Fix tic tac toe action to string

werner-duvaud · werner-duvaud · commit 07ef10203e6c · 2020-09-07T10:30:59.000+02:00
diff --git a/games/tictactoe.py b/games/tictactoe.py
@@ -234,7 +234,7 @@ def action_to_string(self, action_number):
         Returns:
             String representing the action.
         """
-        row = 3 - action_number // 3
+        row = action_number // 3 + 1
         col = action_number % 3 + 1
         return f"Play row {row}, column {col}"
 
diff --git a/muzero.py b/muzero.py
@@ -634,7 +634,7 @@ def hyperparameter_search(
                 lr_init = nevergrad.p.Log(a_min=0.0001, a_max=0.1)
                 discount = nevergrad.p.Log(lower=0.95, upper=0.9999)
                 parametrization = nevergrad.p.Dict(
-                    lr_init=lr_init, discount=discount, training_steps=500
+                    lr_init=lr_init, discount=discount
                 )
                 best_hyperparameters = hyperparameter_search(
                     game_name, parametrization, budget, parallel_experiments, 20

Original file line number	Diff line number	Diff line change
`@@ -634,7 +634,7 @@ def hyperparameter_search(`
`634`	`634`	`lr_init = nevergrad.p.Log(a_min=0.0001, a_max=0.1)`
`635`	`635`	`discount = nevergrad.p.Log(lower=0.95, upper=0.9999)`
`636`	`636`	`parametrization = nevergrad.p.Dict(`
`637`		`- lr_init=lr_init, discount=discount, training_steps=500`
	`637`	`+ lr_init=lr_init, discount=discount`
`638`	`638`	`)`
`639`	`639`	`best_hyperparameters = hyperparameter_search(`
`640`	`640`	`game_name, parametrization, budget, parallel_experiments, 20`