kono/refo
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

add new results of needed timestamps in total

Browse Source
This commit is contained in:
kono
2020-04-04 17:14:12 +02:00
parent a40e279f48
commit a08b8160a3
12 changed files with 113 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/main/java/core/RNG.java
+3 -3
View File
@@ -17,11 +17,12 @@ public class RNG {
private static Random rng;
private static Random rngEnv;
private static int seed = 123;
private static int envSeed = 13;
static {
rng = new Random();
rng.setSeed(seed);
rngEnv = new Random();
rngEnv.setSeed(seed);
rngEnv.setSeed(13);
}
public static Random getRandom() {
@@ -34,7 +35,6 @@ public class RNG {
public static void setSeed(int seed){
RNG.seed = seed;
rng.setSeed(seed);
rngEnv = new Random();
rngEnv.setSeed(seed);
rngEnv.setSeed(13);
}
}
src/main/java/core/algo/td/QLearningOffPolicyTDControl.java
+8 -8
View File
@@ -32,6 +32,7 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
@Override
protected void nextEpisode() {
State state = environment.reset();
try {
Thread.sleep(delay);
@@ -72,19 +73,19 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
}*/
if(reward == Reward.FOOD_DROP_DOWN_SUCCESS){
if(reward == Reward.FOOD_DROP_DOWN_SUCCESS) {
foodCollected++;
foodTimestampsTotal += timestampTilFood;
if(foodCollected % 1000 == 0){
System.out.println(foodTimestampsTotal / 1000f + " " + timestampCurrentEpisode);
File file = new File(ContinuousAnt.FILE_NAME);
//System.out.println(foodCollected + " " + timestampCurrentEpisode);
File file = new File(ContinuousAnt.FILE_NAME);
if(foodCollected % 1000 == 0) {
try {
Files.writeString(Path.of(file.getPath()), foodTimestampsTotal/1000f +",", StandardOpenOption.APPEND);
Files.writeString(Path.of(file.getPath()), timestampCurrentEpisode + ",", StandardOpenOption.APPEND);
} catch (IOException e) {
e.printStackTrace();
}
foodTimestampsTotal = 0;
}
foodTimestampsTotal = 0;
if(foodCollected == 1000){
((EpsilonGreedyPolicy<A>) this.policy).setEpsilon(0.15f);
}
@@ -98,8 +99,7 @@ public class QLearningOffPolicyTDControl<A extends Enum> extends EpisodicLearnin
System.out.println("final 0 expl");
((EpsilonGreedyPolicy<A>) this.policy).setEpsilon(0.00f);
}
if(foodCollected == 15000){
File file = new File(ContinuousAnt.FILE_NAME);
if(foodCollected == 30000) {
try {
Files.writeString(Path.of(file.getPath()), "\n", StandardOpenOption.APPEND);
} catch (IOException e) {