summaryrefslogtreecommitdiff
AgeCommit message (Expand)Author
2026-08-29exp: add physical grid bias crossoverYurenHao0426
2026-08-29results: pass the physical grid clean gateYurenHao0426
2026-08-29exp: add physical grid clean-learning gateYurenHao0426
2026-08-29feat: match physical classification early stoppingYurenHao0426
2026-08-29feat: train the reconstructed physical gridYurenHao0426
2026-08-29data: extract the released Figure 5 protocolYurenHao0426
2026-08-29feat: reconstruct the Dillavou physical gridYurenHao0426
2026-08-29results: validate SDIL on device imperfection modelYurenHao0426
2026-08-29exp: add device-level imperfection crossoverYurenHao0426
2026-08-29results: show SDIL lowers overclamp budgetYurenHao0426
2026-08-29feat: model state-dependent physical learning biasYurenHao0426
2026-08-29exp: add physical clamp budget crossoverYurenHao0426
2026-08-29feat: compose SDIL with physical overclampingYurenHao0426
2026-08-29results: record raw KP traffic scaling collapseYurenHao0426
2026-08-27exp: freeze raw KP traffic scaling controlYurenHao0426
2026-08-10docs: simplify SDIL deck around real figuresYurenHao0426
2026-08-10docs: add context-free SDIL project deckYurenHao0426
2026-08-10results: close BabyAI population predictor rescueYurenHao0426
2026-08-10exp: freeze BabyAI population predictor screenYurenHao0426
2026-08-10results: reject diagonal SDIL on PickupLocYurenHao0426
2026-08-10results: select four-step PickupLoc historyYurenHao0426
2026-08-10feat: add basal history to BabyAI policyYurenHao0426
2026-08-10results: close feedforward PickupLoc gateYurenHao0426
2026-08-10exp: freeze PickupLoc clean feasibility gateYurenHao0426
2026-08-10results: retain GoToObj as diagnostic onlyYurenHao0426
2026-08-10results: select BabyAI clean configurationYurenHao0426
2026-08-10fix: resolve visible CUDA device for BabyAIYurenHao0426
2026-08-10exp: freeze BabyAI shared-feedback protocolYurenHao0426
2026-08-10results: reject Larkum innovation pilotYurenHao0426
2026-08-10fix: separate template train and test instructionsYurenHao0426
2026-08-10exp: freeze Larkum public innovation pilotYurenHao0426
2026-08-10results: reject four-probe OLS SDIL across seedsYurenHao0426
2026-08-10exp: confirm four-probe SDIL across seedsYurenHao0426
2026-08-10results: select four-probe least-squares SDILYurenHao0426
2026-08-10exp: parameterize upfront predictor screenYurenHao0426
2026-08-10feat: add local online least-squares residual predictorYurenHao0426
2026-08-10results: close LMS upfront-calibration screenYurenHao0426
2026-08-10exp: compare equal upfront calibration budgetsYurenHao0426
2026-08-10fix: honor Dillavou upfront calibration budgetYurenHao0426
2026-08-10results: aggregate four full state-bias EP seedsYurenHao0426
2026-08-08exp: parameterize full EP confirmation seedsYurenHao0426
2026-08-08results: compare full state-bias EP trajectoriesYurenHao0426
2026-08-07docs: treat exact constant calibration as oracleYurenHao0426
2026-08-07fix: summarize Rain metric trajectoriesYurenHao0426
2026-08-07exp: test state-aware correction on full EP trajectoriesYurenHao0426
2026-08-07results: reject one-epoch affine advantage across seedsYurenHao0426
2026-08-07exp: pair centered EP corrections across seedsYurenHao0426
2026-08-07results: evaluate centered EP under hardware biasYurenHao0426
2026-08-07exp: move released-profile screen to centered EPYurenHao0426
2026-08-07results: close Rain C0 and released-profile R0YurenHao0426
generated by cgit v1.2.3 (git 2.25.1) at 2026-09-15 13:58:13 +0000