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

Commit Graph

  • 1ea84f5b6e STRAFE: draw the heat field, real bullet danger, corrected ring comment SirStone 2026-09-25 23:25:34 +02:00
  • dc071b83f3 ModularBot: [result] round/battle outcome log lines (TR_RESULT_LOG, default on) SirStone 2026-09-25 22:50:46 +02:00
  • a50c0125d5 STRAFE movement: body pinned perpendicular to the threat, reversals by sign flip SirStone 2026-09-25 22:38:52 +02:00
  • 99cf9e5c82 tfil heat/pillar A/B: record the owner's decision to keep the virtual pillar removed SirStone 2026-09-25 22:24:47 +02:00
  • 48f38b80e7 TFIL heat-time + virtual pillar: live A/B (6 arms x 70 rounds) - neither change beats the pre-change mover SirStone 2026-09-25 22:21:41 +02:00
  • f58d65d2e8 TMComposites gate: per-gun confidence faithful for 3 guns; no pair composes SirStone 2026-09-25 22:02:28 +02:00
  • d0750ab020 TFIL: remove the virtual centre pillar from the shipped default; register j102 env reads SirStone 2026-09-25 22:02:00 +02:00
  • fca899376e TFIL: time-indexed bullet heat (TR_TFIL_HEAT_TIME, DEFAULT OFF) SirStone 2026-09-25 21:47:54 +02:00
  • 4270136948 docs: correct counted-SBC decay amortised cost figure SirStone 2026-09-25 08:43:48 +02:00
  • d85ff53d34 State-window gate: a temporal window of wave-relative states does NOT beat a single state SirStone 2026-09-25 08:43:36 +02:00
  • 40ba96f649 BitBrain SBC: counted mode + global decay (forgetting, probabilities) SirStone 2026-09-25 08:39:10 +02:00
  • 39e06719fb Campaign phase 2 ledger: live lead-gain sweep (gains above 1.0 do NOT beat Pattern) SirStone 2026-09-25 00:24:45 +02:00
  • 2747ebd323 BitBrain: TR_BITBRAIN_GAINS env knob (candidate set + fixed-gain degenerate) SirStone 2026-09-25 00:15:27 +02:00
  • c305ef4212 BitBrain campaign phase 1: the missing gain sweep and BitBrain as a lead-gain corrector SirStone 2026-09-25 00:10:19 +02:00
  • 140fe2519a HeadOn (no-lead) vs Pattern LIVE at long range: clean negative, offline ruler killed SirStone 2026-09-24 23:52:49 +02:00
  • a82c864c60 bitbrain campaign phase 0: offline prediction-quality ruler and the bar SirStone 2026-09-24 23:38:45 +02:00
  • 32a5e72fac Gun mixing (TMHorizon+BitBrain) vs DrussGT: clean negative, no dodge disruption SirStone 2026-09-24 23:05:12 +02:00
  • d93ce444c0 BitBrain verdict: clean negative at 30 runs/arm; analyzer gets MC + Mann-Whitney + MDE SirStone 2026-09-24 23:00:20 +02:00
  • f91e121965 lead capture by range: we under-lead (0.40->0.135) — a RANGE effect, not a power one SirStone 2026-09-24 22:47:18 +02:00
  • 795a0e59fe BitBrain gun (id 16): Pattern-relative ADE+SBC aim corrector, default off SirStone 2026-09-24 22:25:04 +02:00
  • 1fec87537d gitignore: keep the BitBrain primary-source zip local (explicit path, not a pattern) SirStone 2026-09-24 22:09:36 +02:00
  • 17c50159ba tools/ab: reusable A/B runner + analyzer (frozen-HEAD build, exact permutation test, liveness check) SirStone 2026-09-24 22:07:55 +02:00
  • e670788eee env_reference: the ring mover was NEVER measured offline - correct a false label SirStone 2026-09-24 21:44:17 +02:00
  • e40c8493a6 Offline harness: audited, calibrated against live, and one real bug fixed SirStone 2026-09-24 21:40:38 +02:00
  • f41cd08718 BitBrain gate test: fine-grained aim correction vs naive + Pattern (offline) SirStone 2026-09-24 21:23:29 +02:00
  • 1adefaba26 DrussGT dodge vs fired power: no movement response once range is controlled SirStone 2026-09-24 21:22:38 +02:00
  • 77e6dace01 BitBrain: generic clean-room ADE + SBC library with MNIST acceptance SirStone 2026-09-24 20:52:45 +02:00
  • cc11ede824 TFIL: measure the commitment A/B on 490 live DrussGT rounds - no effect SirStone 2026-09-24 20:24:34 +02:00
  • c8b2a8c6c5 docs: reverse-engineer the BitBrain algorithm from its C source; assess as a gun SirStone 2026-09-24 20:17:08 +02:00
  • 19bf4610e4 TFIL: env-gated commitment arms (tile-replan cancel is a bug) SirStone 2026-09-23 23:28:37 +02:00
  • f842ac0f76 Env report Part 2: warn on misnamed env vars (unknown + compile-time defines) SirStone 2026-09-23 08:38:37 +02:00
  • 036979e78e env_reference: verify every knob against the code, fix the trap that cost real time SirStone 2026-09-23 08:29:10 +02:00
  • 9bf3005850 Boot-time env report + fix the env reference SirStone 2026-09-23 08:21:25 +02:00
  • 7f6ccfb015 Ring mover heat retune (author: the user) - dodge bullets, ignore walls/centre, re-plan 3x more often SirStone 2026-09-23 00:30:52 +02:00
  • 69debbe347 Re-adaptation: the sliding WINDOW is a control-validated +9.3pp on late accuracy SirStone 2026-09-23 00:29:37 +02:00
  • b68707c867 Energy economy: the cliff becomes a SLOPE, plus a finishing cap. 11% less energy. SirStone 2026-09-23 00:12:04 +02:00
  • 81af5854df FIX an incomplete commit: TMHORIZON was admitted unconditionally at HEAD SirStone 2026-09-22 23:46:56 +02:00
  • 9ba932d1b1 docs: complete environment variable reference (every knob, its default, and why) SirStone 2026-09-22 23:45:53 +02:00
  • aed579b3af TM horizon: retain learning ACROSS ROUNDS, reset only when the ENEMY changes SirStone 2026-09-22 23:24:11 +02:00
  • be74e369eb Gate 2b: the shift form WORKS - but break-even needs ~80% side accuracy and the problem gives 60%. Thread closed with a number, not a shrug. SirStone 2026-09-22 22:21:32 +02:00
  • fd2f7f608c GATE 2: STOP - the TM learns the side but the correction cannot buy hits SirStone 2026-09-22 22:06:01 +02:00
  • 9064377740 s settled from OUR code (higher = LONGER clauses), and it cannot rescue the gun SirStone 2026-09-22 21:56:06 +02:00
  • e180626b50 Horizon headroom: QUALIFIED PASS at h=10..50, but the signal is one feature SirStone 2026-09-22 21:46:05 +02:00
  • ab8d383121 Automata metrics: settledness alone does NOT separate learning from fidgeting SirStone 2026-09-22 21:40:27 +02:00
  • f9f8d84671 TM diagnostics kit: VALIDATED (finds a known dead input), and it found a real bug SirStone 2026-09-22 21:24:27 +02:00
  • b0654d18eb TM verdict, settled: it loses LIVE and sits at/below its majority class - (c) SirStone 2026-09-22 08:21:49 +02:00
  • eb74f9b2e3 Ram: finisher-only by default, and the bullet-rain abort now measures real energy SirStone 2026-09-22 08:14:03 +02:00
  • 99f9532f55 Melee vs 1v1 racks: mechanism built, but NO gun-level payoff - do not split SirStone 2026-09-22 08:08:43 +02:00
  • bfdcdf8919 The three unmeasured features, A/B'd - and a methodology correction I had wrong SirStone 2026-09-22 02:34:05 +02:00
  • 3142b70aa5 Gate virtual-bullet spawn on rack admission: +68% tick rate, selected gun unchanged SirStone 2026-09-22 02:24:18 +02:00
  • 185a32e9eb Radial offset: STRUCTURALLY incapable of helping, and Pattern does not overshoot SirStone 2026-09-22 02:23:41 +02:00
  • 31c7c01d28 SHIPPED: the default rack is now Pattern-only (+49% hit rate, +66% damage on the boss) SirStone 2026-09-22 02:10:07 +02:00
  • 9cd6e9b8ce Ablation: the radial TM is replaceable by a CONSTANT, and its avenue is dead on the shipped metric SirStone 2026-09-22 02:08:38 +02:00
  • 589a230106 TM radial gun: registered (default OFF) + label-bias fix that removes the bias but retracts its own earlier learning claim SirStone 2026-09-22 01:58:33 +02:00
  • 394b3deeed SETTLED: Pattern alone is the best single gun IN GENERAL, not just vs DrussGT SirStone 2026-09-22 01:49:57 +02:00
  • a54ae6a162 SETTLED: no small rack beats Pattern alone; the selector is negative value on a GOOD rack SirStone 2026-09-22 01:35:51 +02:00
  • 4657fe715e wave pairing: 36-58% of GF/DecayGF/KNN learning samples were MISLABELLED SirStone 2026-09-22 01:33:31 +02:00
  • 1ea72c7f14 TM gun round 2: base was never behind; RADIAL target beats Linear on bmPoint SirStone 2026-09-22 01:27:59 +02:00
  • 78975a35c4 cost: parallel per-enemy virtual bullets are NOT affordable as proposed SirStone 2026-09-22 01:26:12 +02:00
  • e0666a562d The gun selector is NEGATIVE value: Pattern alone beats the full rack (p=0.0012) SirStone 2026-09-22 01:21:14 +02:00
  • 0ede6d12ec selector: arrival-accuracy tie-break measured NEGATIVE; randomness is load-bearing SirStone 2026-09-22 01:07:12 +02:00
  • ca82053a11 TM gun: the discrete-target diagnosis was RIGHT - it learns now. Still loses to Linear. SirStone 2026-09-22 00:56:01 +02:00
  • a73de13458 racks: separate melee and 1v1 gun racks, plus per-mode real hit-rate data SirStone 2026-09-22 00:45:10 +02:00
  • ab86c0481f gun audit: the virtual system is sound; the rack is redundant, not broken SirStone 2026-09-22 00:44:19 +02:00
  • 994f88d7a7 ramming: make the decision PROACTIVE, with a bullet-rain abort SirStone 2026-09-22 00:24:45 +02:00
  • c9825dfb0b power policy: cap power by range and energy, gate 3.0 on above-average chances SirStone 2026-09-21 23:59:56 +02:00
  • 9caf1d3728 movement: range-weighted TFIL variant + tamed heat field (opt-in, default unchanged) SirStone 2026-09-21 23:46:47 +02:00
  • 2daa519e15 TFIL heat field: the safety model keeps us ~400px away, which is our worst range SirStone 2026-09-21 23:39:14 +02:00
  • 391318a7bd cornering/ramming premise REFUTED on three independent measurements SirStone 2026-09-21 22:49:56 +02:00
  • 07f6f3af3f Tsetlin gun: NO configuration adapts faster than random feedback SirStone 2026-09-21 22:45:28 +02:00
  • fb36a0a685 tracker: corpses do not exist - revert the fix and retire the workaround SirStone 2026-09-21 22:41:11 +02:00
  • c091bf3c34 harness: upgrade to server 1.3.1, keep 0.35.5 selectable, re-baseline SirStone 2026-09-21 22:41:06 +02:00
  • 18f778056b gun selector: hysteresis measured NEGATIVE, shipped at the lightest setting SirStone 2026-09-21 22:41:01 +02:00
  • 68e0375be2 feat(radar): adaptive melee radar sweeps only the arc containing all enemies SirStone 2026-09-21 21:35:22 +02:00
  • d3b3c28cdf fix(adversaries): migrate to bot-api 1.0.7 - kills an intermittent crash that corrupted measurements SirStone 2026-09-21 09:14:47 +02:00
  • ab35540035 fix(logging): [config] reported a stale gun - it printed before selection ran SirStone 2026-09-21 08:50:44 +02:00
  • 1e8f0d342a fix(adversaries): the launchers ran STALE binaries - this is why the earlier fix never took effect SirStone 2026-09-21 08:22:04 +02:00
  • c214abcfa8 fix(adversaries): repair four of the five sparring bots SirStone 2026-09-21 08:21:47 +02:00
  • e6653199bb fix(shim): make the generated DrussGT bot dir runnable by hand SirStone 2026-09-21 08:16:09 +02:00
  • 013b9fe01e docs: correct the overstated 'inverted metric' claim; record tonight's fixes SirStone 2026-09-21 06:59:00 +02:00
  • 4cd5618435 fix(test): repair the flaky offline==online acceptance; make the tie-break truly random SirStone 2026-09-21 06:56:44 +02:00
  • 19410164f1 docs: definitive gun-rack report on real measured numbers SirStone 2026-09-21 06:37:20 +02:00
  • 2c94dc221a test(selector): 16 ranking rules A/B'd against the boss - none beat the shipped config SirStone 2026-09-21 06:31:00 +02:00
  • 57b2ac3849 feat(guns): scale-aware power selection (+52% damage); TM classifier gun built, measured, DISABLED SirStone 2026-09-21 05:19:07 +02:00
  • dea4dcb574 feat(gun_harness): scale-aware selector thresholds; default = path + relative SirStone 2026-09-21 04:33:52 +02:00
  • 3b5d70b7c3 feat(gun_harness): runtime metric switch + A/B proving the point metric mis-selects SirStone 2026-09-21 03:58:27 +02:00
  • e2ca2fc7d8 fix(guns): recover the DrussGT regression with a radial-fraction range blend SirStone 2026-09-21 02:14:30 +02:00
  • 17c99f542f feat(fixtures): closed-loop DrussGT captures from real Tank Royale battles SirStone 2026-09-21 01:18:13 +02:00
  • 7f706e5b14 fix(guns): GF family aimed at the wrong RADIUS, not the wrong angle SirStone 2026-09-21 00:56:02 +02:00
  • 8e2be6a4c6 feat(tools): the real DrussGT now plays and wins Tank Royale battles SirStone 2026-09-21 00:29:37 +02:00
  • d5061ee215 test(range): restore the 12/12 offline==online proof; measure TM clause readability SirStone 2026-09-21 00:14:46 +02:00
  • 89370008da fix(tsetlin): make the TM actually learn - saturation 714 -> 13.8 literals/clause SirStone 2026-09-20 23:59:32 +02:00
  • e0bfa9b5e1 feat(tools): drive the real DrussGT jar outside the Robocode engine SirStone 2026-09-20 23:52:51 +02:00
  • 76ae6170f8 docs(research): portability audit of DrussGT 3.1.4159 SirStone 2026-09-20 23:44:42 +02:00
  • 4f18c8ce07 feat(tools): capture real DrussGT movement from classic Robocode as fixtures SirStone 2026-09-20 23:44:42 +02:00
  • 974528d5cf feat(gun_harness): offline gun range, proven equivalent to live play SirStone 2026-09-20 23:44:42 +02:00
  • 3c90a5941d feat(selector): range-aware firing gate fitted to 2611 measured shots SirStone 2026-09-20 23:28:31 +02:00
  • 54e9757567 docs(research): TM learning tracks + note that the TM-DEB source was deleted SirStone 2026-09-20 23:28:31 +02:00
  • 669f9acd41 docs(research): assess TM-DEB paper against the Tsetlin gun's real failure SirStone 2026-09-20 22:57:28 +02:00
  • e53690036b fix(guns): speed-sensitive caches, dead stop-shot branch, exact TM trace pairing SirStone 2026-09-20 22:47:26 +02:00