SirStone/SirRoboGarage
1
0
Fork 0
Code Issues 28 Pull Requests Actions Projects Releases Wiki Activity

fix(PPO_Bot): radar oscillation, Adam persistence, checkpoint order, channel race

Browse Source 
- enemy_tracker: toggle lastOvershootDir each tick; make getRadarTurnRate take var tracker
- training: remove threadvar Adam globals; pass adamStates as var param to ppoUpdate; export ACAdamStates
- PPO_Bot: carry ACAdamStates through TrainingArgs/TrainingResult; drop trainingDone bool and Lock — use resultChan.tryRecv() directly as synchronisation
- weights: sort checkpoint dirs newest-first by mtime instead of hardcoded order
- tests/test_training: pass explicit ACAdamStates to ppoUpdate

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Davide Cappellini
2026-08-16 15:42:06 +02:00
parent 473d67f644
commit aea0724d3a
5 changed files with 74 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files
PPO_Bot/tests/test_training.nim
+2 -1
View File
@@ -86,7 +86,8 @@ block testPpoUpdate:
let v = ac.criticForward(s)
buf.add(Transition(state: s, action: a, logProb: lp, reward: 0.1'f32, value: v))
ppoUpdate(ac, buf, lastValue = 0.0'f32, epochs = 2, miniBatchSize = 5)
var adam: ACAdamStates
ppoUpdate(ac, buf, lastValue = 0.0'f32, adamStates = adam, epochs = 2, miniBatchSize = 5)
# Weights should have changed — compare flattened
let n = ac.actor.w1.shape[0] * ac.actor.w1.shape[1]