1:0 against kenria Booyaa

ultimatetictactoe.py

@@ -141,40 +141,47 @@ class Model(nn.Module):
     def __init__(self):
         super().__init__()
 
-        self.smolChan = 12
-        self.compChan = 7
-
         self.smol = nn.Sequential(
             nn.Conv2d(
                 in_channels=1,
-                out_channels=self.smolChan,
+                out_channels=24,
                 kernel_size=(3,3),
                 stride=3,
                 padding=0,
             ),
             nn.ReLU()
         )
-        self.big = nn.Sequential(
-            nn.Linear(self.smolChan*9, self.compChan),
-            #nn.ReLU(),
-            #nn.Linear(self.compChan, 1),
+        self.comb = nn.Sequential(
+            nn.Conv1d(
+                in_channels=24,
+                out_channels=8,
+                kernel_size=1,
+                stride=1,
+                padding=0,
+            ),
+            nn.ReLU()
+        )
+        self.out = nn.Sequential(
+            nn.Linear(9*8, 32),
             nn.ReLU(),
-            nn.Linear(self.compChan, 3),
+            nn.Linear(32, 8),
             nn.ReLU(),
-            nn.Linear(3, 1),
+            nn.Linear(8, 1),
             nn.Sigmoid()
         )
 
     def forward(self, x):
         x = torch.reshape(x, (1,9,9))
         x = self.smol(x)
-        x = torch.reshape(x, (self.smolChan*9,))
-        y = self.big(x)
+        x = torch.reshape(x, (24,9))
+        x = self.comb(x)
+        x = torch.reshape(x, (-1,))
+        y = self.out(x)
         return y
 
 if __name__=="__main__":
     run = NeuralRuntime(TTTState())
-    run.game(None, 4)
+    run.game([0,1], 4)
 
-    trainer = Trainer(TTTState())
-    trainer.train()
+    #trainer = Trainer(TTTState())
+    #trainer.train()

vacuumDecay.py

@@ -11,6 +11,7 @@ from threading import Event
 from queue import PriorityQueue, Empty
 from dataclasses import dataclass, field
 from typing import Any
+import random
 
 class Action():
     # Should hold the data representing an action
@@ -388,6 +389,7 @@ class Runtime():
     def __init__(self, initState):
         universe = QueueingUniverse()
         self.head = Node(initState, universe = universe)
+        _ = self.head.childs
         universe.newOpen(self.head)
 
     def spawnWorker(self):
@@ -460,9 +462,9 @@ class Trainer(Runtime):
         self.rootNode = Node(initState, universe = self.universe)
         self.terminal = None
 
-    def buildDatasetFromModel(self, model, depth=4, refining=True):
+    def buildDatasetFromModel(self, model, depth=4, refining=True, exacity=5):
         print('[*] Building Timeline')
-        term = self.linearPlay(model, calcDepth=depth)
+        term = self.linearPlay(model, calcDepth=depth, exacity=exacity)
         if refining:
             print('[*] Refining Timeline')
             self.fanOut(term, depth=depth+1)
@@ -475,7 +477,7 @@ class Trainer(Runtime):
             head = head.parent
         head.forceStrong(depth)
 
-    def linearPlay(self, model, calcDepth=7, verbose=True):
+    def linearPlay(self, model, calcDepth=7, exacity=5, verbose=True):
         head = self.rootNode
         self.universe.model = model
         while head.getWinner()==None:
@@ -490,7 +492,10 @@ class Trainer(Runtime):
             for c in head.childs:
                 opts.append((c, c.getStrongFor(head.curPlayer)))
             opts.sort(key=lambda x: x[1])
-            ind = int(pow(random.random(),5)*(len(opts)-1))
+            if exacity >= 10:
+                ind = 0
+            else:
+                ind = int(pow(random.random(),exacity)*(len(opts)-1))
             head = opts[ind][0]
         print('')
         return head
@@ -499,16 +504,20 @@ class Trainer(Runtime):
         head = term
         while True:
             yield head
+            if len(head.childs):
+                yield random.choice(head.childs)
             if head.parent == None:
                 return
             head = head.parent
 
-    def trainModel(self, model, lr=0.01, cut=0.01, calcDepth=4):
+    def trainModel(self, model, lr=0.00005, cut=0.01, calcDepth=4, exacity=5):
         loss_func = nn.MSELoss()
         optimizer = optim.Adam(model.parameters(), lr)
-        term = self.buildDatasetFromModel(model, depth=calcDepth)
+        term = self.buildDatasetFromModel(model, depth=calcDepth, exacity=exacity)
+        print('[*] Conditioning Brain...')
         for r in range(64):
             loss_sum = 0
+            lLoss = 0
             zeroLen = 0
             for i, node in enumerate(self.timelineIter(term)):
                 for p in range(self.rootNode.playersNum):
@@ -524,19 +533,20 @@ class Trainer(Runtime):
                         zeroLen+=1
                     if zeroLen == 5:
                         break
-            print(loss_sum/i)
-            if loss_sum/i < cut:
+            #print(loss_sum/i)
+            if r > 16 and (loss_sum/i < cut or lLoss == loss_sum):
                 return
+            lLoss = loss_sum
 
-    def main(self, model=None, gens=64):
+    def main(self, model=None, gens=1024, startGen=12):
         newModel = False
         if model==None:
             newModel = True
             model = self.rootNode.state.getModel()
         self.universe.scoreProvider = ['neural','naive'][newModel]
-        for gen in range(gens):
+        for gen in range(startGen, startGen+gens):
             print('[#####] Gen '+str(gen)+' training:')
-            self.trainModel(model, calcDepth=3)
+            self.trainModel(model, calcDepth=min(5,3+int(gen/16)), exacity=int(gen/3+1))
             self.universe.scoreProvider = 'neural'
             torch.save(model.state_dict(), 'brains/uttt.pth')
 
@@ -544,4 +554,4 @@ class Trainer(Runtime):
         model = self.rootNode.state.getModel()
         model.load_state_dict(torch.load('brains/uttt.pth'))
         model.eval()
-        self.main(model)
+        self.main(model, startGen=0)