Raven - Revision history

Dsekercioglu: Update to Raven 3.59's tuning method

2021-08-02T09:54:17Z

Update to Raven 3.59's tuning method

Dsekercioglu: Forgot to preview :)

2021-06-08T15:10:59Z

Forgot to preview :)

Dsekercioglu: Add details about the current PyTorch Tuner

2021-06-08T15:10:23Z

Add details about the current PyTorch Tuner

Dsekercioglu: Update page

2021-06-05T16:48:36Z

Update page

Dsekercioglu: Update link

2020-10-03T13:01:49Z

Update link

Dsekercioglu: Surprised to see rankings change after 20k battles.

2020-09-26T09:09:49Z

Surprised to see rankings change after 20k battles.

Dsekercioglu: Raven page update

2020-08-05T21:57:29Z

Raven page update

Dsekercioglu: Raven page link update

2020-07-28T20:30:23Z

Raven page link update

Dsekercioglu: Raven page update

2020-07-28T12:25:21Z

Raven page update

Dsekercioglu: Raven page update

2020-07-27T13:26:55Z

Raven page update

@@ Line 60: / Line 60: @@
 : Improve the pytorch tuned targeting system
-; Pytorch Tuner
+; RTune
-: The current tuning system is very naive and rather experimental.
+: RTune is a tuning system that uses Gradient Descent to optimize the weighting parameters
-: The formula used for transforming the data points is ax + bx^2
+: It has been written in Rust in order to minimize the time spent parsing the files and using kd-trees
-: For each datapoint, Guess Factor pair, it finds the K closest and K furthest Guess Factors in the given match and saves corresponding weights.
+: The system starts by weighting every parameter 1.
-: Then the transformer is trained to minimize (NN(input) - NN(kClosest))^2
+: Then it goes through the pre-recorded battles, dynamically constructing the kd-tree.
-: The obvious flaw with this system is that the optimal solution would be making all weights 0.
+: Unlike [[BeepBoop]]'s training method, RTune uses the weights during search as well rather than only using it for distance.
-: This is prevented in a rather inelegant way:
+: After finding K nearest neighbors, the system weights each match by its distance to the search point and passes it through softmax in order to make sure the sum is 1.
-: All the A terms are normalized so that the sum of their absolute values is 1
+: The loss is calculated by 1 - sum of the weights that hit the enemy robot.
-: The B terms are clipped so that they can't be smaller than 0 so they can only increment the weights
+: Despite not matching the performance of Raven 3.57 due to lost/extra code, Raven 3.59 managed to gain 0.5 APS over the 3.58 version with a training time of 80 seconds.
 ; Does it have any [[White Whale]]s?

@@ Line 70: / Line 70: @@
 : The B terms are clipped so that they can't be smaller than 0 so they can only increment the weights
 : This also makes sure that all transformations are one to one formulas(for the better?).
 ; Does it have any [[White Whale]]s?

@@ Line 59: / Line 59: @@
 : Add a flattener that actually improves its scores against adaptive targeting.
 : Improve the pytorch tuned targeting system
 ; Does it have any [[White Whale]]s?

@@ Line 26: / Line 26: @@
 : [[Chase-san]] for the intercept method I used in my [[PPMEA]] calculations.
 : [[AW]] for giving me the idea of integrating the danger function in order to get the danger value for a given bot width.
 == Strategy ==
@@ Line 57: / Line 58: @@
 : Maybe a pre-trained movement or gun to use in the first ticks of the battle.
 : Add a flattener that actually improves its scores against adaptive targeting.
 ; Does it have any [[White Whale]]s?
 : Drifter has been crushing the latest versions.
 ; What other robot(s) is it based on?

@@ Line 14: / Line 14: @@
 ; Great, I want to try it. Where can I download it?
-: https://www.dropbox.com/s/eu9qpl2ukg17q0z/dsekercioglu.mega.Raven_3.46.jar?dl=1
+: https://www.dropbox.com/s/ln53uvb3ddxe4bv/dsekercioglu.mega.Raven_3.56j8.jar?dl=1
 ; How competitive is it?

@@ Line 17: / Line 17: @@
 ; How competitive is it?
-: Its best is 8th place.
+: Its best is 7th place but considering how close it is to Gilgalad, 7.5th would be the right word =)
 ; Credits

@@ Line 15: / Line 15: @@
 ; Great, I want to try it. Where can I download it?
-: https://www.dropbox.com/s/335p9lcdwgpagry/dsekercioglu.mega.Raven_3.19b.jar?dl=1
+: https://www.dropbox.com/s/v2w0tewmd0g1cxo/dsekercioglu.mega.Raven_3.21a.jar?dl=1
 ; How competitive is it?
-: Its best is 15.
+: Its best is 10th place.
 ; Credits

@@ Line 15: / Line 15: @@
 ; Great, I want to try it. Where can I download it?
-: https://www.dropbox.com/s/8v2x5cot0vekjk0/dsekercioglu.mega.Raven_3.17b.jar?dl=1
+: https://www.dropbox.com/s/335p9lcdwgpagry/dsekercioglu.mega.Raven_3.19b.jar?dl=1
 ; How competitive is it?
@@ Line 30: / Line 30: @@
 ; How does it [[Movement|move]]?
-: True Surfing.
+: A form of Go To Surfing.
 ; How does it fire?
@@ Line 51: / Line 52: @@
 ; What's next for your robot?
 : A proper versioning system so I don't keep accidentally releasing experimental versions.
-: Better Bullet Shadow Calculations.
+: Faster code so it doesn't run slow and doesn't skip turns.
 : Tuning the guns since they haven't been tuned since the first version.
 : Gun Heat Waves.
@@ Line 58: / Line 60: @@
 ; Does it have any [[White Whale]]s?
-: Raven's Anti-Surfer targeting is very weak. Combined with lack of adaptivity and good flattening in movement, other wave surfers become huge problem bots.
+: Raven doesn't seem to have any extreme white whales currently.
 ; What other robot(s) is it based on?