18:27 -0400
Subject: [PATCH 26/39] fix observations for margin factor

---
 pufferlib/ocean/boids/boids.h  | 20 ++++++++++++++++----
 pufferlib/ocean/boids/boids.py |  2 +-
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 6b0311d376..b883d72246 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -95,11 +95,23 @@ void init(Boids *env) {
 static void compute_observations(Boids *env) {
     int idx = 0;
     for (unsigned i=0; i<env->num_boids; i++) {
+        env->observations[idx++] = env->boids[i].x / WIDTH;
+        env->observations[idx++] = env->boids[i].y / HEIGHT;
+        env->observations[idx++] = env->boids[i].velocity.x / VELOCITY_CAP;
+        env->observations[idx++] = env->boids[i].velocity.y / VELOCITY_CAP;
+        for (unsigned j=0; j<4; j++) {
+            env->observations[idx++] = 0;
+        }
         for (unsigned j=0; j<env->num_boids; j++) {
-            env->observations[idx++] = (env->boids[j].x - env->boids[i].x) / WIDTH;
-            env->observations[idx++] = (env->boids[j].y - env->boids[i].y) / HEIGHT;
-            env->observations[idx++] = (env->boids[j].velocity.x - env->boids[i].velocity.x) / VELOCITY_CAP;
-            env->observations[idx++] = (env->boids[j].velocity.y - env->boids[i].velocity.y) / VELOCITY_CAP;
+            if (i == j) continue;
+            env->observations[idx++] = env->boids[j].x / WIDTH;
+            env->observations[idx++] = env->boids[j].y / HEIGHT;
+            env->observations[idx++] = env->boids[j].velocity.x / VELOCITY_CAP;
+            env->observations[idx++] = env->boids[j].velocity.y / VELOCITY_CAP;
+            env->observations[idx++] = (env->boids[i].x - env->boids[j].x) / WIDTH;
+            env->observations[idx++] = (env->boids[i].y - env->boids[j].y) / HEIGHT;
+            env->observations[idx++] = (env->boids[i].velocity.x - env->boids[j].velocity.x) / VELOCITY_CAP;
+            env->observations[idx++] = (env->boids[i].velocity.y - env->boids[j].velocity.y) / VELOCITY_CAP;
         }
     }
 }
diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index a597b46ca3..5582cec81f 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -27,7 +27,7 @@ def __init__(
         self.num_boids = num_boids
 
         self.single_observation_space = gymnasium.spaces.Box(
-            -1000.0, 1000.0, shape=(num_boids*4,), dtype=np.float32
+            -1000.0, 1000.0, shape=(num_boids*8,), dtype=np.float32
         )
         
         #self.single_action_space = gymnasium.spaces.Box(

From 7526366e017bf29aa71f547b5cbd950011dcd4d1 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 20:50:23 -0400
Subject: [PATCH 27/39] remove single agent params

---
 pufferlib/config/ocean/boids.ini | 10 ++++------
 1 file changed, 4 insertions(+), 6 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index b3732b5126..ca757d9cf1 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -9,11 +9,9 @@ rnn_name = Recurrent
 [env]
 num_envs = 64
 num_boids = 64
-; num_envs = 1
-; num_boids = 1
-margin_turn_factor = 1.0
+margin_turn_factor = 0.0
 cohesion_factor = 0.0
-seperation_factor = 0.0
+seperation_factor = 1.0
 alignment_factor = 0.0
 
 [vec]
@@ -22,8 +20,8 @@ num_envs = 2
 batch_size = auto
 
 [train]
-total_timesteps = 150_000_000
-; total_timesteps = 15_000_000
+; total_timesteps = 150_000_000
+total_timesteps = 50_000_000
 gamma = 0.95
 learning_rate = 0.025
 minibatch_size = 16384
\ No newline at end of file

From 2d38d98d1fc72399694d85d6a0717f19d23f9c47 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 21:59:46 -0400
Subject: [PATCH 28/39] update boids.c observations allocation

---
 pufferlib/ocean/boids/boids.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index 8718e70065..ef522c3df1 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -38,9 +38,7 @@ void demo() {
     env.seperation_factor = SEPERATION_FACTOR;
     env.alignment_factor = ALIGNMENT_FACTOR;
     
-    // In the Python binding, these pointers are assigned from NumPy arrays.
-    // Here, we need to allocate them explicitly.
-    size_t obs_size = env.num_boids * env.num_boids * 4; // the 4 = (x, y, vx, vy)
+    size_t obs_size = env.num_boids * env.num_boids * 8; // 8 = (x, y, vx, vy, dx, dy, dvx, dvy)
     size_t act_size = env.num_boids * 2; // the 2 = (dvx, dvy)
     env.observations = (float*)calloc(obs_size, sizeof(float));
     env.actions = (float*)calloc(act_size, sizeof(float));

From 971732bd471271dbf544a23dc4ffba7c785ca76b Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Wed, 11 Jun 2025 22:00:04 -0400
Subject: [PATCH 29/39] update observations and actions comments

---
 pufferlib/ocean/boids/boids.h | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index b883d72246..547b62e4a8 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -39,9 +39,12 @@ typedef struct {
 
 typedef struct Client Client;
 typedef struct {
-    // an array of shape (num_boids, 4) with the 4 values correspoinding to (x, y, velocity x, velocity y)
+    // Flat array of shape (num_boids * 8) values:
+    // - Each boid has 8 values corresponding to (x, y, vx, vy, dx, dy, dvx, dvy)
+    // - The first 8 values are for the boid itself
+    // - All the other 8 values for the other boids
     float* observations;
-    // an array of shape (num_boids, 2) with the 2 values correspoinding to (velocity x, velocity y)
+    // an array of shape (num_boids, 2) with the 2 values correspoinding to (dvx, dvy)
     float* actions;
     // an array of shape (1) with the summed up reward for all boids
     float* rewards;

From bcecdd22eb0855549692d2e142efea6d2e87a34c Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Fri, 4 Jul 2025 01:39:29 -0400
Subject: [PATCH 30/39] remove commented parameters and update parameters to
 current best

---
 pufferlib/config/ocean/boids.ini | 13 +++++--------
 1 file changed, 5 insertions(+), 8 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index ca757d9cf1..cf03a8aa8d 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -1,18 +1,16 @@
 [base]
 package = ocean
 env_name = puffer_boids 
-; policy_name = Boids
 policy_name = Policy
 rnn_name = Recurrent
-; rnn_name = None
 
 [env]
 num_envs = 64
 num_boids = 64
-margin_turn_factor = 0.0
-cohesion_factor = 0.0
-seperation_factor = 1.0
-alignment_factor = 0.0
+margin_turn_factor = 2.0
+cohesion_factor = 0.0048
+separation_factor = 0.0128
+alignment_factor = 0.2
 
 [vec]
 num_workers = 2
@@ -20,8 +18,7 @@ num_envs = 2
 batch_size = auto
 
 [train]
-; total_timesteps = 150_000_000
-total_timesteps = 50_000_000
+total_timesteps = 100_000_000
 gamma = 0.95
 learning_rate = 0.025
 minibatch_size = 16384
\ No newline at end of file

From a9f1b98c88e4524e43e968e29f945a96c059461c Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Fri, 4 Jul 2025 01:41:22 -0400
Subject: [PATCH 31/39] fix to "separation_factor" instead of
 "seperation_factor"

---
 pufferlib/ocean/boids/binding.c | 2 +-
 pufferlib/ocean/boids/boids.c   | 4 ++--
 pufferlib/ocean/boids/boids.h   | 2 +-
 pufferlib/ocean/boids/boids.py  | 4 ++--
 4 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/pufferlib/ocean/boids/binding.c b/pufferlib/ocean/boids/binding.c
index 89cb389759..d8e5820169 100644
--- a/pufferlib/ocean/boids/binding.c
+++ b/pufferlib/ocean/boids/binding.c
@@ -8,7 +8,7 @@ static int my_init(Env* env, PyObject* args, PyObject* kwargs) {
     env->report_interval = unpack(kwargs, "report_interval");
     env->margin_turn_factor = unpack(kwargs, "margin_turn_factor");
     env->cohesion_factor = unpack(kwargs, "cohesion_factor");
-    env->seperation_factor = unpack(kwargs, "seperation_factor");
+    env->separation_factor = unpack(kwargs, "separation_factor");
     env->alignment_factor = unpack(kwargs, "alignment_factor");
     init(env);
     return 0;
diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index ef522c3df1..4e4937666c 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -12,7 +12,7 @@
 #define ACTION_SCALE 3.0f   // Corresponds to action space [-3.0, 3.0]
 #define MARGIN_TURN_FACTOR 1.0
 #define COHESION_FACTOR 0.0
-#define SEPERATION_FACTOR 0.0
+#define SEPARATION_FACTOR 0.0
 #define ALIGNMENT_FACTOR 0.0
 
 // Dummy action generation: random velocity changes for each boid
@@ -35,7 +35,7 @@ void demo() {
     env.report_interval = REPORT_INTERVAL_DEMO;
     env.margin_turn_factor = MARGIN_TURN_FACTOR;
     env.cohesion_factor = COHESION_FACTOR;
-    env.seperation_factor = SEPERATION_FACTOR;
+    env.separation_factor = SEPARATION_FACTOR;
     env.alignment_factor = ALIGNMENT_FACTOR;
     
     size_t obs_size = env.num_boids * env.num_boids * 8; // 8 = (x, y, vx, vy, dx, dy, dvx, dvy)
diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 547b62e4a8..8c17f8712f 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -53,7 +53,7 @@ typedef struct {
     unsigned int num_boids;
     float margin_turn_factor;
     float cohesion_factor;
-    float seperation_factor;
+    float separation_factor;
     float alignment_factor;
     unsigned tick;
     Log log;
diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index 5582cec81f..6eadcc8627 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -19,7 +19,7 @@ def __init__(
         num_boids=1,
         margin_turn_factor=1.0,
         cohesion_factor=0.0,
-        seperation_factor=0.0,
+        separation_factor=0.0,
         alignment_factor=0.0
     ):
         ACTION_SPACE_SIZE = 2
@@ -58,7 +58,7 @@ def __init__(
                 report_interval=self.report_interval,
                 margin_turn_factor=margin_turn_factor,
                 cohesion_factor=cohesion_factor,
-                seperation_factor=seperation_factor,
+                separation_factor=separation_factor,
                 alignment_factor=alignment_factor,
             ))
         

From 9dc328fe0e704eb7f5a74dec796e60bdd2653b53 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Fri, 4 Jul 2025 01:42:33 -0400
Subject: [PATCH 32/39] update preset env parameters

---
 pufferlib/ocean/boids/boids.h | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 8c17f8712f..460b4531b9 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -8,13 +8,13 @@
 
 #include "raylib.h"
 
-#define TOP_MARGIN 100
-#define BOTTOM_MARGIN 100
-#define LEFT_MARGIN 100
-#define RIGHT_MARGIN 100
+#define TOP_MARGIN 50
+#define BOTTOM_MARGIN 50
+#define LEFT_MARGIN 50
+#define RIGHT_MARGIN 50
 #define VELOCITY_CAP 5
 #define VISUAL_RANGE 400
-#define PROTECTED_RANGE 100
+#define PROTECTED_RANGE 60
 #define WIDTH 1080
 #define HEIGHT 720
 #define BOID_WIDTH 32

From 4b339a36eca35a8a174c395232a21d4dbc609b94 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Fri, 4 Jul 2025 01:42:50 -0400
Subject: [PATCH 33/39] condence controlled boid observation loop

---
 pufferlib/ocean/boids/boids.h | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 460b4531b9..f84e25b4c6 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -102,9 +102,7 @@ static void compute_observations(Boids *env) {
         env->observations[idx++] = env->boids[i].y / HEIGHT;
         env->observations[idx++] = env->boids[i].velocity.x / VELOCITY_CAP;
         env->observations[idx++] = env->boids[i].velocity.y / VELOCITY_CAP;
-        for (unsigned j=0; j<4; j++) {
-            env->observations[idx++] = 0;
-        }
+        for (unsigned j=0; j<4; j++) { env->observations[idx++] = 0; }
         for (unsigned j=0; j<env->num_boids; j++) {
             if (i == j) continue;
             env->observations[idx++] = env->boids[j].x / WIDTH;

From e56d28f6eb5abb6301d561dc20ed21f43d749f58 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sun, 6 Jul 2025 23:15:44 -0400
Subject: [PATCH 34/39] remove use of protected range diff

---
 pufferlib/ocean/boids/boids.h | 8 +++++---
 1 file changed, 5 insertions(+), 3 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index f84e25b4c6..8b45dd463a 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -130,7 +130,7 @@ void c_step(Boids *env) {
     Boid* current_boid;
     Boid observed_boid;
     float vis_vx_sum, vis_vy_sum, vis_x_sum, vis_y_sum, vis_x_avg, vis_y_avg, vis_vx_avg, vis_vy_avg;
-    float diff_x, diff_y, dist, current_boid_reward, protected_range_diff;
+    float diff_x, diff_y, dist, current_boid_reward;
     unsigned visual_count, protected_count;
     bool manual_control = IsKeyDown(KEY_LEFT_SHIFT);
     float mouse_x = (float)GetMouseX();
@@ -173,8 +173,10 @@ void c_step(Boids *env) {
             }
         }
         if (protected_count > 0) {
-            protected_range_diff = (float)(env->num_boids - protected_count) - protected_count;
-            current_boid_reward += protected_range_diff * env->seperation_factor;
+            // protected_range_diff = (float)(env->num_boids - protected_count) - protected_count;
+            // current_boid_reward += protected_range_diff * env->seperation_factor;
+
+            current_boid_reward -= protected_count * env->separation_factor;
         }
         if (visual_count) {
             vis_x_avg  = vis_x_sum  / visual_count;

From cc8a397b18870090f90e3f56d5225eb25a8f2e45 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Mon, 7 Jul 2025 00:36:54 -0400
Subject: [PATCH 35/39] change reward normalization number

---
 pufferlib/ocean/boids/boids.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 8b45dd463a..f7391126d2 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -197,7 +197,10 @@ void c_step(Boids *env) {
         }
 
         // Normalization
-        env->rewards[current_indx] = current_boid_reward / 2.0f;
+        // env->rewards[current_indx] = current_boid_reward;
+        env->rewards[current_indx] = current_boid_reward / 6.0f;
+        // env->rewards[current_indx] = current_boid_reward / 205.0f;
+        // env->rewards[current_indx] = current_boid_reward / 10.0f;
 
         //log updates
         if (env->tick == env->report_interval) {

From 0eda88952542d85cab5a63549acd0a77299014b0 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Tue, 8 Jul 2025 19:52:52 -0400
Subject: [PATCH 36/39] update puffer resource path

---
 pufferlib/ocean/boids/boids.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index f7391126d2..9e9b87db89 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -19,7 +19,7 @@
 #define HEIGHT 720
 #define BOID_WIDTH 32
 #define BOID_HEIGHT 32
-#define BOID_TEXTURE_PATH "./resources/puffers_128.png"
+#define BOID_TEXTURE_PATH "./resources/shared/puffers_128.png"
 
 typedef struct {
     float score;

From 6158014a91f38f971308c86b0fe1a54d3748ede8 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sun, 27 Jul 2025 16:30:53 -0400
Subject: [PATCH 37/39] enable all factors

---
 pufferlib/config/ocean/boids.ini | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/pufferlib/config/ocean/boids.ini b/pufferlib/config/ocean/boids.ini
index cf03a8aa8d..683d8e9189 100644
--- a/pufferlib/config/ocean/boids.ini
+++ b/pufferlib/config/ocean/boids.ini
@@ -6,8 +6,8 @@ rnn_name = Recurrent
 
 [env]
 num_envs = 64
-num_boids = 64
-margin_turn_factor = 2.0
+num_boids = 16
+margin_turn_factor = 1.0
 cohesion_factor = 0.0048
 separation_factor = 0.0128
 alignment_factor = 0.2
@@ -19,6 +19,7 @@ batch_size = auto
 
 [train]
 total_timesteps = 100_000_000
+; total_timesteps = 80_000_000
 gamma = 0.95
 learning_rate = 0.025
 minibatch_size = 16384
\ No newline at end of file

From e67ad51947ff882c537f9977e132a7460b71b6bf Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sun, 27 Jul 2025 16:31:39 -0400
Subject: [PATCH 38/39] add euclidean distance to observations

---
 pufferlib/ocean/boids/boids.h  | 17 ++++++++++++++---
 pufferlib/ocean/boids/boids.py |  2 +-
 2 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/pufferlib/ocean/boids/boids.h b/pufferlib/ocean/boids/boids.h
index 9e9b87db89..a56f489437 100644
--- a/pufferlib/ocean/boids/boids.h
+++ b/pufferlib/ocean/boids/boids.h
@@ -20,6 +20,7 @@
 #define BOID_WIDTH 32
 #define BOID_HEIGHT 32
 #define BOID_TEXTURE_PATH "./resources/shared/puffers_128.png"
+#define MAX_DIST 2000
 
 typedef struct {
     float score;
@@ -97,20 +98,30 @@ void init(Boids *env) {
 
 static void compute_observations(Boids *env) {
     int idx = 0;
+    float diff_x, diff_y, dist;
     for (unsigned i=0; i<env->num_boids; i++) {
+        // observations for the current boid
         env->observations[idx++] = env->boids[i].x / WIDTH;
         env->observations[idx++] = env->boids[i].y / HEIGHT;
         env->observations[idx++] = env->boids[i].velocity.x / VELOCITY_CAP;
         env->observations[idx++] = env->boids[i].velocity.y / VELOCITY_CAP;
-        for (unsigned j=0; j<4; j++) { env->observations[idx++] = 0; }
+        // zeros for relative observations since comparing to itself will always be 0
+        for (unsigned j=0; j<5; j++) { env->observations[idx++] = 0; }
+
+        // observations for the other boids compared to the current boid
         for (unsigned j=0; j<env->num_boids; j++) {
             if (i == j) continue;
+            diff_x = env->boids[i].x - env->boids[j].x;
+            diff_y = env->boids[i].y - env->boids[j].y;
+            dist = sqrtf(diff_x*diff_x + diff_y*diff_y);
+
             env->observations[idx++] = env->boids[j].x / WIDTH;
             env->observations[idx++] = env->boids[j].y / HEIGHT;
             env->observations[idx++] = env->boids[j].velocity.x / VELOCITY_CAP;
             env->observations[idx++] = env->boids[j].velocity.y / VELOCITY_CAP;
-            env->observations[idx++] = (env->boids[i].x - env->boids[j].x) / WIDTH;
-            env->observations[idx++] = (env->boids[i].y - env->boids[j].y) / HEIGHT;
+            env->observations[idx++] = diff_x / WIDTH;
+            env->observations[idx++] = diff_y / HEIGHT;
+            env->observations[idx++] = dist / MAX_DIST;
             env->observations[idx++] = (env->boids[i].velocity.x - env->boids[j].velocity.x) / VELOCITY_CAP;
             env->observations[idx++] = (env->boids[i].velocity.y - env->boids[j].velocity.y) / VELOCITY_CAP;
         }
diff --git a/pufferlib/ocean/boids/boids.py b/pufferlib/ocean/boids/boids.py
index 6eadcc8627..5e9b522226 100644
--- a/pufferlib/ocean/boids/boids.py
+++ b/pufferlib/ocean/boids/boids.py
@@ -27,7 +27,7 @@ def __init__(
         self.num_boids = num_boids
 
         self.single_observation_space = gymnasium.spaces.Box(
-            -1000.0, 1000.0, shape=(num_boids*8,), dtype=np.float32
+            -1000.0, 1000.0, shape=(num_boids*9,), dtype=np.float32
         )
         
         #self.single_action_space = gymnasium.spaces.Box(

From 0bf4d7410fda7354cce4d49e344a8e50c3616687 Mon Sep 17 00:00:00 2001
From: PLAZMAMA <maorioriori@gmail.com>
Date: Sun, 27 Jul 2025 16:34:44 -0400
Subject: [PATCH 39/39] add euclidean distance to local build observations

---
 pufferlib/ocean/boids/boids.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/pufferlib/ocean/boids/boids.c b/pufferlib/ocean/boids/boids.c
index 4e4937666c..0573cb9735 100644
--- a/pufferlib/ocean/boids/boids.c
+++ b/pufferlib/ocean/boids/boids.c
@@ -38,7 +38,7 @@ void demo() {
     env.separation_factor = SEPARATION_FACTOR;
     env.alignment_factor = ALIGNMENT_FACTOR;
     
-    size_t obs_size = env.num_boids * env.num_boids * 8; // 8 = (x, y, vx, vy, dx, dy, dvx, dvy)
+    size_t obs_size = env.num_boids * env.num_boids * 9; // 9 = (x, y, vx, vy, dx, dy, dist, dvx, dvy)
     size_t act_size = env.num_boids * 2; // the 2 = (dvx, dvy)
     env.observations = (float*)calloc(obs_size, sizeof(float));
     env.actions = (float*)calloc(act_size, sizeof(float));