Update README for v13b

README.md

@@ -11,26 +11,38 @@ tags:
 
 PPO-trained agent for [OpenFront.io](https://openfront.io), a multiplayer territory control game.
 
+## Model Version: v13b
+
+Current best model trained with normalized elimination reward and winner bonus.
+
 ## Training Details
 
 - **Algorithm:** PPO (Proximal Policy Optimization)
 - **Architecture:** Actor-Critic with shared backbone (512→512→256)
-- **Observation dim:** 80
-- **Max neighbors:** 16
-- **Maps:** plains, big_plains, world, giantworldmap, ocean_and_land, half_land_half_ocean, europe, europeclassic, northamerica, africa, asia, australia, southamerica, mediterranean, britannia, britanniaclassic, eastasia, oceania, pangaea, mena, aegean, alps, amazonriver, amazonriverwide, arctic, baikal, beringstrait, betweentwoseas, blacksea, bosphorusstraits, deglaciatedantarctica, falklandislands, faroeislands, fourislands, gatewaytotheatlantic, gulfofstlawrence, halkidiki, hawaii, iceland, italia, japan, lemnos, lisbon, manicouagan, niledelta, passage, sanfrancisco, straitofgibraltar, straitofhormuz, surrounded, thebox, theboxplus, tourney1, tourney2, tourney3, tourney4, tradersdream, twolakes, worldrotated, yenisei, achiran, mars, milkyway, montreal, newyorkcity, pluto, reglaciatedantarctica (random per episode)
-- **Opponents:** 2 Easy bots
-- **Parallel envs:** 8
-- **Learning rate:** 0.00015
-- **Rollout steps:** 512
-- **Updates trained:** 2750
-- **Global steps:** 11264000
-- **Best mean reward:** 2761.6309762120245
-
-## Final Training Metrics
-
-- **Mean reward:** 1866.676198823452
-- **Mean episode length:** 35228.65
-- **Loss:** 17.78866958618164
+- **Observation dim:** 80 (16 player stats + 16 neighbors × 4 features)
+- **Action space:** MultiDiscrete [17 action types, 16 targets, 5 troop fractions]
+- **Maps:** plains, big_plains, world, giantworldmap, ocean_and_land, half_land_half_ocean (random per episode)
+- **Parallel envs:** 16
+- **Learning rate:** 1.5e-4 (constant)
+- **Rollout steps:** 1024
+- **Batch size:** 16,384
+- **Value function coefficient:** 0.5
+- **Updates trained:** 1550 (ongoing)
+
+## Reward Design (v13)
+
+Normalized elimination reward — total reward sums to +1.0 on a full win regardless of opponent count:
+- **Per-kill:** `+1/N` per opponent eliminated (N = starting opponents)
+- **Winner bonus:** remaining alive opponents credited as `aliveCount/N` when `game.getWinner()` fires
+- **Death penalty:** -1.0
+
+## Curriculum
+
+Win-rate-gated 12-stage curriculum advancing through Easy → Medium → Hard difficulty and 2 → 15 opponents. Stages advance only when rolling win rate exceeds per-stage threshold (75% down to 45%) over 200 episodes.
+
+## Eval Results
+
+- **Easy/2 opponents:** 100% win rate (20/20 games)
 
 ## Usage
 
@@ -39,7 +51,8 @@ from train import ActorCritic
 import torch
 
 model = ActorCritic(obs_dim=80, max_neighbors=16, hidden_sizes=[512, 512, 256])
-model.load_state_dict(torch.load("best_model.pt", weights_only=True))
+checkpoint = torch.load("best_model.pt", map_location="cpu", weights_only=False)
+model.load_state_dict(checkpoint["model_state_dict"])
 model.eval()
 ```