Update simplified__first_attack checkpoint metadata

metadata/total_config.py

@@ -21,11 +21,11 @@ exp_config = {
         'stop_value':
         10000000000,
         'n_evaluator_episode':
-        10,
+        20,
         'env_id':
         'simplified__first_attack',
         'battle_mode':
-        'self_play_mode',
+        'play_with_bot_mode',
         'battle_mode_in_simulation_env':
         'self_play_mode',
         'bot_action_type':
@@ -38,6 +38,10 @@ exp_config = {
         0,
         'prob_random_action_in_bot':
         0.0,
+        'collector_bot_mode_seat_swap':
+        True,
+        'bot_mode_live_seat':
+        1,
         'channel_last':
         False,
         'scale':
@@ -56,9 +60,13 @@ exp_config = {
             'custom_games_simplified.simplified__first_attack.envs.first_attack_env'
         ],
         'collector_env_num':
-        4,
+        32,
         'evaluator_env_num':
-        10
+        20,
+        'collector_bot_mode_live_seat_weights': {
+            '1': 0.25,
+            '2': 0.75
+        }
     },
     'policy': {
         'model': {
@@ -119,7 +127,7 @@ exp_config = {
                 'figure_path': None,
                 'cfg_type': 'InteractionSerialEvaluatorDict',
                 'stop_value': 10000000000,
-                'n_episode': 10
+                'n_episode': 20
             }
         },
         'other': {
@@ -166,13 +174,13 @@ exp_config = {
         'sampled_algo': False,
         'gumbel_algo': False,
         'mcts_ctree': True,
-        'collector_env_num': 4,
-        'evaluator_env_num': 10,
+        'collector_env_num': 32,
+        'evaluator_env_num': 20,
         'env_type': 'board_games',
         'action_type': 'varied_action_space',
-        'battle_mode': 'self_play_mode',
+        'battle_mode': 'play_with_bot_mode',
         'monitor_extra_statistics': True,
-        'game_segment_length': 12,
+        'game_segment_length': 5,
         'eval_offline': False,
         'calculate_dormant_ratio': False,
         'analysis_sim_norm': False,
@@ -182,9 +190,9 @@ exp_config = {
         'use_augmentation': False,
         'augmentation': ['shift', 'intensity'],
         'ignore_done': False,
-        'update_per_collect': 2,
+        'update_per_collect': 25,
         'replay_ratio': 0.25,
-        'batch_size': 128,
+        'batch_size': 256,
         'optim_type': 'Adam',
         'learning_rate': 0.003,
         'target_update_freq': 100,
@@ -192,11 +200,11 @@ exp_config = {
         'weight_decay': 0.0001,
         'momentum': 0.9,
         'grad_clip_value': 0.5,
-        'n_episode': 4,
+        'n_episode': 32,
         'num_segments': 8,
-        'num_simulations': 20,
+        'num_simulations': 50,
         'discount_factor': 1,
-        'td_steps': 12,
+        'td_steps': 5,
         'num_unroll_steps': 5,
         'reward_loss_weight': 1,
         'value_loss_weight': 0.25,
@@ -230,15 +238,19 @@ exp_config = {
         'type': 'efficientzero',
         'import_names': ['lzero.policy.efficientzero'],
         'model_path':
-        '/workspace/combinatorial_reasoning_post_training/models/simplified5_immediate_300k_continuation_20260518/round-01/simplified__first_attack/attempt-01_260518_231816/ckpt/envstep_130000.pth.tar',
+        '/home/molfetta/combinatorial_reasoning_post_training/models/main5_hf_bot_mode_200k_collector_tuned_20260522/simplified__first_attack/attempt-01_260522_142417/ckpt/envstep_100000.pth.tar',
         'reanalyze_ratio': 0.0,
-        'eval_freq': 100001,
+        'eval_freq': 150001,
         'replay_buffer_size': 50000,
         'best_ckpt_strategy': 'raw',
         'best_ckpt_ema_alpha': 0.3,
         'best_ckpt_min_episodes': 20,
         'battle_mode_in_simulation_env': 'self_play_mode',
         'eval_opponent_type': 'env_bot',
+        'fixed_bot_evaluator': {
+            'type': 'arena',
+            'seat_swap': True
+        },
         'previous_best_checkpoint': {
             'path': None,
             'selector': 'best',
@@ -252,6 +264,6 @@ exp_config = {
         'device': 'cuda'
     },
     'exp_name':
-    '/workspace/combinatorial_reasoning_post_training/models/simplified5_constant_lr_fixed_20260519/round-01/simplified__first_attack/attempt-01_260519_093543',
+    '/home/molfetta/combinatorial_reasoning_post_training/models/main5_hf_bot_mode_recovery_50k_20260522/simplified__first_attack/attempt-01_260522_170112',
     'seed': 0
 }