Added feature telling if back is possible or not

Networks.py

@@ -2,6 +2,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 import Features
+import Transition
 
 ################################################################################
 def createNetwork(name, dicts, outputSizes, incremental) :
@@ -51,7 +52,7 @@ class BaseNet(nn.Module):
       self.add_module("emb_"+name, nn.Embedding(len(dicts.dicts[name]), self.embSize))
     self.fc1 = nn.Linear(self.inputSize * self.embSize, hiddenSize)
     for i in range(len(outputSizes)) :
-      self.add_module("output_"+str(i), nn.Linear(hiddenSize, outputSizes[i]))
+      self.add_module("output_"+str(i), nn.Linear(hiddenSize+1, outputSizes[i]))
     self.dropout = nn.Dropout(0.3)
 
     self.apply(self.initWeights)
@@ -61,6 +62,9 @@ class BaseNet(nn.Module):
 
   def forward(self, x) :
     embeddings = []
+    canBack = x[...,0:1]
+    x = x[...,1:]
+
     for i in range(len(self.columns)) :
       embeddings.append(getattr(self, "emb_"+self.columns[i])(x[...,i*self.nbTargets:(i+1)*self.nbTargets]))
     y = torch.cat(embeddings,-1).view(x.size(0),-1)
@@ -79,6 +83,7 @@ class BaseNet(nn.Module):
       curIndex = curIndex+self.suffixSize
     y = self.dropout(y)
     y = F.relu(self.dropout(self.fc1(y)))
+    y = torch.cat([y,canBack], 1)
     y = getattr(self, "output_"+str(self.state))(y)
     return y
 
@@ -95,7 +100,8 @@ class BaseNet(nn.Module):
     historyValues = Features.extractHistoryFeatures(dicts, config, self.historyNb)
     prefixValues = Features.extractPrefixFeatures(dicts, config, self.prefixSize)
     suffixValues = Features.extractSuffixFeatures(dicts, config, self.suffixSize)
-    return torch.cat([colsValues, historyValues, prefixValues, suffixValues])
+    backAction = torch.ones(1, dtype=torch.int) if Transition.Transition("BACK 1").appliable(config) else torch.zeros(1, dtype=torch.int)
+    return torch.cat([backAction, colsValues, historyValues, prefixValues, suffixValues])
 ################################################################################
 
 ################################################################################
@@ -121,7 +127,7 @@ class SemiNet(nn.Module):
     self.fc1 = nn.Linear(self.inputSize * self.embSize, hiddenSize)
     for i in range(len(outputSizes)) :
       self.add_module("output_hidden_"+str(i), nn.Linear(hiddenSize, hiddenSize))
-      self.add_module("output_"+str(i), nn.Linear(hiddenSize, outputSizes[i]))
+      self.add_module("output_"+str(i), nn.Linear(hiddenSize+1, outputSizes[i]))
     self.dropout = nn.Dropout(0.3)
 
     self.apply(self.initWeights)
@@ -131,6 +137,8 @@ class SemiNet(nn.Module):
 
   def forward(self, x) :
     embeddings = []
+    canBack = x[...,0:1]
+    x = x[...,1:]
     for i in range(len(self.columns)) :
       embeddings.append(getattr(self, "emb_"+self.columns[i])(x[...,i*self.nbTargets:(i+1)*self.nbTargets]))
     y = torch.cat(embeddings,-1).view(x.size(0),-1)
@@ -150,6 +158,7 @@ class SemiNet(nn.Module):
     y = self.dropout(y)
     y = F.relu(self.dropout(self.fc1(y)))
     y = self.dropout(getattr(self, "output_hidden_"+str(self.state))(y))
+    y = torch.cat([y,canBack], 1)
     y = getattr(self, "output_"+str(self.state))(y)
     return y
 
@@ -166,7 +175,8 @@ class SemiNet(nn.Module):
     historyValues = Features.extractHistoryFeatures(dicts, config, self.historyNb)
     prefixValues = Features.extractPrefixFeatures(dicts, config, self.prefixSize)
     suffixValues = Features.extractSuffixFeatures(dicts, config, self.suffixSize)
-    return torch.cat([colsValues, historyValues, prefixValues, suffixValues])
+    backAction = torch.ones(1, dtype=torch.int) if Transition.Transition("BACK 1").appliable(config) else torch.zeros(1, dtype=torch.int)
+    return torch.cat([backAction, colsValues, historyValues, prefixValues, suffixValues])
 ################################################################################
 
 ################################################################################