feat: 优化DQN训练配置并支持Dueling网络评估

- 将学习率调整为5e-5，批次大小增加到64，经验回放缓冲区扩大到500,000
- 启用优先经验回放，调整目标网络更新频率为1000步
- 评估时使用Dueling网络架构，训练时评估模式的ε设为0
- 更新评估结果以反映配置改进后的性能变化

强化学习个人项目报告（Atari 游戏方向）/evaluation_results.json

@@ -2,79 +2,79 @@
   {
     "model": "models/dqn_step_100000.pt",
     "step": 100000,
-    "avg_reward": 17.8,
-    "std_reward": 5.2306787322488075
+    "avg_reward": 20.9,
+    "std_reward": 11.235657524150511
   },
   {
     "model": "models/dqn_step_200000.pt",
     "step": 200000,
-    "avg_reward": 14.0,
-    "std_reward": 6.603029607687671
+    "avg_reward": 23.05,
+    "std_reward": 8.361967471833408
   },
   {
     "model": "models/dqn_step_400000.pt",
     "step": 400000,
-    "avg_reward": 16.4,
-    "std_reward": 4.24735211631906
+    "avg_reward": 14.5,
+    "std_reward": 9.418067742376884
   },
   {
     "model": "models/dqn_step_600000.pt",
     "step": 600000,
-    "avg_reward": 19.0,
-    "std_reward": 4.123105625617661
+    "avg_reward": 22.0,
+    "std_reward": 11.218288639538564
   },
   {
     "model": "models/dqn_step_800000.pt",
     "step": 800000,
-    "avg_reward": 13.2,
-    "std_reward": 3.944616584663204
+    "avg_reward": 24.95,
+    "std_reward": 11.617766566771772
   },
   {
     "model": "models/dqn_step_1000000.pt",
     "step": 1000000,
-    "avg_reward": 15.7,
-    "std_reward": 4.960846701924985
+    "avg_reward": 32.65,
+    "std_reward": 14.44134689009304
   },
   {
     "model": "models/dqn_step_1200000.pt",
     "step": 1200000,
-    "avg_reward": 18.4,
-    "std_reward": 6.216108107168021
+    "avg_reward": 21.5,
+    "std_reward": 12.188108959145385
   },
   {
     "model": "models/dqn_step_1400000.pt",
     "step": 1400000,
-    "avg_reward": 14.2,
-    "std_reward": 4.467661580737736
+    "avg_reward": 16.15,
+    "std_reward": 13.950000000000001
   },
   {
     "model": "models/dqn_step_1600000.pt",
     "step": 1600000,
-    "avg_reward": 17.8,
-    "std_reward": 4.686149805543993
+    "avg_reward": 30.5,
+    "std_reward": 15.55795616396961
   },
   {
     "model": "models/dqn_step_1800000.pt",
     "step": 1800000,
-    "avg_reward": 21.5,
-    "std_reward": 4.984977432245807
+    "avg_reward": 34.25,
+    "std_reward": 16.40464873138099
   },
   {
     "model": "models/dqn_step_2000000.pt",
     "step": 2000000,
-    "avg_reward": 14.6,
-    "std_reward": 5.276362383309167
+    "avg_reward": 23.65,
+    "std_reward": 14.120995007434852
   },
   {
     "model": "models/dqn_best.pt",
     "step": -1,
-    "avg_reward": 19.9,
-    "std_reward": 6.920260110718383
+    "avg_reward": 16.6,
+    "std_reward": 9.606768447297977
   },
   {
     "model": "models/dqn_final.pt",
     "step": -2,
-    "avg_reward": 11.3,
-    "std_reward": 3.3778691508109073
+    "avg_reward": 20.2,
+    "std_reward": 11.185258155268478
   }
 ]