|
|
@@ -18,6 +18,7 @@ struct MarioNeuralTrainer: public Container
|
|
|
mario::GameplaySimulation simulator;
|
|
|
pika::pikaImgui::FileSelector currentMap;
|
|
|
pika::pikaImgui::FileSelector rezFile;
|
|
|
+ pika::pikaImgui::FileSelector graphFile;
|
|
|
|
|
|
//todo user can request imgui ids; shortcut manager context; allocators
|
|
|
static ContainerStaticInfo containerInfo()
|
|
|
@@ -64,6 +65,28 @@ struct MarioNeuralTrainer: public Container
|
|
|
requestedInfo.writeEntireFileBinary(rezFile.file, &deadSimulations[0].network, sizeof(deadSimulations[0].network));
|
|
|
}
|
|
|
|
|
|
+ if (graphFile.file[0] != '\0' && save)
|
|
|
+ {
|
|
|
+
|
|
|
+ ::pika::memory::pushCustomAllocatorsToStandard();
|
|
|
+ {
|
|
|
+ std::ofstream file(graphFile.file);
|
|
|
+
|
|
|
+ if (file.is_open())
|
|
|
+ {
|
|
|
+ for (auto &g : generations)
|
|
|
+ {
|
|
|
+ file << g.leastFit << " " << g.averageFit << " " << g.fitest << "\n";
|
|
|
+ }
|
|
|
+ };
|
|
|
+
|
|
|
+ file.close();
|
|
|
+ }
|
|
|
+ ::pika::memory::popCustomAllocatorsToStandard();
|
|
|
+
|
|
|
+ }
|
|
|
+
|
|
|
+ //we keep the first people unchanged
|
|
|
std::vector<SimulationNetwork> fittest;
|
|
|
fittest.push_back(deadSimulations[0]);
|
|
|
fittest.push_back(deadSimulations[1]);
|
|
|
@@ -79,10 +102,23 @@ struct MarioNeuralTrainer: public Container
|
|
|
simulations.push_back(i);
|
|
|
}
|
|
|
|
|
|
- while (simulations.size() < 1000)
|
|
|
+ //we also add some of the less fit people
|
|
|
+ for (int i = 0; i < 20; i++)
|
|
|
{
|
|
|
- int a = mario::getRandomInt(rng, 0, 4);
|
|
|
- int b = mario::getRandomInt(rng, 0, 4);
|
|
|
+ float a = mario::getRandomFloat(rng, 0, 1);
|
|
|
+ float b = mario::getRandomFloat(rng, 0, 1);
|
|
|
+
|
|
|
+ a *= b; //we bias the people to keep the fitter ones
|
|
|
+
|
|
|
+ fittest.push_back(deadSimulations[5 + a * (deadSimulations.size()/2)]);
|
|
|
+ }
|
|
|
+
|
|
|
+ int size = fittest.size() - 1;
|
|
|
+
|
|
|
+ while (simulations.size() < 10000)
|
|
|
+ {
|
|
|
+ int a = mario::getRandomInt(rng, 0, size);
|
|
|
+ int b = mario::getRandomInt(rng, 0, size);
|
|
|
|
|
|
SimulationNetwork newNetwork = fittest[a];
|
|
|
newNetwork.network.combine(rng, fittest[b].network);
|
|
|
@@ -91,7 +127,7 @@ struct MarioNeuralTrainer: public Container
|
|
|
newNetwork.player.p.position.position = {1,1};
|
|
|
newNetwork.player.p.lastPos = {1,1};
|
|
|
|
|
|
-
|
|
|
+ //combine the 2 individuals
|
|
|
if (mario::getRandomChance(rng, 0.25))
|
|
|
{
|
|
|
int type = mario::getRandomInt(rng, 0, 100);
|
|
|
@@ -133,6 +169,7 @@ struct MarioNeuralTrainer: public Container
|
|
|
currentMap.setInfo("Training map", PIKA_RESOURCES_PATH "/mario", {".mario"});
|
|
|
|
|
|
rezFile.setInfo("Neural network file", PIKA_RESOURCES_PATH "/mario", {".neural"});
|
|
|
+ graphFile.setInfo("Graph File", PIKA_RESOURCES_PATH "/mario", {".txt"});
|
|
|
|
|
|
|
|
|
if (commandLineArgument.size() != 0)
|
|
|
@@ -170,6 +207,14 @@ struct MarioNeuralTrainer: public Container
|
|
|
return v - (int)v;
|
|
|
}
|
|
|
|
|
|
+ struct Generation
|
|
|
+ {
|
|
|
+ float fitest = 0;
|
|
|
+ float leastFit = 0;
|
|
|
+ float averageFit = 0;
|
|
|
+ };
|
|
|
+
|
|
|
+ std::vector<Generation> generations;
|
|
|
|
|
|
bool update(pika::Input input, pika::WindowState windowState,
|
|
|
RequestedContainerInfo &requestedInfo)
|
|
|
@@ -182,9 +227,15 @@ struct MarioNeuralTrainer: public Container
|
|
|
float maxFit = 0;
|
|
|
for (int i = 0; i < simulations.size(); i++)
|
|
|
{
|
|
|
- if (simulations[i].player.p.position.position.x - simulations[i].player.jumpCount / 10.f > maxFit)
|
|
|
+ //if (simulations[i].player.p.position.position.x - simulations[i].player.jumpCount / 10.f > maxFit)
|
|
|
+ //{
|
|
|
+ // maxFit = simulations[i].player.p.position.position.x - simulations[i].player.jumpCount/10.f;
|
|
|
+ // maxIndex = i;
|
|
|
+ //}
|
|
|
+
|
|
|
+ if (simulations[i].player.maxFit > maxFit)
|
|
|
{
|
|
|
- maxFit = simulations[i].player.p.position.position.x - simulations[i].player.jumpCount/10.f;
|
|
|
+ maxFit = simulations[i].player.maxFit;
|
|
|
maxIndex = i;
|
|
|
}
|
|
|
}
|
|
|
@@ -211,7 +262,7 @@ struct MarioNeuralTrainer: public Container
|
|
|
|
|
|
if (fast)
|
|
|
{
|
|
|
- newDelta = 1.f / 50.f;
|
|
|
+ newDelta = 1.f / 20.f;
|
|
|
}
|
|
|
|
|
|
for (int i = 0; i < simulations.size(); i++)
|
|
|
@@ -231,6 +282,32 @@ struct MarioNeuralTrainer: public Container
|
|
|
|
|
|
if (simulations.empty())
|
|
|
{
|
|
|
+
|
|
|
+ Generation result = {};
|
|
|
+
|
|
|
+ result.fitest = 0;
|
|
|
+ result.leastFit = 999999999999;
|
|
|
+ result.averageFit = 0;
|
|
|
+
|
|
|
+ for (auto &g : deadSimulations)
|
|
|
+ {
|
|
|
+ if (g.player.maxFit > result.fitest)
|
|
|
+ {
|
|
|
+ result.fitest = g.player.maxFit;
|
|
|
+ }
|
|
|
+
|
|
|
+ if (g.player.maxFit < result.leastFit)
|
|
|
+ {
|
|
|
+ result.leastFit = g.player.maxFit;
|
|
|
+ }
|
|
|
+
|
|
|
+ result.averageFit += g.player.maxFit;
|
|
|
+ }
|
|
|
+
|
|
|
+ result.averageFit /= deadSimulations.size();
|
|
|
+
|
|
|
+ generations.push_back(result);
|
|
|
+
|
|
|
recreateGenerations(requestedInfo);
|
|
|
}
|
|
|
|
|
|
@@ -260,7 +337,8 @@ struct MarioNeuralTrainer: public Container
|
|
|
ImGui::Checkbox("fixed framerate", &fast);
|
|
|
|
|
|
rezFile.run(2);
|
|
|
- ImGui::SameLine();
|
|
|
+ graphFile.run(3);
|
|
|
+ //ImGui::SameLine();
|
|
|
ImGui::Checkbox("save", &save);
|
|
|
|
|
|
}
|
vlod