add vdac map

xuance/configs/vdac/sc2/1c3s5z.yaml

@@ -0,0 +1,68 @@
+agent: "VDAC"
+env_name: "StarCraft2"
+env_id: "1c3s5z"
+fps: 15
+policy: "Categorical_MAAC_Policy_Share"
+representation: "Basic_RNN"
+vectorize: "Subproc_StarCraft2"
+runner: "StarCraft2_Runner"
+
+# recurrent settings for Basic_RNN representation
+use_recurrent: True
+rnn: "GRU"
+recurrent_layer_N: 1
+fc_hidden_sizes: [64, ]
+recurrent_hidden_size: 64
+N_recurrent_layers: 1
+dropout: 0
+normalize: "LayerNorm"
+initialize: "orthogonal"
+gain: 0.01
+
+actor_hidden_size: []
+critic_hidden_size: []
+activation: "ReLU"
+
+mixer: "QMIX"  # choices: VDN (sum), QMIX (monotonic)
+hidden_dim_mixing_net: 32  # hidden units of mixing network
+hidden_dim_hyper_net: 64  # hidden units of hyper network
+
+seed: 1
+parallels: 8
+n_size: 8
+n_epoch: 1
+n_minibatch: 1
+learning_rate: 0.0007  # 7e-4
+weight_decay: 0
+
+vf_coef: 1.0
+ent_coef: 0.01
+target_kl: 0.25
+clip_range: 0.2
+clip_type: 1  # Gradient clip for Mindspore: 0: ms.ops.clip_by_value; 1: ms.nn.ClipByNorm()
+gamma: 0.99  # discount factor
+
+# tricks
+use_linear_lr_decay: False  # if use linear learning rate decay
+end_factor_lr_decay: 0.5
+use_grad_norm: True  # gradient normalization
+max_grad_norm: 10.0
+use_value_clip: True  # limit the value range
+value_clip_range: 0.2
+use_value_norm: True  # use running mean and std to normalize rewards.
+use_huber_loss: True  # True: use huber loss; False: use MSE loss.
+huber_delta: 10.0
+use_advnorm: True  # use advantage normalization.
+use_gae: True  # use GAE trick to calculate returns.
+gae_lambda: 0.95
+
+start_training: 1
+running_steps: 2000000
+train_per_step: True
+training_frequency: 1
+
+test_steps: 10000
+eval_interval: 10000
+test_episode: 16
+log_dir: "./logs/vdac/"
+model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/25m.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 5000000  # 5M
 training_frequency: 1
 
-eval_interval: 50000
+eval_interval: 25000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/2m_vs_1z.yaml

@@ -62,7 +62,7 @@ train_per_step: True
 training_frequency: 1
 
 test_steps: 10000
-eval_interval: 10000
-test_episode: 5
+eval_interval: 5000
+test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/2s3z.yaml

@@ -0,0 +1,68 @@
+agent: "VDAC"
+env_name: "StarCraft2"
+env_id: "2s3z"
+fps: 15
+policy: "Categorical_MAAC_Policy_Share"
+representation: "Basic_RNN"
+vectorize: "Subproc_StarCraft2"
+runner: "StarCraft2_Runner"
+
+# recurrent settings for Basic_RNN representation
+use_recurrent: True
+rnn: "GRU"
+recurrent_layer_N: 1
+fc_hidden_sizes: [64, ]
+recurrent_hidden_size: 64
+N_recurrent_layers: 1
+dropout: 0
+normalize: "LayerNorm"
+initialize: "orthogonal"
+gain: 0.01
+
+actor_hidden_size: []
+critic_hidden_size: []
+activation: "ReLU"
+
+mixer: "QMIX"  # choices: VDN (sum), QMIX (monotonic)
+hidden_dim_mixing_net: 32  # hidden units of mixing network
+hidden_dim_hyper_net: 64  # hidden units of hyper network
+
+seed: 1
+parallels: 8
+n_size: 8
+n_epoch: 1
+n_minibatch: 1
+learning_rate: 0.0007  # 7e-4
+weight_decay: 0
+
+vf_coef: 1.0
+ent_coef: 0.01
+target_kl: 0.25
+clip_range: 0.2
+clip_type: 1  # Gradient clip for Mindspore: 0: ms.ops.clip_by_value; 1: ms.nn.ClipByNorm()
+gamma: 0.99  # discount factor
+
+# tricks
+use_linear_lr_decay: False  # if use linear learning rate decay
+end_factor_lr_decay: 0.5
+use_grad_norm: True  # gradient normalization
+max_grad_norm: 10.0
+use_value_clip: True  # limit the value range
+value_clip_range: 0.2
+use_value_norm: True  # use running mean and std to normalize rewards.
+use_huber_loss: True  # True: use huber loss; False: use MSE loss.
+huber_delta: 10.0
+use_advnorm: True  # use advantage normalization.
+use_gae: True  # use GAE trick to calculate returns.
+gae_lambda: 0.95
+
+start_training: 1
+running_steps: 2000000
+train_per_step: True
+training_frequency: 1
+
+test_steps: 10000
+eval_interval: 10000
+test_episode: 16
+log_dir: "./logs/vdac/"
+model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/3m.yaml

@@ -55,7 +55,7 @@ start_training: 1
 running_steps: 1000000
 training_frequency: 1
 
-eval_interval: 10000
+eval_interval: 5000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/5m_vs_6m.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 10000000  # 10M
 training_frequency: 1
 
-eval_interval: 100000
+eval_interval: 50000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/8m.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 1000000  # 1M
 training_frequency: 1
 
-eval_interval: 10000
+eval_interval: 5000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/8m_vs_9m.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 10000000  # 10M
 training_frequency: 1
 
-eval_interval: 100000
+eval_interval: 50000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/MMM2.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 10000000  # 10M
 training_frequency: 1
 
-eval_interval: 100000
+eval_interval: 50000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"

xuance/configs/vdac/sc2/corridor.yaml

@@ -62,7 +62,7 @@ start_training: 1
 running_steps: 10000000  # 10M
 training_frequency: 1
 
-eval_interval: 100000
+eval_interval: 50000
 test_episode: 16
 log_dir: "./logs/vdac/"
 model_dir: "./models/vdac/"