Training Coach

Doctrine

A report may cite any of these as doctrine-N.

1. doctrine-1Contract freeze and fingerprint discipline

   The policy I/O contract (observation layout, scales, history semantics, action pipeline) is frozen and fingerprinted; every exported policy is stamped and verified; contract changes ship as new versioned profiles that leave old artifacts bit-identical, and old policies run forever under their era's pinned profile.

   Case. The 215-dim omni contract was frozen with a three-machine digest; the one contract-level extension (lateral feed-forward) went in as a new `omni_ff` profile with the old profile provably untouched, and the contract checker caught two real wiring bugs before any training (`contract-freeze-and-checker`). A silently changed gait-clock default would have fed old policies a 25% slower clock - closed by pinned legacy profiles (`legacy-profile-pinning`). A stale derived USD forked plant mass 2.2% until an automated source-vs-derived instrument gated it (`derived-asset-staleness-check`). A gain profile is part of the closed loop a policy was trained in and belongs in its stamp; the recovery line's anchored authority was left out of its manifest and recorded as the gap not to repeat (`gain-profile-belongs-in-the-stamp`), and a second policy behind a deploy-side switch made the handoff state itself a contract (`recovery-two-policies-and-a-state-machine`, `walk-recovery-fsm-handoff`).

   Coach application. On any proposal touching obs/action semantics, defaults, or derived assets: demand the version/profile plan, the fingerprint update, and the checker extension in the same change; flag any old artifact that would run under new defaults.

   contract-freeze-and-checkerlegacy-profile-pinningderived-asset-staleness-checkgain-profile-belongs-in-the-stamprecovery-two-policies-and-a-state-machinewalk-recovery-fsm-handoff

2. doctrine-2Attribution by resolved training params - never eval-override knobs

   Capability differences between lineages are explained only by digging each lineage's *resolved* training configuration and eliminating columns; evaluation-side override knobs (kd-scale, power-scale, cycle-time) act on the plant for *every* policy and may serve as deployment mitigations but never as explanations.

   Case. Low-friction robustness across 8 lineages x 3840 cells was traced to kd DR *bandwidth* - every lineage had ground friction pinned to (1.0,1.0), so "trained friction" could not be the axis; the parameter axis and the plant axis were explicitly separated after the first attribution conflated them (`kd-bandwidth-mu-law-attribution`). "Weak turning" on hardware was a power-scale plant effect, not a training gap (`deploy-knob-attribution-before-retraining`); slowing the deploy clock was out-of-distribution, not a feature (`cycle-time-override-is-ood`). The ground truth for what a run trained under is the logged per-run config, not the source tree (`resolved-config-is-source-of-truth`).

   Coach application. Whenever asked "why is lineage A better", require the resolved-param table first; kill zero-variance columns; refuse explanations phrased in eval-knob terms; when a knob helps, label it deployment mitigation.

   kd-bandwidth-mu-law-attributiondeploy-knob-attribution-before-retrainingcycle-time-override-is-oodresolved-config-is-source-of-truth

3. doctrine-3PASS gates become constraints; FAIL gates become objectives

   Once a skill passes its gate, that gate converts into a standing regression constraint (budget <= 2/20 against the parent baseline) for all later training; gates currently failing are the only legitimate objectives of the next rung.

   Case. The C ladder ran one frozen 13-cell x 20-seed matrix at every rung with promotion = "new skill PASS and old skills within regression budget"; C1 was stopped and re-rooted precisely because it trained away the root's backward PASS (`fixed-acceptance-matrix-per-rung`, `preregistered-stop-criteria-per-rung`). The C4 product shipped only at 260/260 cells with zero regression.

   Coach application. Keep the ledger: every PASS adds a constraint row; propose rungs only against FAIL rows; treat any constraint violation as stop-and-attribute, never "the next rung might win it back".

   fixed-acceptance-matrix-per-rungpreregistered-stop-criteria-per-rung

4. doctrine-4One variable per ladder rung - counted against what the checkpoint saw

   A rung changes one variable, where "one" is counted against the checkpoint's actual training state, not against the current config's diff; batching is allowed only when each change owns a disjoint symptom space with a pre-registered ablation order.

   Case. Two rungs failed identically because resuming s1e-500 under the evolved config silently added four plant variables the checkpoint had never seen ("单变量纪律不只看「我改了什么」,还要看「checkpoint 见过什么」" - `resume-state-dr-audit`). v8 legally batched four orthogonal fixes with a written ablation order (`orthogonal-batch-with-ablation-order`); v9 spent one run completing a 2x2 factorial so either outcome convicted a factor (`fill-the-missing-factorial-cell`); v10b's three-way ablation wrongfully convicted the clock and had to be retried fairly.

   Coach application. Before any resume: diff cfg against the checkpoint's logged training state. Before any batch: require the symptom-ownership map and ablation order in writing.

   resume-state-dr-auditorthogonal-batch-with-ablation-orderfill-the-missing-factorial-cell

5. doctrine-5Pre-register risks, readings, and stop criteria before the ladder

   Before a ladder or risky rung, write down the known risks, the interpretation of every plausible outcome, and hit-any-one stop criteria - frozen before training, tightened when priors say results should come fast.

   Case. The C ladder opened with three numbered risks including the exact falsification condition for its own root choice; A/B arms carried "预注册读法(事后不改)" tables; a level expected to fail was run anyway for its pre-registered diagnostic value (`preregister-risks-and-fork-readings`). Stop criteria caught C4-redo rungs at +200 instead of full caps (`preregistered-stop-criteria-per-rung`); hardware sessions pre-registered per-config expected signatures and the disagreement rule "不改结论改账" (`preregistered-real-expectations`, `feasibility-accounts-lock-design-point`).

   Coach application. Refuse to open a rung without the written risk/reading/ stop block; after results, read conclusions off the pre-registered table and flag any post-hoc reinterpretation.

   preregister-risks-and-fork-readingspreregistered-stop-criteria-per-rungpreregistered-real-expectationsfeasibility-accounts-lock-design-point

6. doctrine-6Plant parameters are measured, never invented

   Every plant number carries measurement provenance: armature = N^2 x rotor inertia from no-load tests, friction split by rig and by API column, torque limits shaped by per-joint gait peaks, latency traced through the real pipeline, masses weighed - and DR bands are additive around the measured nominal, sized to the measured dispersion.

   Case. Guessed friction was 2.5x low and guessed damping 5x high (`friction-measured-not-guessed`); armature had been 0 with a 9:1 gearbox (81x reflected inertia, `armature-n2-rotor-inertia`); a uniform torque derating was "the wrong shape" vs measured peaks (`torque-limit-shape-by-measured-peaks`); the delay implementation itself was a wrong plant for a whole lineage (`latency-lerp-reverse-extrapolation`); the run design point was locked by three accounts including the tau_limit/kd speed ceiling (`feasibility-accounts-lock-design-point`); identified friction had to land in the right simulator API columns to act at all (`sim-api-friction-columns`). The recovery and one-leg lines opened with the same kind of accounts before any reward existed - a connected static path and the torque along it for an armless get-up, and the gains single support needs to be holdable at all (`get-up-feasibility-accounts-before-training`, `single-support-gain-authority-probe`).

   Coach application. For any plant value in a config review, ask "measured how?"; reject absolute ranges with no nominal; check API column mapping and derived-asset regeneration whenever measured values land.

   friction-measured-not-guessedarmature-n2-rotor-inertiatorque-limit-shape-by-measured-peakslatency-lerp-reverse-extrapolationfeasibility-accounts-lock-design-pointsim-api-friction-columnsget-up-feasibility-accounts-before-trainingsingle-support-gain-authority-probe

7. doctrine-7Sim2sim gate before sim2real - under deployment conditions

   Every checkpoint passes a second, independently built simulator before hardware, and both the gate and the smoke loop run under the measured deployment conditions (real pipeline delay, honest contact parameters, the deployment gain/power profile).

   Case. The standing order "先sim2sim 再sim2real" (`sim2sim-gate-before-sim2real`); acceptance flipped to match hardware only under measured condim/torsional friction (`eval-plant-honesty-contact-params`); gates moved permanently to `--delay 2` after the kicking incident (`pipeline-latency-is-plant-not-dr`); and the harness itself must be audited - a frame-convention bug in the cross-sim evaluator invalidated a whole line of verdicts (`body-frame-velocity-api-audit`). The recovery line's second simulator caught a torque penalty paid for by bracing the legs together (`torque-penalty-bought-by-leg-bracing`), and a 1.8x torque disagreement between the two plants stayed binding because its one surviving explanation was never tested (`torque-disagreement-between-simulators-unresolved`).

   Coach application. Block any hardware request lacking a second-sim PASS at deployment conditions; when sim2sim and training-side metrics disagree, treat the evaluator as a suspect too.

   sim2sim-gate-before-sim2realeval-plant-honesty-contact-paramspipeline-latency-is-plant-not-drbody-frame-velocity-api-audittorque-penalty-bought-by-leg-bracingtorque-disagreement-between-simulators-unresolved

8. doctrine-8Observation honesty - the actor's inputs are a hardware contract

   The actor observes only signals the real robot produces with realistic noise; privileged truths go to the critic; history windows are estimators and must train under plant variation; rewards on quantities the actor cannot observe buy only average suppression, never closed-loop correction.

   Case. Ground-truth velocity/forces went critic-only (`observation-honesty-critic-only`); frame_hist under zero DR memorized the trainer's plant fingerprint - 0/20 transfer (`history-obs-needs-plant-variation`); world-frame yaw rewards could not teach pull-back because heading is unobservable to the actor - correction was routed to the deploy outer loop instead of breaking the contract (`reward-observability-limit`, `deploy-heading-loop-and-align-training`).

   Coach application. Audit every actor-obs element for hardware existence; require minimal plant jitter whenever history/recurrence exists; for each reward, ask "can the actor see this error?" and route correction tasks to outer loops.

   observation-honesty-critic-onlyhistory-obs-needs-plant-variationreward-observability-limitdeploy-heading-loop-and-align-training

9. doctrine-9Reward economics are audited in realized currency

   Reward design decisions are made on realized per-step magnitudes under the actual policy and command distribution: price the do-nothing optimum before adding a mode, compare achieved values to the computed ignore-floor, calibrate thresholds between measured healthy and sick distributions, and ship every new penalty with a withdrawal clause.

   Case. feet_air_time at weight 2.0 realized 0.038 vs tracking 1.2 - drag was rational (`realized-contribution-audit`); ignoring a vy command cost 28-180x less than ignoring vx until a gated tracking term was added (`reward-cost-of-ignoring-audit`, `gate-new-reward-terms-by-command`); achieved-vs-floor separated "never learned" from "priced out" (`ignore-floor-diagnosis`); the foot-distance wall was placed between measured healthy (0.6% tax) and sick (55%) policies (`calibrate-threshold-between-healthy-and-sick`); the landing penalty carried a pre-registered stand-down condition and actually stood down (`calibration-threshold-with-withdrawal-clause`); two clearance terms were inert until zero-points and gate occupancy were checked (`inert-reward-term-audit`). A get-up policy sat because three gated terms paid the seated pose 84% of the return and the one term that could tell sitting from standing was an exp kernel reading 4.6e-5 at the real error (`seated-basin-dead-exp-kernel`); a torque-tail term was weighted by its measured steady value beside a peer term after the estimate proved 12x off (`tail-torque-needs-hinge-on-computed-demand`).

   Coach application. Never discuss weights in the abstract: demand the realized-contribution table, the ignore-floor number, and the healthy-pay calibration before any reward edit is approved.

   realized-contribution-auditreward-cost-of-ignoring-auditgate-new-reward-terms-by-commandignore-floor-diagnosiscalibrate-threshold-between-healthy-and-sickcalibration-threshold-with-withdrawal-clauseinert-reward-term-auditseated-basin-dead-exp-kerneltail-torque-needs-hinge-on-computed-demand

10. doctrine-10The zero-cost option must be the desired behavior

    For every penalty, name what the zero-cost option is; penalize failure events (slip, saturation excess, contact in flight windows), never the motion or joints that healthy behavior uses; make degenerate strategies fatal via termination where penalties cannot price them out.

    Case. Joint-usage penalties for drift taxed a 1.4%-of-momentum channel 2.7/step and collapsed training; the slip penalty costs a non-slipping gait exactly zero (`penalize-the-slip-not-the-joint`). A frozen-at-clamp joint pays zero action-rate forever - only a pre-clip saturation penalty flips the cheat economics (`saturation-cheating-zero-rate-cost`). Ungated phase shaping made standing 42x more expensive than stepping and cooked the hip motors (`moving-gate-42x-stand-tax`); crouch-shuffling lived until a height termination deleted it (`termination-closes-degenerate-basin`). A gated penalty is an exit: the policy parked just outside an uprightness gate, then just under a height gate, to stop paying a stance tax, and only a positive band plus an always-on guard closed both (`penalty-gate-is-an-escape-hatch`); a soft-limit penalty that charged the standing pose itself bought a 4.1 deg lean (`soft-limit-penalty-charges-nominal-pose`); an unpriced foot attitude was spent on edge-standing (`unpriced-foot-attitude-is-a-free-variable`); and the one-leg line listed its cheapest cheats before training and still met one through a zero-gradient band (`enumerate-cheapest-cheats-before-training`, `binary-band-reward-fake-touchdown`).

    Coach application. Run the "零代价的选项是什么" audit on every proposed term; convert motion taxes into event-conditional penalties; check the termination set against each known degenerate strategy.

    penalize-the-slip-not-the-jointsaturation-cheating-zero-rate-costmoving-gate-42x-stand-taxtermination-closes-degenerate-basinpenalty-gate-is-an-escape-hatchsoft-limit-penalty-charges-nominal-poseunpriced-foot-attitude-is-a-free-variableenumerate-cheapest-cheats-before-trainingbinary-band-reward-fake-touchdown

11. doctrine-11Measurement discipline: independent referees, signs, distributions

    A disputed measurement is adjudicated only by an independent algorithm from raw state; directional ability requires sign-antisymmetry under command reversal; bimodal metrics are reported as mode shares (never medians, never 3 seeds); ratios are not comparable when totals change; reward values compare only within one command distribution; single chaotic events never cross machines.

    Case. The triple reversal - a good metric was "refuted" by a sibling metric that shared the disease (`independent-referee-for-metric-disputes`, `body-frame-velocity-api-audit`); same-signed +/- responses were bias, not turning (`same-sign-response-is-yaw-bias`); the swing median sat in a bimodal gap (`median-hides-bimodal-distribution`); "v6 is jitterier" died on absolute energies (`ratio-metrics-need-absolute-check`); yaw gain measured 15x wrong in an oscillating frame (`heading-integral-not-body-rate`); a 44% improvement evaporated under same-distribution comparison (`same-distribution-reward-comparison`); drift direction was a limit cycle (`multiseed-sign-test-for-drift`); a cross-machine push cliff was chaos (`single-impulse-recovery-is-chaotic`).

    Coach application. Before accepting any surprising number: ask for the independent recomputation, the sign pair, the distribution shape, and the comparison conditions. Retract in writing when a metric falls.

    independent-referee-for-metric-disputesbody-frame-velocity-api-auditsame-sign-response-is-yaw-biasmedian-hides-bimodal-distributionratio-metrics-need-absolute-checkheading-integral-not-body-ratesame-distribution-reward-comparisonmultiseed-sign-test-for-driftsingle-impulse-recovery-is-chaotic

12. doctrine-12The deployment pipeline is plant

    Irreducible pipeline properties - action latency, rate limits, power/torque scaling, teleop command mappings - are part of the nominal plant, modeled from day one and reproduced in every gate; deploy-side scalings are crutches that flag unmodeled plant, and they cannot be algebraically folded into training constants.

    Case. Right-leg kicking was over-trained-delay x loop gain; power 0.8 was a gain-reduction crutch that retired when the delay was modeled (`pipeline-latency-is-plant-not-dr`); power derating damages non-forward axes first (`power-scale-hurts-nonforward-axes`); training at 0.4 scale as the "twin" of deploying 0.5 x 0.8 collapsed 0/20 (`deploy-scaling-not-training-equivalent`); one shared teleop speed sent an out-of-band lateral command and the robot clipped its own foot (`teleop-command-band-per-axis`); the latency DR range had not even covered the measured pipeline (`latency-dr-covers-measured-pipeline`). A rate limiter added at deployment only clipped a policy that kept commanding (`deploy-rate-limiter-windup`); moved into training and anchored on the last command it became an integrator in the balance loop (`slew-anchor-is-an-integrator`); anchored on the measured angle it bounded torque and kept the bandwidth (`beta-anchored-action-target`). The walking lines' safe setting, power-scale 0.8, cut the ends of the recovery policy's full-range travel and left its spikes alone; a gain inside the trained band did the job (`power-derating-cuts-full-range-contract`).

    Coach application. Demand the measured pipeline latency/limits in the plant model and in gate conditions; treat every deploy-side derating as a question ("what is this compensating?"); block per-axis command sources that exceed training bands.

    pipeline-latency-is-plant-not-drpower-scale-hurts-nonforward-axesdeploy-scaling-not-training-equivalentteleop-command-band-per-axislatency-dr-covers-measured-pipelinedeploy-rate-limiter-windupslew-anchor-is-an-integratorbeta-anchored-action-targetpower-derating-cuts-full-range-contract

13. doctrine-13DR budget is finite; its distribution is the measured support

    Robustness is a conserved budget: disturbance training on an already-hardened lineage borrows from existing margins; DR ranges span the measured deployment support - no fictitious tails (they buy degenerate gaits), no single constants (they allow thin-margin specialization); harden the plant only after the task distribution is final.

    Case. The same push dose helped a narrow lineage and damaged a balanced one - budget conservation (`push-dr-conditional-budget-conservation`); wide latency tails bought drag-glide, constant values shipped 60% thinner tilt margins - the answer is a narrow band on the measured support (`dr-tail-plant-continuation`, `constant-value-dr-overfits-margin`); task-first ordering because hardening a soon-to-change task wastes budget (`task-shaping-before-plant-hardening`); COM randomization used deliberately as a behavior-shaping tool, and rolled back on symptom per its own contract (`com-randomization-forces-leg-spread`, `com-dr-rollback-on-symptom`). DR that is switched on can still be thin: the run policy fell in the frontal plane its gain-and-latency randomization never touched (`thin-dr-judged-by-channel-coverage`), and a friction priority settled under one action contract had to be re-measured under the next (`friction-priority-re-measured-after-plant-change`).

    Coach application. Before any DR rung: check the untrained policy against the spec, the lineage's current DR load, and the measured real-world range; after it: audit retained margins, not just the new tolerance.

    push-dr-conditional-budget-conservationdr-tail-plant-continuationconstant-value-dr-overfits-margintask-shaping-before-plant-hardeningcom-randomization-forces-leg-spreadcom-dr-rollback-on-symptomthin-dr-judged-by-channel-coveragefriction-priority-re-measured-after-plant-change

14. doctrine-14Gates measure what hardware feels: posture, margins, stripped assists

    Acceptance batteries carry posture-class rows (tilt max median, per-joint L/R asymmetry, temperature) beside task rows, graded margin columns beside binary gates, chirality scored per side, at least one condition that removes the environment's free stabilization, and validated predictive scalars promoted into the gate.

    Case. Three same-shaped judging errors - survival, displacement, wz-difference - all missed what the operator felt; posture metrics had the predictive power (`task-metrics-vs-posture-metrics`, `stand-gate-posture-not-survival`); binary survival saturated and hid a 60% margin gap (`constant-value-dr-overfits-margin`); v5 passed everything on the ground and failed suspended (`suspension-probe-removes-free-stabilizer`); the hip_roll (l+r) scalar predicted real drift direction and ordering and entered the battery (`hip-roll-sum-predicts-lateral-drift`); averages hide chirality (`chirality-scored-separately`); gait-quality gates are judged at speeds that demand a gait (`low-speed-commands-reward-dragging`). The recovery line added the rest of the kit: where failed episodes end, not only where they started (`end-state-confusion-matrix`); a frozen acceptance distribution with a pinned seed (`frozen-acceptance-distribution-and-pinned-seed`); video of the metric rollout itself (`video-as-acceptance-record`); and the admission that a 10 s episode cannot see a stance that fails after a minute (`episode-length-bounds-what-a-gate-sees`). The one-leg line removed a foot-spacing wall that no gate measured, and the feet met on hardware (`removed-wall-returns-on-hardware`).

    Coach application. Review every battery for posture rows, margin columns, per-side scoring, and an assist-stripped condition; when operator feel and gates disagree, suspect the metric class first.

    task-metrics-vs-posture-metricsstand-gate-posture-not-survivalconstant-value-dr-overfits-marginsuspension-probe-removes-free-stabilizerhip-roll-sum-predicts-lateral-driftchirality-scored-separatelylow-speed-commands-reward-draggingend-state-confusion-matrixfrozen-acceptance-distribution-and-pinned-seedvideo-as-acceptance-recordepisode-length-bounds-what-a-gate-seesremoved-wall-returns-on-hardware

15. doctrine-15Fork and root selection: recoverability, maturity, frozen rewards

    Choose fork roots by which candidate's deficits the coming training can pay back (precision is recoverable; lost plasticity, symmetry, and margins are not); prefer mature checkpoints as roots even when younger ones score better as products; never fine-tune through a reward change - continuation is legal only with the reward frozen and plant/DR widening one rung at a time.

    Case. s1e-500 beat higher-precision candidates because its exclusive strengths were unrecoverable (`fork-root-recoverable-shortfall`); the b300 arm proved maturity is capital against adaptation shock (`root-maturity-vs-product-quality`); the B-arm scatter/half-recover/collapse signature falsified reward-change fine-tuning and drew the legal boundary for S2 continuation (`fine-tune-reward-change-falsified`).

    Coach application. For root debates, build the exclusive-strengths table and ask "which side can be trained back?"; require dual-arm evidence for maturity claims; classify any proposed continuation as reward-frozen or not before approving.

    fork-root-recoverable-shortfallroot-maturity-vs-product-qualityfine-tune-reward-change-falsified

16. doctrine-16Curricula: verified engagement, lineage counters, disease-phase gating

    Automatic curricula must prove they engage (a saturated ratchet is constant DR wearing a curriculum's name); every ramp counts lineage-cumulative progress, not per-process steps; penalties aimed at late-stage pathologies ramp in after exploration noise decays; difficulty rises on measured per-stratum success, never on schedule.

    Case. The s1f ratchet capped at iter 248 and never engaged (`auto-curriculum-engagement-check`); the saturation ramp re-fired at +600 after every resume and no shipped product ever saw the penalty (`curriculum-counter-lineage-steps`); the same penalty worked once gated to the disease phase and became an untouchable mechanism (`gate-penalties-to-the-disease-phase`); record-high aggregate reward hid a fully-failing delay stratum (`aggregate-metrics-mask-subgroup-failure`); bucket share is not a gradient lever (`bucket-share-is-not-a-gradient-lever`). An assist curriculum keyed to a pooled success share was withdrawn on the strength of the categories that already worked (`curriculum-criterion-conditioned-on-lagging-category`); a pace set by per-step income moved only when that income was time-gated (`per-step-income-drives-speed-time-gate`), and the same gate had to be retired in a lineage without the disease (`time-gate-vs-wide-stance-retire-the-fix`).

    Coach application. Ask every curriculum three questions: does it engage (show the internal state)? what does it count (process or lineage)? when is it present (against the pathology's phase)? Check where shipped checkpoints sit relative to every ramp.

    auto-curriculum-engagement-checkcurriculum-counter-lineage-stepsgate-penalties-to-the-disease-phaseaggregate-metrics-mask-subgroup-failurebucket-share-is-not-a-gradient-levercurriculum-criterion-conditioned-on-lagging-categoryper-step-income-drives-speed-time-gatetime-gate-vs-wide-stance-retire-the-fix

17. doctrine-17Probe before training: feasibility first, hypotheses in tables

    After two failed training attempts at a skill, stop training: demonstrate the behavior open-loop, enumerate hypotheses in a written table audited against actual configs cheapest-first, race one probe per side of the sim2real boundary for hardware-only pathologies, and use suspended tests to acquit or convict actuators before blaming authority.

    Case. "在黑暗里试钥匙" - four sidewalk rungs failed until an open-loop probe separated exploration/waveform/authority in one experiment (`open-loop-probe-before-reward-tuning`); the foot-drag mystery fell to a seven-hypothesis config audit (`hypothesis-table-code-audit`); the period-doubling was resolved by racing a reward-side and a plant-side evidence line - and both paid off, one per sub-case (`period-doubling-evidence-race`); the suspended test acquitted the roll actuator in one measurement (`suspended-test-isolates-actuator-authority`). A read-only configuration probe told a wall from a slope in the recovery line's seated basin (`configuration-probe-wall-not-slope`), and the fix it pointed to - where the feet are - took prone from 0/159 to 158/159 (`prone-dead-end-is-foot-placement`); a knob that did not move its variable was recorded as no test of the idea (`dof-vel-penalty-is-not-a-pacing-knob`).

    Coach application. When a skill resists training, prescribe the probe before any further reward edits; require verified target trajectories before imitation terms; keep a falsified-fixes list so closed roads stay closed (`amplitude-cut-falsified-yaw-fix`).

    open-loop-probe-before-reward-tuninghypothesis-table-code-auditperiod-doubling-evidence-racesuspended-test-isolates-actuator-authorityconfiguration-probe-wall-not-slopeprone-dead-end-is-foot-placementdof-vel-penalty-is-not-a-pacing-knobamplitude-cut-falsified-yaw-fix

18. doctrine-18External advice is recomputed locally; values transfer as ratios

    Every external suggestion is classified adopt / already-have / modify / trap by recomputing its claim on the local reward table and probe data; numeric values transfer only as dimensionless ratios (to tracking weight, leg length, sqrt(gL), control rate); citations are verified to exist.

    Case. "Start vy very small" would have destroyed sidewalk learning on this reward table - the gradient scales quadratically (`external-advice-audit-against-own-arithmetic`); swing-height targets and weights transferred correctly only through leg-length and tracking-ratio scaling (`transfer-ratios-not-absolutes`); the "6-step delay" was refused for lacking a control rate (`latency-dr-covers-measured-pipeline`); a borrowed reference's structure was FK-verified and its amplitude re-derived from the division of labor (`reference-structure-fk-amplitude-division`); retrieval agents fabricated verbatim arXiv quotes - only source-verifiable material was used; and one dismissed suggestion later proved right for a different mechanism, and was credited (`cycle-average-tracking-for-gait-quantities`). An advisor's staged state machine turned out to exist in none of the three papers it cited, and reading them changed the plan (`advisor-paraphrase-vs-paper`).

    Coach application. Intercept every "paper X does Y" with the local recomputation; convert absolutes to ratios before comparison; verify quotes; revisit dismissed advice when new mechanisms appear.

    external-advice-audit-against-own-arithmetictransfer-ratios-not-absoluteslatency-dr-covers-measured-pipelinereference-structure-fk-amplitude-divisioncycle-average-tracking-for-gait-quantitiesadvisor-paraphrase-vs-paper

19. doctrine-19Hardware sessions are scripted experiments, not tuning sessions

    Real-robot time executes a pre-registered matrix: risk-ordered (baseline first, fragile last with a spotter), stage-gated (suspended smoke before ground), A/B sessions bracketed by a repeated reference run, operators briefed on measured zero-command and untrained-axis behavior, chirality-aware disturbance protocols, no field tuning - the only legal field changes are scripted, single-variable, and self-reversing.

    Case. The S2 acceptance sheet (`risk-ordered-real-deployment`, `battery-bracketed-real-ab`, `know-zero-command-behavior`, `push-test-chirality-protocol`, `no-field-tuning-protocol`); the RAM-only torque experiment with automatic power-cycle rollback (`reversible-single-variable-field-experiments`); and the sim-veto rule - even sim's condemnations get one safeguarded hardware check when they judge the purpose-built configuration (`sim-veto-needs-real-confirmation`). The recovery line's first real run went ahead with its preconditions unmet and was stopped as dangerous (`first-real-get-up-violent-stage-one-policy`); after it: a staged hang, mat and floor protocol (`staged-hang-mat-floor-for-get-up`), a fixed power-cycle pre-flight and two-machine discipline (`power-cycle-preflight`, `two-machine-config-discipline`), a fall guard replaced rather than switched off (`fall-guard-becomes-a-state`), and logs that are part of the run (`hardware-log-is-the-attribution-input`).

    Coach application. Turn every hardware request into a runbook with order, gates, brackets, briefing, and anomaly plays; refuse improvised parameter changes on the floor.

    risk-ordered-real-deploymentbattery-bracketed-real-abknow-zero-command-behaviorpush-test-chirality-protocolno-field-tuning-protocolreversible-single-variable-field-experimentssim-veto-needs-real-confirmationfirst-real-get-up-violent-stage-one-policystaged-hang-mat-floor-for-get-uppower-cycle-preflighttwo-machine-config-disciplinefall-guard-becomes-a-statehardware-log-is-the-attribution-input

20. doctrine-20Close questions in writing; restart when the debt is structural

    Audited questions get frozen verdicts with citable wording and an explicit reopening bar; hardware verdicts are dated by deployment-stack and calibration state and expire when those change; and when successive rungs shuffle symptoms without net progress, freeze the lineage as regression baselines, pay the structural debts, and retrain minimal - carrying laws and instruments, not weights.

    Case. The chirality and COM questions were closed with frozen wording and "no reopening without new hard evidence" (`frozen-verdicts-semantic-boundaries`); v5/v6's condemnations expired with the deploy stack (`stale-verdicts-under-old-stack`); a 2-degree calibration fix moved the whole runnable envelope (`zero-offset-calibration-shifts-envelope`); plant upgrades are era boundaries with paired re-baselining (`plant-swap-invariants-vs-shifts`); and the 2026-08-05 reset froze v5-v11, fixed the latency FIFO / manifest / sampling / reward-table debts, and restarted - producing the lineage that reached hardware SOTA (`freeze-lineage-fix-structure-restart`, `minimal-reward-table-with-provenance`). The recovery line's real-robot verdicts ended up in three places that disagree, one of them an undated note in a command file (`write-hardware-verdicts-back`).

    Coach application. Maintain the closed-questions ledger and quote it when symptoms recur; stamp verdicts with stack/calibration versions; when a team is three rungs into symptom-shuffling, raise the restart question explicitly with the freeze-fix-restart pattern.

    frozen-verdicts-semantic-boundariesstale-verdicts-under-old-stackzero-offset-calibration-shifts-envelopeplant-swap-invariants-vs-shiftsfreeze-lineage-fix-structure-restartminimal-reward-table-with-provenancewrite-hardware-verdicts-back

21. doctrine-21Name the quantity in the space it lives in

    A goal, reward term or acceptance criterion about the feet, the base or the contact state is computed from the quantity itself - world poses, forces, per-category outcomes - never through a joint-angle, single-signal or pooled stand-in that assumes everything else sits at nominal; and every detector is validated on a behaviour known not to contain the event before it becomes a gate.

    Case. The recovery line was caught three times: |ankle roll| as "flat feet" sold stance width and the real robot slid into the splits, a hip-roll criterion was confounded by 50 deg of yaw, and the joint table said 0.271 m where the feet were 0.159 m apart; task-space terms produced the first flat, wide stance (`joint-space-proxy-for-task-space-quantity`). Flight detection lied in both directions across two lines - foot height flagged 40% false flight on a walking gait, contact force alone flagged slip chatter as hops (`contact-detector-single-signal-lies`). A pooled height average described a robot that did not exist - six in ten standing, four in ten sitting (`zero-partial-credit-is-not-an-iteration-problem`) - and the walking line had learned the same lesson on yaw rate (`heading-integral-not-body-rate`).

    Coach application. For every reward term and gate row, ask what physical quantity it stands for and whether it is measured directly; flag joint-space or single-signal stand-ins for task-space goals, ask for a detector validated on a negative control, and split pooled metrics by category before reading them.

    joint-space-proxy-for-task-space-quantitycontact-detector-single-signal-lieszero-partial-credit-is-not-an-iteration-problemheading-integral-not-body-rate

22. doctrine-22Continuation needs a live gradient; a release is chosen by a scan

    Continue a converged policy only on a change that creates a live gradient, on a short budget, with every checkpoint scanned on the transfer axis; choose a release by running the full battery over a band of checkpoints and stop on signals, never by taking the last one; and when edits to the terminal phase cannot move a behaviour, roll back and retrain with the constraint present from the start, keeping the order in which the lineage acquired its mechanisms as explicit curriculum phases.

    Case. A continuation with no new gradient drifted MuJoCo transfer from 100/98% to 80/28% while every Isaac gate stayed perfect, and a live-gradient continuation at the same depth kept it (`converged-continuation-is-poison`). One-leg checkpoints 100 iterations apart failed 1 and 38 of 40 cells, and late ones degraded (`checkpoint-choice-is-a-full-gate-scan`). Four in-lineage stance fixes failed because the stance was the end of the get-up path, and from scratch it grew right (`stance-decided-by-get-up-path`); fixes stacked on degraded states were rolled back by the user (`stop-stacking-roll-back-and-audit`); and the lineage's final recipe, trained from scratch in one run, sat at 0% because the order of its curriculum was part of the product (`curriculum-history-is-part-of-the-product`). The omni line's short adaptation budgets and mature roots are the same law seen from the other side (`continuation-budget-not-from-zero`, `root-maturity-vs-product-quality`).

    Coach application. Before approving a continuation, ask for the new gradient, the budget and the transfer axis in the scan; before approving a release, ask for the scan; after three rungs without progress on the target, propose rolling back to the last good checkpoint and a from-scratch phase plan instead of a fourth patch.

    converged-continuation-is-poisoncheckpoint-choice-is-a-full-gate-scanstance-decided-by-get-up-pathstop-stacking-roll-back-and-auditcurriculum-history-is-part-of-the-productcontinuation-budget-not-from-zeroroot-maturity-vs-product-quality

Experience cards

66 cards matching “com-randomization-forces-leg-spread”.

• Multiple changes may share one rung only if their symptom spaces are orthogonal - with the ablation order written in advanceorthogonal-batch-with-ablation-order

  Observed oncewalkprocessprocessattribution

  Batch changes into one rung only when you can name each change's private symptom space in writing; pre-register the ablation order (numeric before structural) and per-change escalation plans, so a mixed outcome decomposes without new decisions.

  Symptom

  v8 needed four repairs at once (saturation cheating, landing impact, leg narrowing, heading alignment) - strict one-variable laddering would have cost four training cycles for wounds that were all already diagnosed.

  Context

  The batch was allowed because each change owns a disjoint symptom space, stated explicitly: "A→形态/饱和, B→落地/步态高度, C→腿距/roll 摇摆, D→偏航/转向" - so a single run can attribute each outcome to its change by which symptom moved. For the failure case, the ablation order was pre-registered (A2 -> A1 -> B -> D -> C, "先撤数 值改动" - retract numeric tweaks before structural ones), and every change carried its own escalation/rollback plan (e.g. A insufficient: joint_pos_ref 1.6->2.0 or widen the exp kernel; B overdone - robot afraid to land: v_ok 0.30->0.45; D unstable: rel 0.5->0.25, not back to 0).

  Change

  Four-change rung executed as one run with per-change symptom ownership, per-change contingency plans, and a pre-registered global ablation order for unattributable regressions.

  Outcome

  The rung retained single-run attributability without paying 4x training cost; the contingency table meant no failure mode would require improvising an ablation under pressure.

  Mechanism

  The one-variable rule exists to keep attribution possible, not as an end in itself; attribution survives batching exactly when the changes' observable effects are separable. Orthogonality is a claim that must be argued per pair in advance - and the pre-registered ablation order is the escape hatch for the case the claim fails.

  Applies when

  • several diagnosed fixes are queued and ladder time is scarce
  • deciding between strict laddering and a combined rung
  • a combined rung shows a regression no single change explains

  “四个改动症状空间基本正交,可单 run 归因:A→形态/饱和,B→落地/步态高度,C→腿距/roll 摇摆,D→偏航/转向。出现无法归因的整体退化时消融顺序 A2→A1→B→D→C(先撤数值改动)。”

• The walk phase machine structurally cannot express flight - rebuild the representation for duty < 0.5, teach flight with a mask tax, never a cliff bountyphase-machine-structural-limits

  Mechanism understoodrunreward-shapingreward-shapingcurriculumgate-battery

  When a new gait changes the contact pattern's structure, audit whether the phase/mask representation can express it and rebuild the representation if not; teach the new contact pattern with graded mask-mismatch pressure and count it in acceptance with artifact-proof definitions (minimum segment length), never with cliff bounties.

  Symptom

  Running requires both feet airborne, but the walk-era phase machine switches legs by the sign of sin(phase) - with duty < 0.5 the two swing windows must OVERLAP during flight, which a sign-switching representation cannot express at all.

  Context

  The run phase machine was re-architected rather than patched: per-leg phases (left = phi, right = phi+0.5 mod 1) with leg_phase < duty defining stance, aligned to the walk sin convention at duty=0.5 so the machines agree where their domains overlap. Flight is taught by the SAME mechanism that once cured foot-dragging, direction reversed: in the two planned flight windows the contact mask is (0,0) and feet_contact_number_duty charges -0.3 per foot still on the ground - a mild ~0.16/step tax, deliberately NOT a cliff: "悬崖式腾空奖励诱发跳跃 hack,v4-clearance 家族老课文". The reference shape (half-sine bump over swing progress) is zero at window boundaries by construction, eliminating the clearing-window that the C4 probe measured to cost 13-19% on non-sinusoidal references. The shape self-check ("零代价选项是什么") was run on three behaviors: standing pays ref everywhere (known cmd=0 stepping risk, booked), walking pays only the flight-window tax, proper running collects full marks.

  Change

  New rewards.py run section (leg_phase_duty / stance_mask_duty / ref_run / clearance_run / contact_duty) with the walk versions untouched byte-for-byte; flight acceptance metric defined with a segment-length floor (>=40 ms to count) so numeric contact flicker cannot fake flight.

  Outcome

  Flight became expressible and taught by a calibrated mild pressure; the walk lineage's phase code stayed frozen as its own contract.

  Mechanism

  A phase representation defines which contact patterns exist in the reward's vocabulary; duty cycling below 0.5 introduces states (double-flight) outside a half-period sign convention's language, so no weight tuning can teach them. And rare desirable events taught by cliff-shaped bounties invite hacks (jumping in place); the graded mask tax prices the planned pattern without creating a jackpot.

  Applies when

  • extending a walking stack to running/jumping (duty < 0.5)
  • a desired contact pattern never appears despite reward increases
  • defining flight/contact acceptance metrics

  “duty<0.5 时摆动窗 (1−duty)T > T/2,两腿摆动窗在腾空段重叠 —— sin 符号切腿的机制结构上表达不了"双脚同时在空中"。… feet_contact_number_duty 对"还踩着地"持续 −0.3/脚 —— 与 walk 治拖地同一机制,方向相反。不腾空的税 ~0.16/步 … 梯度温和不构成悬崖(悬崖式腾空奖励诱发跳跃 hack,v4-clearance 家族老课文)。”

• The 1.5 Hz step-frequency gate was retracted - the author had misread his own actuator data, and the limit fought pendulum dynamicsgate-threshold-retracted-frequency

  Mechanism understoodwalkgate-batterygate-batteryactuator-modelingprocess

  Every gate threshold must cite its measurement and survive a first-principles sanity check; when a gate keeps failing otherwise healthy behavior, re-derive the threshold from the raw data before enforcing it again - and retract wrong gates in writing.

  Symptom

  An acceptance criterion "step frequency <= 1.5 Hz" kept failing healthy policies (walk_v4 at 2.33 Hz), and an earlier attempt to force slower stepping (v3) had killed stepping altogether.

  Context

  The threshold had been derived from the author's own actuator frequency-response measurements - but re-reading the raw table showed the misread: amplitude ratio at 2.0 Hz is 0.88 (knee) / 0.83 (ankle), acceptable; the genuinely bad point was walk_v2's 3.8 Hz at 0.58. Mechanism check agreed: the leg as a compound pendulum (L ~ 0.30 m) has a natural frequency ~1.1 Hz, swing half-period 0.45 s - the observed 2.1-2.4 Hz sits near where the leg wants to swing, and forcing 1.5 Hz "是跟摆动动力学对着干" (fights the swing dynamics). The frequency definition itself was pinned by two independent methods (contact-event counting 2.39 Hz vs FFT 2.33 Hz, agreeing): reported numbers are cycle frequency = steps per leg per second.

  Change

  Gate retracted in writing: "步频 ≤1.5 Hz 应删除或放宽到 ≤2.5 Hz"; frequency definition standardized before entering any config.

  Outcome

  walk_v4/v6's 2.1-2.4 Hz reclassified from disease to normal; the v3 failure got its probable explanation (suppressing stepping to meet a wrong gate).

  Mechanism

  A gate is only as good as the measurement and the reading behind it; thresholds inherited from a misread plot become invisible design constraints that later training obeys at real cost. Cross-checking a threshold against first-principles dynamics (pendulum frequency) is a cheap way to catch such misreads.

  Applies when

  • an acceptance threshold repeatedly fails policies that look healthy
  • thresholds were set from a single person's reading of raw data
  • a forced compliance with a gate degrades the behavior it guards

  “我当初依据自己测的执行器频响定的,但看错了区间。… 2.0~2.4 Hz 的幅值比 0.83~0.88 是可接受的;真正不行的是 walk_v2 的 3.8 Hz。… 腿按复摆算(L≈0.30 m)自然频率约 1.1 Hz … 把它压到 1.5 Hz 是跟摆动动力学对着干(walk_v3 把迈步压没了,可能正是这个原因)。”

• Drop the frozen policy into chosen configurations - a squat 2.7 cm lower than the stuck pose stood 52% of the time, the stuck W-sit 0%, and the interpolation between them showed a wall, not a slopeconfiguration-probe-wall-not-slope

  Mechanism understoodrecoveryattributionattributionmeasurementcurriculum

  When a policy is stuck, probe the frozen policy from a grid of hand-placed start configurations, including interpolations between the stuck state and a nearby state it escapes from; one read-only experiment separates height, torque, sampling and configuration and tells you whether to prevent entry or train the exit.

  Symptom

  After R0.3 the policy stood from 62% of starts and never from the W-sit it fell into; height, torque, missing samples and reward were all plausible suspects.

  Context

  A read-only probe placed the R0.3 policy directly into specified configurations. Squats (hip, knee, ankle) = (-.65,-1.3,-.65) stood 100%, (-1.0,-2.0,-1.0) 89.8%, (-1.2,-2.4,-1.2) at 0.176 m 52.3%; the measured W-sit at 0.203 m 0.0%; the account-(3) hand-over state (146 deg tilt) 34.4%; linear interpolations from the W-sit toward the squat at 25/50/75% stood 0.0/0.0/3.1%. The squat family's quasi-static torque is 16% of the limits, and the W-sit was visited ~9 s per episode in training. FK showed the squat family (-a,-2a,-a) keeps the torso vertical, the feet flat and the COM over the feet all the way from 0.146 m to 0.384 m.

  Change

  Height, torque and sampling were eliminated in one experiment; the next rungs targeted entering the W-sit (foot placement) instead of escaping it, and seeding the dead point itself was ruled out because it was already visited every episode.

  Outcome

  Pure configuration: the W-sit (hips externally rotated +/-47 deg, knees folded 110 deg, shins flat, feet beside the body) is a different place from the sagittal squat (feet flat under the COM). The policy's standing skill was bound to a narrow sagittal family, and the wall was confirmed by the interpolation. The foot-placement rungs that followed took prone from 0/159 to 158/159.

  Mechanism

  A learned skill covers the neighbourhood of the states it succeeded from; a start state outside that neighbourhood fails regardless of height or torque, and an interpolation that stays at zero until close to a working state shows the boundary is sharp.

  Applies when

  • a policy stalls in a specific posture and several causes are plausible
  • deciding between reverse-curriculum seeding and entry-prevention shaping
  • a feasibility account says a path exists but the policy does not take it

  “**决定性对比:比死点矮 2.7 cm 的蹲姿站立 52.3%,死点 0.0%。** 所以不是高度、 不是力矩(蹲姿族准静态力矩膝 1.96/12、踝 1.24/17,只占 16%)、也不是训练采样 (死点每局被访问 ~9 s)。**是纯位形问题** … 插值实验进一步显示这**不是坡是墙** —— 走到 75% 仍只有 3.1%”

• Diagnose a behavior failure by enumerating hypotheses and auditing each against the actual config, cheapest firsthypothesis-table-code-audit

  Replicatedwalkattributionattributionprocessreward-shaping

  Before changing anything, write the full hypothesis list for the symptom and audit each against the resolved config and measured magnitudes, cheapest check first; train only on the survivors.

  Symptom

  Real robot leaned forward "wanting to walk" but dragged its feet instead of lifting them - a symptom with many plausible causes and no obvious single fix.

  Context

  Seven hypotheses were listed and each checked against the actual training config files (velocity_env_cfg.py, isaac_values.py), ordered by check cost: missing foot clearance term (CONFIRMED, primary - feet_air_time existed but no swing-height term at all); energy penalties dominating (REJECTED - energy terms total -0.19 vs tracking +1.2, 16%); command range too narrow (CONFIRMED - (0.15,0.35)); nominal pose too crouched / action scale too small (HALF - knee 0.5 rad = 28.6 deg deep, scale fine); mixed PD across motor types (REJECTED - already grouped); missing base-height reward (REJECTED - present at -5.0); height-drop termination (REJECTED - none exists, which itself became finding #4 of the fix list).

  Change

  The audit produced a ranked fix list (add clearance penalty; widen speed range; reduce nominal crouch) with each rejected hypothesis documented so it would not be re-litigated.

  Outcome

  Three confirmed causes fixed over v5/v6: swing height went 22-23 mm -> 34 mm, tracking 81% -> 87%; the rejected hypotheses stayed rejected (no wasted rungs on energy weights or PD grouping).

  Mechanism

  Multi-cause symptoms invite guess-and-train loops; a written hypothesis table forces each candidate to be confirmed or rejected against actual values (not impressions), and cost-ordering the checks means most hypotheses die for the price of reading a config.

  Applies when

  • a real or sim behavior failure has multiple plausible causes
  • the team is about to "try a fix" without an audit
  • post-mortems keep re-proposing already-rejected causes

  “真机现象:躯干前倾像要走,脚抬不起来(拖着蹭)。按成本从低到高逐条核查 … | 1 | 缺 foot clearance | ✅ 成立,首要 | 有 feet_air_time,无任何摆动足高度项 | | 2 | 能量惩罚压过跟踪 | ❌ 不成立 | 能量类合计 −0.19,跟踪 +1.2,只占 16% |”

• When the training reset distribution changes, freeze the acceptance distribution separately and pin its seed - the same checkpoint measured twice differed by 3.8 and 10.2 pointsfrozen-acceptance-distribution-and-pinned-seed

  Mechanism understoodrecoverysim-evalgate-batterymeasurement

  Acceptance distributions are frozen artifacts, decoupled from whatever the training distribution becomes and evaluated with a pinned seed, and every gate row carries its sample size so that a small-n row is never read as a regression.

  Symptom

  R0.3 added easier roll-arc start states to the training resets, and the acceptance script shared the training category table; separately, the same checkpoint scored side 3.8 points and mid 10.2 points differently on two runs of the same script.

  Context

  Acceptance ran 512 parallel envs drawn from the fall categories. Had it kept following the training table, 15% of acceptance samples would have landed on states easier than prone - inflated scores and generations that could not be compared. The two same-checkpoint runs had identical per-item height medians, so the ruler had not changed; the spread was reset resampling noise (side n ~ 169, sigma 2.3%; mid n ~ 35, sigma 8.3%). The spec's "20 seeds" had always meant controlled seeds.

  Change

  accept_recovery.ACCEPT_CATEGORIES pinned to the four R0-R0.2 categories and decoupled from the training FALL_CATEGORIES; --seed 20260809 pinned, after which two consecutive runs were bit-identical. The mid row (n ~ 35) was labelled the bluntest gate.

  Outcome

  Every later generation (R0.3 through V3.1) was scored on the frozen distribution and seed, which is what let R0.3's intermediate state be read as "no measurable gain" (62.7 -> 62.1%) and R0.2's mid drop be booked as noise rather than a regression.

  Mechanism

  An acceptance set that follows the training distribution measures a moving target, and an unpinned reset draw adds sampling noise that small-n rows cannot absorb.

  Applies when

  • the training reset or command distribution changes between generations
  • repeated evaluations of one checkpoint disagree
  • a small category drives a pass/fail decision

  “**分布冻结**:`accept_recovery.ACCEPT_CATEGORIES` 钉死 §5 四类 … 与训练侧 `FALL_CATEGORIES` **解耦**。 … **side 差 3.8 点、mid 差 10.2 点**(h 中位逐项一致,证明不是尺子变了)—— 纯 reset 重采样噪声 … 钉死后两次连跑逐位相同。”

• Tightening the bridge's rate limiter under an unchanged policy cut torque peaks 30-50% and made other things worse - the policy cannot see the limiter, keeps commanding and winds up; a deploy-side limiter is a safety net, not a curedeploy-rate-limiter-windup

  Mechanism understoodrecoverysim2sim-gateactuator-modelingsim2simreal-acceptance

  A rate or torque limiter added at deployment lowers peaks but the policy still commands as if unconstrained (saturation, windup, new contacts); use it as a safety net mirrored in evaluation, and put the constraint where the policy can learn around it.

  Symptom

  After the violent first real-robot get-up, the cheapest candidate fix was to tighten the bridge's slew (rate) limit for the recovery policy without retraining.

  Context

  Probe on R3.1 in MuJoCo (5 categories x 3 seeds, mu 1.0), monkeypatching the limiter with no repository change: TIGHT = RS06 4.0 / RS02 3.0 / RS00 2.0 rad/s (about 0.08/0.06/0.04 rad per policy step) against the current vel_limit setting.

  Change

  The probe decided the role of the limiter rather than a deployment.

  Outcome

  Success 14/15 -> 12/15; get-up median 2.35 -> 3.53 s (max 9.30); torque demand peak median hip_pitch 164% -> 111%, knee 166% -> 86%; action saturation still 100%; leg-leg contact 558 -> 860 frames. The limiter was kept only as a real-robot safety net (mirrored into sim2sim evaluation); the cure moved into training - where the next lesson was that a limiter anchored on the last command is itself an integrator (slew-anchor-is-an-integrator).

  Mechanism

  A policy that never trained with the limiter keeps issuing the targets it learned; the limiter clips them, the target window runs ahead (windup), and the robot follows a trajectory the policy never evaluated.

  Applies when

  • a trained policy is too violent on hardware and a quick deploy-side fix is tempting
  • adding slew, torque or velocity limits in a bridge or firmware
  • evaluation and deployment use different limiter settings

  “判读:**链路侧收紧立等可取地把 τ 峰值砍 30~50%,但成功率掉、饱和率仍 100%、 腿-腿接触反升** —— 策略感知不到限速器,目标窗口继续狂奔。⇒ 收紧 slew 只配当 **真机侧安全网**(必须同步进 sim2sim 口径,基础设施现成),**不配当治法; 治法必须进训练**。”

• A proposal in the runbook - torque and action limits as versioned safety tiers (classroom / research / expert) written to motor RAM and read back, separate from the reward's effort penalty - recorded as a proposal, its implementation unrecordedsafety-limits-are-a-layer-not-a-reward

  Hypothesisinfrareal-deployhardwareprocessreal-acceptance

  Keep hardware limits as an explicit, versioned safety layer (tiers written and read back at start, the persisted default the safest one) and the effort penalty as a behaviour layer; when a skill needs more torque, change tier deliberately rather than trading one layer against the other.

  Symptom

  Running and jumping need more torque than the deployed limits allow, and the temptation is to trade the training-side effort penalty against the hardware limit, or to hand a new user a robot "tuned however the last person left it".

  Context

  A message pasted into the operator runbook (undated, citing Berkeley's practice of storing the full motor configuration as JSON with write and read-back scripts) proposes: configuration is a versioned artifact, not a verbal agreement; three safety tiers in robot.yaml beside the gain and policy profiles - classroom (RS06 limited to 10 N*m, lateral joints clamped: "however bad the policy, it only moves awkwardly"), research (14 N*m, clamps at twice the measured need, the default) and expert (the 36 N*m rating, joint limits only, requiring an explicit flag); deploy writes the tier to motor RAM at start and reads it back, while the stored copy stays classroom so a power cut returns to the safest state. It frames limits as the safety layer and the effort penalty as the behaviour layer - more torque for running means switching tier, not weakening the penalty.

  Change

  None recorded: the message ends by asking which to do first, a rollback or the tiers.

  Outcome

  The sources do not record the tiers being implemented; the deployed limits stayed at 12/17/11 N*m through the recovery and one-leg lines (the one-leg spec treats raising the RS06 limit as a separate, unapproved hardware decision). Related and recorded elsewhere: torque limits were written to RAM only in a scripted, self-reversing field experiment.

  Mechanism

  Hardware limits bound the damage any policy can do; reward terms shape what a policy prefers. Mixing them either weakens safety to buy behaviour or distorts behaviour to buy safety.

  Applies when

  • a new skill needs more torque than the deployed limits
  • robots are handed to students or new users
  • motor configuration lives in people's heads or in the firmware only

  “配置是版本化的产物,不是口头约定。 … deploy_policy 启动时按档写进电机 RAM 并读回校验(落盘的那份永远保持 classroom,断电自动回到最安全状态)。 … 限幅是安全层,dof_torques_l2 是行为塑造层,它们在不同的层,不冲突。跑步要更大力矩就换档,而不是去动训练里的省力惩罚。”

• Torque caps cannot soften footfalls - impact is falling-mass momentum, only the reward can treat itlanding-impact-not-fixed-by-torque-caps

  Mechanism understoodwalkreward-shapingreward-shapinghardwareactuator-modeling

  Classify each hardware symptom by the physics that sets it: quantities fixed by ballistic momentum at contact must be treated through the policy's trajectory (reward terms on approach velocity/force), never through actuator caps - and size such penalty weights against your own tracking reward, not a lighter robot's.

  Symptom

  Footfalls slammed at 1.78x body weight in sim baseline (human walking: 1.2-1.5x); the tempting hardware-side fix was cutting actuator torque limits.

  Context

  Measured directly: scaling torque limits from x1.0 down to x0.4 left peak landing force essentially unchanged (1.75 -> 1.78x body weight) - the impact force comes from the momentum of the falling mass at touchdown, not from motor effort. The fix has to change the trajectory, i.e. the policy, i.e. the reward: feet_contact_forces penalty above a threshold of 113 N (= 1.2x the 9.58 kg robot's weight), clipped, weight -0.005. The weight was sized locally, not copied: the reference robot's -0.001 would amount to 0.9% of tracking reward on this robot ("策略不会理它" - the policy would ignore it); -0.005 gives 4.4%.

  Change

  Added threshold-type contact-force penalty (-0.005, threshold 1.2x body weight) as one of v6-minimal's three changes; hardware torque cuts explicitly rejected as a footfall treatment.

  Outcome

  Landing force 1.72x -> 1.55x by v6 (target <1.5x, missed by 3% - progress booked honestly); the torque-cap dead end was documented so it would not be retried.

  Mechanism

  At touchdown the ground stops a ballistic mass; the impulse is set by approach velocity and effective inertia, which motors can no longer influence in the final instant. Only earlier trajectory choices (approach velocity, timing) reduce it - and those are selected by the reward, not by actuator limits.

  Applies when

  • footfall impact or landing noise on hardware
  • proposals to derate torque as a softness fix
  • importing contact-force penalty weights from another robot

  “⚠️ 硬件限扭降不了落脚力 —— 砸地力来自下落质量的动量: 实测 tau ×1.0→×0.4, 落脚力 1.75→1.78× 体重纹丝不动。只有这条奖励能治。… ⚠️ 权重不能用 Pi 的 −0.001 —— 实测在我们身上只占跟踪奖励的 0.9%, 策略不会理它 (Pi 6.94 kg 更轻)。−0.005 给到 4.4%。”

• Add a single-point-suspension test to acceptance - the ground is a free stabilizer that hides divergencesuspension-probe-removes-free-stabilizer

  Mechanism understoodwalkgate-batterygate-batteryreal-acceptancesim2sim

  Include at least one acceptance condition that strips the environment's free stabilization (suspension, or equivalent) - the sim-passing policy that fails on hardware is often failing a condition the battery never posed.

  Symptom

  walk_v5 looked healthy in every on-ground sim test yet diverged on the real robot - the acceptance battery had never measured a condition that would have revealed it.

  Context

  The battery gained a single-point-suspension probe (robot hung, feet free): measure torso tilt while the policy runs without ground contact. v5 scored 45.9 deg mean tilt suspended - wildly unstable - which the file calls "最灵敏的失稳探针(拿掉地面这个免费稳定器)": ground reaction forces passively stabilize a marginal policy, so on-ground metrics saturate long before the policy's internal balance is actually sound. v6 halved it (23.0 deg, target <10 deg) - progress visible on a scale where on-ground numbers showed nothing.

  Change

  Suspended-tilt added as a standing acceptance row; run under the honest contact parameters battery (accept_v2 with measured condim 4 / torsional friction 0.035), under which v5 correctly FAILS in agreement with the real robot.

  Outcome

  The sim battery's verdict on v5 flipped from pass to fail, matching hardware; suspended tilt became the discriminating metric between v5 and v6 (45.9 vs 23.0 deg) when ground metrics differed little.

  Mechanism

  Contact with the ground closes a stabilizing feedback loop the policy gets for free; removing it exposes the policy's own attitude control authority. A metric measured only in the assisted condition cannot rank policies by the unassisted quantity that hardware will actually demand during perturbations and flight phases.

  Applies when

  • sim acceptance passes but hardware diverges
  • designing an acceptance battery for a legged robot
  • two candidates tie on ground metrics

  “单点吊那条是最灵敏的失稳探针(拿掉地面这个"免费稳定器"), v5 在地上一切正常却在真机发散, 就是因为验收从没测过这个工况。”

• When hardware underperforms, audit deployment knobs before prescribing retrainingdeploy-knob-attribution-before-retraining

  Mechanism understoodomniattributionattributionreal-acceptanceprocess

  Before any "retrain it" decision, reproduce the symptom in sim under the exact deployment configuration; if the symptom follows the deployment knob rather than the checkpoint, fix the knob or randomize it in training - never top-up-train the skill.

  Symptom

  Real-robot feedback after the C4 deployment - "turning is weak" - with two retraining options on the table: top up turn training, or restart from the s1e root.

  Context

  The sim account showed the policy turned well (75-81% at pw1.0); the robot was deployed at power-scale 0.8. The 3-6 pp difference between C2 and C4 policies at the same power was noise; the 40-50 pp difference between power levels was the entire effect. Both proposed retraining paths would have burned budget on a non-existent training gap, and restarting from s1e would additionally have discarded the sidewalk skill that took four rungs and a coordinate-bug hunt to obtain.

  Change

  Decision: retrain nothing. (1) Try pw1.0 on hardware first - sim says net gain; (2) only if 1.0 is unacceptable (heat/feel), the correct training fix is power/torque randomization in the S2 plant line (one variable, fixes turn and backward together) - not skill top-up; (3) restart-from-root explicitly ranked worst.

  Outcome

  The "weakness" was fully explained by the deployment knob; the sim/real signatures matched the earlier power-derating law verbatim ("与 C2 时代 power 衰减主要伤非前进轴 逐字吻合").

  Mechanism

  The policy's competence is defined under its training plant; deployment knobs (power scale, teleop mapping, command bands) silently define a different plant. Attributing a deploy-plant effect to a training gap produces exactly the wrong fix - more training on the wrong variable.

  Applies when

  • real robot underperforms a skill that sim says is fine
  • proposals on the table include retraining or re-rooting
  • deployment uses any override the trainer never saw (power scale, remapped commands, different control rate)

  “正确的训练修法不是补训转向，而是训练时加 power/力矩随机化让策略在 0.8 下自己补偿 —— 单变量，属 S2 plant 线，一次同时修好转向与后退；从 s1e 重训是最差选项：丢掉四轮 + 一个指标 bug 才换来的侧走，而 C2 的转向本来就没问题。”

• Adapting a lineage to one plant increment needs hundreds of iterations, not thousands - long runs only buy specializationcontinuation-budget-not-from-zero

  Mechanism understoodomnitraining-runcurriculumprocess

  Budget continuation rungs by increment class (hundreds of iterations for plant pins and smooth shifts, ~1000-1500 only for behavior-demanding changes like push), enforce a hard cap with frequent evaluation, and treat remaining budget as a reason to stop, not to continue.

  Symptom

  The default "6000 iterations per rung" (a from-zero-scale budget) was about to be applied to continuation rungs whose only change is one plant/DR increment - overspending compute and, worse, giving each rung thousands of iterations to specialize away retained skills.

  Context

  The 2026-08-07 budget table replaced the default with "最低适应窗口 + 每 100 iter 验收 + hard cap" scaled to the increment's difficulty: fixed-latency levels 300-500 (cap 500-800; the base has already seen in-band values, this only pins the plant); PD full-band 700 (cap 1000; kp+/-20%/kd+/-30% clearly widens the actuator family); COM +/-20 mm 500 (cap 800; a smooth dynamics shift); friction DR 700 (cap 1000; contact and actuator friction change the gait/contact solution together); push 1000 (cap 1500; a non-static plant change requiring recovery behavior - hardest). Rationale: "续训适应一个 plant 增量不需要从零量级的预算,跑长了只是给特化时间". The C ladder reused the scheme (per-rung caps 500-2000 by increment type), and the deep-training hazard got its own name when long runs sold quality ("深适应卖质量" - deep adaptation sells quality).

  Change

  Per-rung iteration budgets set by increment class with hard caps and 100-iter watch loops; checkpoint selection inside the window by the smoke curve, never "run to cap because budget remains".

  Outcome

  S2/C rungs completed in 300-1500 iterations each; the recurring late-run degradations (collapse valleys at 1500+, vx+0.30 decay) fell outside most rungs' caps instead of inside their runs.

  Mechanism

  A continuation rung's learning problem is local robustification around an existing optimum - low sample complexity; iterations past adaptation are spent sharpening onto the current distribution, which is exactly how retained skills and margins erode. Budgets sized to the increment bound both compute and the specialization damage window.

  Applies when

  • planning iteration budgets for a robustification or command ladder
  • a continuation run keeps improving its training metric late
  • retained skills decay in the back half of long continuation runs

  “「最低适应窗口 + 每 100 iter 验收(watch_ckpt --every 100)+ hard cap」—— 续训适应一个 plant 增量不需要从零量级的预算,跑长了只是给特化时间 … ⑥ push | 1000 | 1500 | 非静态 plant 变化,要学 recovery 行为,最难”

• Fine-tuning through a reward-table change was falsified (scatter, half-recover, collapse) - continuation training is legal only with the reward frozenfine-tune-reward-change-falsified

  Replicatedomnicurriculumcurriculumfork-selectionreward-shapingprocess

  Never fine-tune through a reward-table change - retrain from zero; reserve checkpoint continuation for frozen-reward plant/DR widening, reset noise_std when branching, and watch for the scatter/half-recover/collapse signature as the abort trigger.

  Symptom

  The s1c A/B experiment: arm B fine-tuned from an existing checkpoint under the revised reward (same contract, same network, changed reward + small DR) and failed with a characteristic signature - scatter, half-recover, fall back ("打散→半恢复→摔回"); arm A trained from zero under the same config won decisively (full shaping lifted swing to 21.6 mm within 500 iters; shipped at 5500).

  Context

  Verdict recorded: "从零 + 强塑形是本机唯一验证过的发育路径" (from-zero plus strong shaping is this machine's only validated development path). The signature became a standing stop criterion in every later rung that touched a reward ("s1c B 臂签名,出现即停"). Crucially the boundary of the law was drawn explicitly when S2 continuation training was proposed: "当年证伪的是「奖励表中途改版的 fine-tune」… S2 续训奖励表全程冻结、只逐级加宽 plant/DR,属另一类" - continuing a checkpoint with the reward FROZEN while widening plant/DR one rung at a time is a different class and was allowed (and then worked, powering the whole S2/C lineage) - with the honest fallback that if frozen-reward continuation ever collapses, that rung retrains from zero and the doctrine gets re-examined with data. Fine-tune arms also need mechanical care: reset the checkpoint's collapsed noise_std (terminal 0.033 "会杀死探索") and account for iteration counters re-zeroing (curriculum gates fire immediately).

  Change

  Reward changes and lineage continuation permanently separated: reward revisions -> from-zero retrain; plant/DR widening -> frozen-reward continuation with per-rung gates; the B-arm signature promoted to a universal tripwire.

  Outcome

  No later reward revision was attempted by fine-tune; frozen-reward continuation carried S2 (PD/COM/friction rungs) and the C command ladder successfully from the s1e root.

  Mechanism

  A trained policy sits in an optimum of its reward's geometry; changing the reward moves the optimum but leaves the policy's exploration noise near-zero and its value function calibrated to the old returns - it disassembles the old solution faster than it can assemble the new one. Widening DR under a frozen reward instead keeps the optimum's identity and asks only for local robustification.

  Applies when

  • proposing to fine-tune an existing policy under a revised reward
  • planning a robustification ladder from a validated checkpoint
  • a continued run scatters then partially recovers then collapses

  “B 臂 fine-tune 证伪(打散→半恢复→摔回——从零 + 强塑形是本机唯一验证过的发育路径)。… 当年证伪的是「奖励表中途改版的 fine-tune」(B 臂,塑形突变致终盘摔回);S2 续训奖励表全程冻结、只逐级加宽 plant/DR,属另一类;若 s2_lag1 续训本身塌方,回退方案 = 该级从零重训,续训教义再议(拿数据说话)。”

• A quadratic clearance reward stalled for 3500 iters near target - switch to an indicator on accumulated heightindicator-reward-avoids-gradient-decay

  Mechanism understoodwalkreward-shapingreward-shaping

  When a shaped term plateaus near its target, check the gradient profile: replace vanishing-gradient forms with threshold/indicator forms for the final approach, and prefer delta-accumulation over absolute positions to immunize against frame offsets.

  Symptom

  The quadratic-error clearance term froze at -0.002 from iteration 2000 to 5500 - thousands of iterations with no progress on foot lift.

  Context

  Diagnosis: a quadratic penalty's gradient vanishes as the error approaches target, so exactly where the last millimeters must be earned the incentive fades to nothing. The replacement (Humanoid-Gym form): accumulate the swing-phase height climb per foot, reward a BINARY indicator |accumulated - target| < 0.01 masked to the planned swing window, reset on contact, weight +1.6 as a positive reward. Two properties: the indicator's incentive is constant until the threshold is crossed (no decay zone), and accumulating height DELTAS makes any constant sole-frame offset cancel automatically - which structurally sidesteps the earlier 0.0585 m zero-point bug ("顺带绕开我先前那个'忘了减 0.0585 导致惩罚恒为 0'的坑").

  Change

  Clearance reformulated from quadratic penalty on instantaneous height to indicator on per-swing accumulated climb (target 0.03 m by leg-length scaling, weight +1.6).

  Outcome

  Part of the v5 package under which lift finally moved (v5 29 mm, v6 34 mm vs the stalled 18-24 mm era); the offset-cancellation property removed one whole bug class from the term.

  Mechanism

  Policy-gradient learning follows the reward's local slope; quadratic shaping concentrates slope far from target and starves it near target, so convergence stalls precisely at the finish line. An indicator pays a constant bounty until the goal is met; formulating on deltas rather than absolutes removes sensitivity to reference- frame constants.

  Applies when

  • a reward term's value freezes short of target for thousands of iters
  • designing clearance/height/precision terms
  • reward code depends on absolute link positions

  “现行二次型在接近 target 时梯度趋零 —— 这正是 clearance 从 iter 2000 到 5500 卡在 −0.002 不动的原因。… 二值指示在跨过阈值前梯度恒定,没有衰减区 … 累积 delta 让 SOLE_OFFSET 自动抵消”

• Removing a foot-spacing wall passed every simulated gate and made the feet collide on the real robot - nothing priced stance width in the two-foot phase, the policy narrowed to the simulator's self-collision floor, and real calibration offsets closed the last millimetres; the wall came back with a gateremoved-wall-returns-on-hardware

  Observed onceonelegreal-deployreward-shapinggate-batteryreal-acceptance

  When a constraint is removed, name what will govern that quantity instead and add a gate for it; never let a simulator's collision floor be the margin, and when a gate is exceeded by a hair, record the exact numbers and hand the release decision to a person instead of quietly passing it.

  Symptom

  On the first real-robot try of oneleg_v0 (2026-09-16) the two feet collided; the user also judged the folded foot not high enough.

  Context

  The V0 reward table had dropped the feet_lateral_distance wall because it seemed to conflict with the hip adduction single support needs. In the two-foot command bucket no remaining term governed stance width, so the policy drifted narrower until the simulator's self-collision stopped it; the sim acceptance had no foot-spacing gate, so 40/40 said nothing about it. On the robot, calibration offsets consumed the margin.

  Change

  V0.1: the wall restored (-10, minimum 0.16 m), re-checked against measured numbers (a swing-phase lateral spacing of ~148 mm costs 0.12 per step, acceptable); fold weight 0.8 -> 2.0; a ninth gate: minimum foot spacing >= 100 mm and zero leg-contact frames. The removal was kept on record.

  Outcome

  oneleg_v0_1 (V0r2 model_2200) passed 39/40 with the spacing gate 40/40. The single miss (a 15.4 deg tilt transient against a < 15 deg limit during a side switch, steady 6.9 deg, everything else green) was recorded with its numbers and released for the user to overrule.

  Mechanism

  An unpriced degree of freedom drifts to wherever the simulator stops it; if that stop is the simulator's own collision model, the policy's margin on hardware is whatever the calibration error leaves.

  Applies when

  • dropping a reward term that looked redundant or conflicting
  • hardware shows a failure no simulated gate measures
  • a release candidate misses one gate row by a small amount

  “V0 撤墙被真机证伪（2026-09-16）：双脚桶没有任何项管站宽，策略贴 sim 自碰撞底线收窄，真机标定偏差一吃**双脚相碰**。 … min ≥ 100 mm 且腿碰 0 帧（eval_straight 同判据）—— … V0 真机双脚相碰暴露 sim 门未看脚距的缺口 … L s2 标称 tilt 瞬态 15.4°（门限 <15, 超 0.4°, 稳态 6.9°, 该跑其余全绿）——换侧瞬态蹭线, 判定放行留档, 用户可否决。”

• Every power cycle starts with the same read-only pre-flight - read the buses, check the torque limits against 12/17/11, verify the IMU axes, check the ports after any new USB device - and any reassembly re-measures the joint zerospower-cycle-preflight

  Observed onceinfrareal-deployreal-acceptancehardwareprocess

  Start every powered session with a fixed, read-only pre-flight - bus responses, torque limits equal to the simulated ones, IMU axes, device identities - and re-measure joint zeros after any mechanical reassembly before running a policy.

  Symptom

  Hardware state drifts between sessions in ways no policy can see: a motor that stops answering after a power cycle, a torque limit that differs from the one simulated, an IMU axis flipped, two USB devices swapping identities, a joint zero moved by reassembly.

  Context

  The runbook's session order before any policy runs: read every motor on both CAN buses without enabling them (the first command after every power cycle); set_torque --check, all twelve motors must read 12/17/11 N*m, and any difference is written back; imu_reader --verify-axes, where the operator tilts the robot forward and to the right and every check must pass before continuing; check_ports after plugging in any new USB device (the IMU and a CAN adapter once collided on USB identity). After re-mounting motors: read the buses, then re-measure the calibration offsets (three repeats, written back) - "skipping it means running everything on the wrong zero". Hanging checklists repeat the torque-limit check (the deploy script also self-checks at start).

  Change

  A fixed, read-only pre-flight run in the same order every session.

  Outcome

  The runbook records one earlier hardware check in the same spirit: all 12 motors' implied kp fell within 18.4-22.0 for a commanded 20, inside the kp randomization range used in training.

  Mechanism

  A policy transfers only if the plant matches the one it was evaluated on; the pre-flight turns silent hardware drift into a failed check before the robot moves.

  Applies when

  • the first command after powering a robot on
  • after swapping adapters, cables or motors
  • a policy that worked last session suddenly behaves differently

  “python tools/set_torque.py --check # 12 颗应全对 12/17/11, 有 diff 就 --write … 插任何新 USB 设备后都先跑一次 check_ports.py(IMU 和 CANable 的 USB 身份撞过车) … python tools/calib_stance.py --repeat 3 --write # 重标 offset —— 8/9/10 重新装, 机械零位变了”

• Training-log reward values and fixed-command eval values live on different distributions - comparing them once claimed a 44% improvement that was really 6-10%same-distribution-reward-comparison

  Mechanism understoodwalksim-evalmeasurementattributionprocess

  Quote reward-term values only with their distribution attached (command range, DR on/off, environment), and compare across runs only when those match; re-measure in a common environment before claiming any improvement percentage.

  Symptom

  A v6-era analysis concluded slip had dropped 44% by comparing the training log's Episode_Reward against values calibrated in a fixed-command play environment; a same-condition re-measurement showed the true improvement was 6-10%.

  Context

  The training log's reward is an expectation over the training command distribution (vx 0.15-0.5, yaw +/-0.6, with pushes and domain randomization); play-environment calibrations are taken at a single fixed command with DR off. Subtracting one from the other compares apples to oranges - the warning was written into the v7 pre-flight: "奖励数值只能在同一指令分布下比较 … 拿它和固定 cmd 的 play 环境标定值相减会得出错误结论(v6 那轮已经栽过一次)".

  Change

  Rule adopted: any before/after reward-term comparison must hold the command distribution, DR state, and evaluation environment fixed; training-log values compare only against training-log values of runs with identical command/DR configs.

  Outcome

  The phantom 44% improvement was retracted; later term-level accounting (e.g. the C4 ignore-floor work) consistently specified its distribution before quoting numbers.

  Mechanism

  A reward term's expectation depends on the visited-state distribution as much as on the policy; changing the command distribution or DR moves every term's baseline. Cross-distribution differences therefore measure the distributions, not the policy change.

  Applies when

  • comparing reward telemetry across training runs or vs play evals
  • claiming improvement percentages from training logs
  • term-level reward accounting for diagnosis

  “奖励数值只能在同一指令分布下比较。训练日志的 Episode_Reward 是在训练指令分布上算的(vx 0.15~0.5 / 偏航 ±0.6 / 带推力与域随机化), 拿它和固定 cmd 的 play 环境标定值相减会得出错误结论(v6 那轮已经栽过一次: 据此以为滑移降了 44%, 同条件对拍只有 6~10%)。”

• Calibrate a wall penalty by measuring healthy and sick policies - healthy pays ~0, the disease pays a wallcalibrate-threshold-between-healthy-and-sick

  Mechanism understoodwalkreward-shapingreward-shapinggate-battery

  Calibrate every threshold penalty by evaluating its exact formula on replays of at least one healthy and one sick policy: place the threshold between their distributions, size the weight so the sick policy pays a decisive fraction of tracking while the healthy one pays ~0, and pre-compute neighboring thresholds for cheap adjustment.

  Symptom

  Real walk_v7 occasionally clipped its own legs (stance narrowed to 133 mm mean vs nominal 214.5); a foot-distance penalty was needed, but an uncalibrated threshold/weight risked either doing nothing or becoming a reverse barrier.

  Context

  The term (relu(d_min - lateral foot distance), measured in the base yaw frame because world-frame y is meaningless after turning) was calibrated before training by replaying three known policies through the exact reward formula at the acceptance operating point: healthy v5 (183 mm) pays 0.6% of tracking - effectively free; narrowed v7 (133 mm) pays 22% - effective widening pressure; collapsed v8 (93 mm) pays 55% - a wall. d_min 0.16 was placed deliberately between healthy and sick, with alternative thresholds (0.14/0.18) pre-computed in the tool output for later adjustment. The shape self-check was named as a standing question: "先问'零代价的选项是什么'" - the zero-cost region must be exactly the desired behavior.

  Change

  feet_lateral_distance added at d_min 0.16 / weight -10, with telemetry expectation pre-registered (should decay toward 0 as stance learns >160 mm; if bow-legged over-widening >214 appears, only then discuss an upper bound).

  Outcome

  v9_probe on hardware: no leg contact ("没碰腿(N2 兑现)"), stance min 145/126 mm green; the term's zero-cost design left healthy gait untaxed.

  Mechanism

  A relu threshold penalty defines a free region and a priced region; its correctness is entirely in where the boundary sits relative to the healthy and pathological distributions. Replaying known-good and known-bad policies through the exact formula measures both distributions in the term's own currency, making the threshold and weight a placement decision instead of a guess.

  Applies when

  • adding any relu/threshold-style wall penalty
  • a safety margin (foot distance, joint limit, clearance) needs enforcement without taxing normal behavior
  • choosing between candidate thresholds for a new term

  “形状自检 (v6 横向组/v8-B 的教训 —— 先问"零代价的选项是什么"): 标称站距付 0, 健康步态付 ~0, 收窄才付费 … v5(健康) 183 mm … 0.6% ≈ 免费 | v7(收窄) 133 mm … 22% —— 有效推宽 | v8(塌陷) 93 mm … 55% —— 墙 … d_min=0.16 恰在 v5(183)与 v7(133)之间”

• Model CAN polling skew - joint observations are 6-9 ms stale by read ordercan-timing-skew-modeling

  Observed oncewalkactuator-modelingactuator-modelingsim2simhardware

  If joints are read sequentially over a shared bus, reproduce the per-group observation staleness in sim (or randomize it over the measured range) - synchronous observations are a privileged fiction.

  Symptom

  Policies trained with synchronous joint observations degrade on hardware where motors are polled sequentially over CAN - hip data is already 6-9 ms old by the time ankle data arrives.

  Context

  Menlo's core sim2real finding on a leg platform of almost identical mass to Lucen's. CAN is a sequential bus: one poll cycle reads motors in a fixed order, so the observation vector mixes timestamps. Lucen runs 6:6 motors on a dual CAN split, so the problem transfers one-to-one.

  Change

  Explicitly model CAN timing skew in sim: give the joint groups different observation delays matching physical read order. Menlo went further - running real firmware in the loop with a motor simulator between MuJoCo and firmware injecting 0.4-2 ms uniformly distributed delay.

  Outcome

  Reported by the reference team as their core sim2real enabler on a same-scale platform; recorded in Lucen's experience log as directly applicable ("这个问题一模一样").

  Mechanism

  A policy exploits any cross-joint temporal coherence present in training observations; when hardware breaks that coherence per bus position, the learned feedback acts on inconsistent state estimates, injecting phase error exactly at the control bandwidth.

  Conflicts

  Second-hand episode: outcome numbers are the reference team's report, not a Lucen-run experiment; Lucen adopted the requirement but the corpus has no Lucen A/B of skew-on vs skew-off.

  Applies when

  • robot polls actuators sequentially over CAN/RS485 or similar shared bus
  • sim2real degradation appears as jitter or oscillation not seen in sim
  • designing the observation/delay model before a training run

  “电机走 CAN 是顺序轮询的，髋部电机的数据到踝部电机上报时已经陈旧了 6-9 ms，他们直接在仿真里显式建模了 CAN 时序偏斜，按读取顺序给三组关节不同的观测延迟。… 注入 0.4–2 ms 的均匀分布延迟。你们是 6:6 双 CAN 分总线，这个问题一模一样”

• A walking policy's tilt cutoff is a legal state for a recovery policy - the default 45 deg fall guard had to be raised for recovery tests and is disabled once the switch owns falls, so the abort chain becomes the recovery timeout, the operator's cut, and the firmware torque limitsfall-guard-becomes-a-state

  Observed oncerecoveryreal-deployreal-acceptancehardwareprocess

  When a new skill makes a safety cutoff's trigger a legal state, replace the cutoff with a bound of the skill's own (a timeout ending in a safe stop) instead of just switching it off; keep the operator's cut and the firmware limits as independent layers, and write every flag change into the run sheet.

  Symptom

  deploy_policy's default protection stops the robot beyond 45 deg of tilt. A recovery policy starts lying at roughly 90-97 deg, so under the default it is refused on the spot - a flag the first hanging checklist forgot.

  Context

  The layers in the sources: deploy_policy's tilt cutoff (default 45 deg, a line in the safety chain); for standalone recovery tests the cutoff was raised (110 deg in the spec's A/B sheet; 181 deg, effectively off, in some runbook commands); with --recovery-policy the cutoff is disabled because a fall is now a state, not an exception, and RECOVERY lasting over 15 s ends in a safe stop (the runbook calls it the line where the spotter steps in). Independent of the policy: firmware torque limits checked at start (12/17/11 N*m, set_torque --check), the operator cutting enable at any kicking or oscillation, and in the one-leg teleop a space-bar stop that puts the foot down.

  Change

  The flag was added to the run sheets, and the FSM replaced the removed cutoff with its own bound (the timeout).

  Outcome

  The spec records the flag omission and its fix; it does not record the FSM's timeout being exercised on hardware.

  Mechanism

  A safety cutoff encodes one policy's notion of "abnormal"; a new skill whose normal operation lies beyond it either cannot run or runs with the cutoff off, and only a replacement bound keeps the chain closed.

  Applies when

  • deploying recovery, fall-damage or acrobatic skills behind existing safety checks
  • a run sheet disables a protection flag
  • listing the abort chain for a hardware session

  “--max-tilt-deg(默认 45°,安全链第 13 行写的那个)。recovery 的合法状态覆盖整个倾角域,把它抬到 181 = 实效关闭 … RECOVERY 超时 15s 会自动安全停(看护介入线)”

• Three hardware accounts locked the run design point - and the knee's real speed ceiling is tau_limit/kd, not the firmware limitfeasibility-accounts-lock-design-point

  Mechanism understoodrunplant-calibrationplant-calibrationactuator-modelinghardwareprocess

  Before opening a dynamic-gait training line, compute the full account set - tau_limit/kd effective speed ceilings, joint ROM under the intended reference geometry, and thermal RMS at the duty cycle - and let the accounts lock the design point; move only to pre-registered in-table alternates, re-running the accounts first.

  Symptom

  The run line was believed to require a firmware raise of the RS06 speed limit (10 rad/s) as a hard precondition, and the feasibility script's motor-envelope scan had marked 80/100 mm foot-lift cells "physically feasible".

  Context

  Three added accounts re-decided everything. (1) Damping tax: in MIT mode tau = kp*(q_des-q) - kd*qd, so sustained rotation is capped at tau_limit/kd = 12/1.5 = 8 rad/s - below the firmware's 10; at peak speeds 6.7-7.9 rad/s the damping term alone eats 10.1-11.9 N*m (84-99% of the torque limit). "提固件 limit_spd 越不过这道税 —— 它是 kd 与限扭的比,不是固件旋钮." (2) Joint ROM: the feasibility script had checked motor envelopes but NOT joint range - the ankle-pitch ROM caps 1:2:1 leg-shortening lift at 62 mm (soft) / 77 mm (hard), so the 80/100 mm "feasible" cells were voided; also firmware-independent. (3) Ankle thermal: duty 0.40 puts ankle RMS at 87% of continuous rating (0.35 -> 93%); long-period big-stride cells hit both ankle torque peak and heat. Verdict: firmware raise DEQUEUED (50 mm design point needs knee 6.7-7.3 < the 8 rad/s effective ceiling < firmware 10); vel_limit stays 10 so sim == robot. The three accounts uniquely lock the design point - 50 mm lift / T 0.60 s / duty 0.40 - "三笔账 唯一锁定,不是调参空间", with pre-registered alternates allowed only inside the table and only after re-running the accounts.

  Change

  Design point frozen from accounts; hardware precondition reversed by arithmetic rather than by test; reference amplitude (0.84 rad = FK inverse of 50 mm) derived, per-joint action scales sized to the required travel (knee 0.9, hip_pitch 0.6, ankle deliberately NOT amplified - hard limit is adjacent).

  Outcome

  A firmware work item left the critical path; an infeasible region of the design space was closed before any training; the remaining risk (knee tracking lag from the damping tax) was pre-registered with its own criterion and in-table fallback (duty 0.35) - "这不是'奖励没调好', 是 plant 账".

  Mechanism

  PD actuators in MIT mode pay kd*velocity out of the same torque budget that tracks position, so the effective speed ceiling is a ratio of configuration constants, invisible to firmware settings; and feasibility is the intersection of ALL constraint families (torque envelope, joint ROM, thermal RMS) - a scan that omits one family certifies impossible cells.

  Applies when

  • planning running/jumping or any high-rate gait on PD actuators
  • a firmware or hardware upgrade is assumed as a training precondition
  • a feasibility scan covers motor limits but not ROM or heat

  “膝的有效速度顶 = τ_limit/kd = 12/1.5 = 8 rad/s,不是固件的 10。… 提固件 limit_spd 越不过这道税 —— 它是 kd 与限扭的比,不是固件旋钮。… 可行性脚本只查了电机包络没查关节 ROM —— 其 80/100mm 的"物理可行"格作废。… 判决:RS06 提固件对 run v0 不是前置,出队”

• Real robot walked at half the sim clock for two generations - resolved by racing a reward-side and a plant-side evidence line, not by guessingperiod-doubling-evidence-race

  Observed oncewalkattributionattributionactuator-modelingplant-calibrationprocess

  For a hardware-only pathology, refuse to guess: pre-register one probe per side of the sim2real boundary (can the reward mechanism change it on hardware? can fitted plant parameters reproduce it in sim?) and let the first positive result direct the next version.

  Symptom

  The number-one sim2real gap: on hardware v6/v7 stepped at 1.23-1.32 Hz - almost exactly half the 2.50 Hz gait clock they were trained and simulated at; sim never reproduced it, two generations running.

  Context

  Instead of committing training budget to a guess, v8 pre-registered two mutually controlled evidence lines and kept the clock OUT of the training variables: (a) reward-side - if the v8 saturation fix revives joint_pos_ref (the term that pins the gait to the clock), re-run hardware and see whether frequency returns to 2.5 Hz (hypothesis: v7's frozen actions meant NO reward was pinning the gait to the clock, and the real plant - with armature and friction making high frequencies expensive - slid down to the leg's pendulum natural frequency ~1.1 Hz); (b) plant-side - record suspended joint data (fit_actuator), fit armature/friction, load the fitted values into sim2sim and see whether the 1.25 Hz reproduces IN SIM. Decision rule fixed in advance: "谁先给出阳性结果谁定 v9 的方向 (奖励侧 vs plant 侧)" - whichever line goes positive first sets the next version's direction.

  Change

  Period-doubling excluded from the v8 change set; both diagnostic lines scheduled in parallel as non-blocking work; frequency reported factually in acceptance with no pass/fail attached ("倍周期是否消失 不设判定,它是 §9 的关键证据").

  Outcome

  The gap was routed into a decisive-experiment structure rather than a speculative retrain; the plant-side line pointed at exactly the unmodeled armature/friction that were later measured and installed as the plant baseline. Resolution (era-2c full-plant retest): the family had TWO causes - v8's low-speed period-doubling vanished once measured armature+friction were installed (1.30 -> 2.50 Hz, bifurcation-edge machine sensitivity), while v7's stood untouched at 1.20 Hz (saturation-freeze-driven policy property) - both evidence lines paid off, one per case.

  Mechanism

  A behavior appearing only on hardware has candidate causes on both sides of the sim2real boundary; changing training to fix it tests only one side per expensive cycle. Two cheap parallel probes - one intervening on the reward mechanism, one making sim reproduce the real behavior - localize the cause to a side before any training money is spent, and sim-reproduction of a real pathology is itself the strongest form of plant validation.

  Applies when

  • a gait pathology appears on hardware but never in any simulator
  • deciding whether a sim2real gap is reward-side or plant-side
  • tempted to change the gait clock/reward to chase a hardware symptom

  “倍周期(真机 1.23~1.32 Hz ≈ 时钟一半,v6/v7 连续两代;sim 从不出现):两条证据线互为对照——(a)… 真机重跑看频率是否回 2.5 Hz(假说:v7 没有任何奖励把步态钉在时钟上,真机 plant 有 armature/摩擦、高频贵,自由滑落到复摆自然频率 ~1.1 Hz);(b)真机吊挂录 fit_actuator.py … 看能否在仿真里复现 1.25 Hz。谁先给出阳性结果谁定 v9 的方向。”

• Audit rewards by realized contribution (weight x achieved value) - a weight of 2.0 was really paying 0.04realized-contribution-audit

  Mechanism understoodwalkreward-shapingreward-shapingattribution

  Evaluate a reward table by each term's realized per-step contribution under the current policy, never by its weight column; if a term's realized value is ~0, escalating its weight is a no-op - change the term's structure instead.

  Symptom

  Foot dragging persisted through repeated weight escalation: feet_air_time had been raised 0.25 -> 1.0 -> 2.0 across versions with no behavioral change, and the training-side comment even recorded the fact ("几乎没有 单支撑相, 是拖着脚蹭") without the fix changing form.

  Context

  Computing realized per-term contributions in the trained state exposed the economy: feet_air_time contributed weight 2.0 x achieved 0.019 = 0.038 per step, against tracking's +1.20 - lifting the leg earned 3% of what tracking earned, so dragging was the rational optimum no matter the weight escalation. The same table acquitted the energy penalties (total -0.30 negative vs +1.74 positive) that a naive read of weights (-5.0 orientation!) would have blamed.

  Change

  Fix redirected from "raise the weight again" to "add a term whose realized contribution changes the optimum": a clearance penalty sized so its realized magnitude (~0.018/foot when dragging) is comparable to feet_air_time's, enough to flip the optimum without drowning tracking.

  Outcome

  With the term economy corrected (plus posture/range fixes), swing height reached 34 mm and tracking 87% by v6; weight escalation of the old term was abandoned.

  Mechanism

  A reward weight is only a multiplier on whatever the policy currently achieves on that term; when the achieved value is near zero (behavior absent), escalating the weight multiplies near-zero. Optimizer behavior is governed by realized per-step magnitudes, so audits must be conducted in that currency.

  Applies when

  • a behavior persists despite repeated weight increases
  • auditing whether penalties are "too strong" or rewards "too weak"
  • sizing a new reward term against existing ones

  “把 feet_air_time 权重从 0.25 → 1.0 → 2.0 一路加,但没有加高度项。量级算下来:feet_air_time 权重 2.0 × 实得 0.019 = 0.038,而跟踪奖励是 1.2。抬腿的边际收益只有跟踪的 3%,拖地当然是最优解。… 正项 +1.74,负项 −0.30。能量惩罚不是瓶颈,抬腿没收益才是。”

• A soft joint-limit penalty charged the standing pose itself - the geometric-zero knee sat on its hard limit, so stand_v1 bent its knees to dodge 0.419 per step and leaned 4.1 deg forward; excluding the knee gave 0.24 degsoft-limit-penalty-charges-nominal-pose

  Mechanism understoodstandreward-shapingreward-shapingplant-calibration

  Before training, evaluate every penalty at the nominal pose; if a joint's soft limit sits inside the pose the task requires (a straight knee on its hard stop), exclude that joint from the soft-limit penalty and let the action clip enforce the hard limit.

  Symptom

  stand_v1 (retrained after the default pose moved to the CAD geometric zero and mirror augmentation was added) fixed left/right asymmetry (6.8 -> 0.0 deg) but settled at a 4.1 deg forward lean, where pure PD at the same default settled at 0.1 deg - the policy was actively pushing itself forward, which is exactly the real robot's failure direction.

  Context

  soft_joint_pos_limit_factor = 0.9 shrank the knee's soft limit to -/+0.1047 rad, while the geometric-zero default has the knee at q = 0, exactly on the hard limit. Standing in the nominal pose therefore paid 0.2094 x 2.0 = 0.419 per step in dof_pos_limits (alive earned only 0.5). The policy's way out was to bend the knees to -/+0.1013 rad, and the cost was the forward lean.

  Change

  stand_v1b: the knees excluded from dof_pos_limits (a straight knee IS the standing pose; the hard limit is still enforced by the action clip). No other change.

  Outcome

  stand_v1b: max tilt 0.3 deg, steady tilt 0.24 deg, asymmetry 0.1 deg, height 0.384 m - exactly nominal - with knees at -0.0007 / +0.0005 rad. It became the standing release used on the real robot, and later the standing side of the recovery switch. The same exclusion was carried into the recovery contract (knee at the clip in the standing pose) and the one-leg reward table.

  Mechanism

  A limit penalty whose soft boundary lies inside the nominal pose turns the nominal into a taxed state, and the policy buys its way out with whatever posture change is cheapest - here a knee bend paid for with lean.

  Applies when

  • a standing or default pose has a joint at or near its hard limit
  • a policy settles in a small steady tilt that pure PD does not show
  • soft-limit factors shrink limits uniformly across joints

  “`soft_joint_pos_limit_factor=0.9` 把膝软限位内缩到 ∓0.1047,而几何零位 default **膝盖 q=0 正好压在硬限位上** ⇒ 站在标称姿态每步白扣 `0.2094 × 2.0 = 0.419` (alive 才 +0.5)。策略只能屈膝到 ∓0.1013 躲罚,代价是躯干前倾 —— 恰好是真机的 失效方向。 … 修掉"软限位罚标称姿态"后重训(`dof_pos_limits` 排除膝盖)。**前倾问题彻底消失** … 不再屈膝躲惩罚,高度正好落回标称 0.3840。”

• Real-robot trials of a new skill were staged by risk - a hanging dry run with the robot posed by hand, then one short try per category on a mat with the hardest last, then the composed behaviour (switch + walking) last - with the user present and a log every timestaged-hang-mat-floor-for-get-up

  Replicatedrecoveryreal-deployreal-acceptanceprocess

  Stage a new skill's hardware trials by risk - hanging dry run posed by hand, one short try on a mat per start category with the hardest last, the composed behaviour last - with an operator ready to cut enable and a log for every try; relax a safety ban only for short, attended runs and say so in writing.

  Symptom

  A get-up policy acts violently near the ground by design, and the first unstaged real run of the line was stopped as dangerous.

  Context

  The hanging checklist written with the first stamped recovery product (v2_5, 2026-08-11), to be ticked item by item with the user present: both machines on the same commit and firmware torque limits checked; the robot hung from a single point about 0.1 m off the ground; a dry run with the robot posed by hand into supine and prone to watch that the target stream is gentle (the beta contract keeps targets within +/-0.25 rad of the measured pose, so enabling causes no homing fling); the first floor try is supine only, on a mat, once, with torque and joint logs; categories are added one at a time, prone last; any kicking or oscillation cuts enable immediately. For the switch (08-14) the runbook orders: hang with the standing policy as the locomotion side, then on a mat push the robot over and let it recover, and only last swap in the walking policy. An exemption was also written: edge-standing policies stay banned from long or unattended runs, but a short single A/B with the user present, hung or on a mat, is allowed. The one-leg line reused the same order (hang, then floor with a spotter, 60 s segments with a temperature check).

  Change

  Real trials as a checklist of stages, each gated on the previous one, with the composed behaviour last.

  Outcome

  The line's first real get-up (v2_6, 08-11) came through this protocol and was reported "fairly stable"; no further hardware outcomes of the switch are recorded in the spec.

  Mechanism

  Each stage exposes one new risk (commanded targets without contact, a single category with contact, harder categories, then the interaction of two policies), so a failure is attributable and cheap.

  Applies when

  • first hardware trial of a recovery, jumping or other high-impact skill
  • switching between two policies on hardware for the first time
  • a policy with a known posture defect needs a comparison run

  “吊挂空跑: 手动摆到 supine/prone 姿态, 看目标流是否温和 (β 帽 7.5 N·m, 目标永远贴着当前 q ±0.25 rad —— 使能瞬间无归位甩动, 这是 β 契约附带保证) … 落地首试: supine 一类, 垫子, 单次; τ/q --log 全程记录 … 逐类别扩展 (prone 最后), 每类先单次”

• A stronger action_rate penalty cut the median torque demand under the gate and left the p99 at 4x the limit - only a hinge on the pre-clip (computed) torque, weighted by comparison with a peer term, collapsed the tailtail-torque-needs-hinge-on-computed-demand

  Mechanism understoodrecoveryreward-shapingactuator-modelingreward-shapingmeasurement

  Judge actuator demand against the deployed limit, read the pre-clip demand (applied torque is censored and gives no gradient on the excess), use an L2 rate penalty for the median and a thresholded hinge on computed demand for the tail, and set a new term's weight from its measured steady magnitude next to a peer term rather than from a back-of-envelope estimate.

  Symptom

  After R0.5 hip_pitch delivered torque sat at its 12 N*m limit in a typical get-up (demand 119-125% of the limit, p99 4.2x) - zero control margin at exactly the moment modelling error matters.

  Context

  The 12/17/11 N*m limits are deployment limits written into robot.yaml by set_torque (RS06 at 33% of rated), and simulation uses the same effort_limit - so the gate is judged against them, not the 36 N*m rating (an early reading against the rating was retracted). Applied torque is clipped at the limit - censored data - so demand must be read from computed_torque. With the full-range action contract (hip_pitch scale 1.309, kp 30) a single-step action change of 0.306 already saturates hip_pitch, and action_rate penalizes exactly that change.

  Change

  R3.0: action_rate_l2 -0.01 -> -0.03 (child-run). R3.1: new torque_headroom = sum relu(|tau_computed|/limit - 0.9)^2, normalized so three motor types share a scale. Its weight was first estimated at -0.1, measured in a 12-iteration run at an effective -0.019 (12x smaller - the estimate had mixed a per-episode-peak p99 with a per-step p99, and at 1% of upright it would have been numerically absent), and set to -0.5 so its steady value (-0.095) matched action_rate's (-0.097).

  Outcome

  R3.0: sum |da|^2 -64%, success 99.6 -> 100%, delivered median = demand median (the clamp no longer fired in a typical episode), gate PASS at worst 79.9% - but p99 unchanged (hip_pitch 419-432% -> 427-436%). R3.1: p99 hip_pitch -> 148-189% (-57 to -65%), knee 422-439% -> 233-234%, saturation duty -60%, success 100%; the worst joint became hip_roll at 67.7%. Its cost appears in torque-penalty-bought-by-leg-bracing.

  Mechanism

  A squared-rate penalty presses the whole-episode sum and moves the typical step, not rare spikes; the spikes came from the kp term (large targets while a limb is blocked by the ground - velocity alone could not reach them under vel_limit), and a penalty on applied torque cannot see demand above the clip because every excess sample reads as exactly the limit.

  Applies when

  • torque demand saturates actuator limits in high-effort skills
  • a smoothness penalty improves medians but not peaks
  • a new reward term's weight is set by estimate alone

  “**必须用 `computed_torque` 而不是 `applied_torque`**:后者被 `effort_limit` 削平, 是删失数据,超限样本全被压成"恰好等于限",对超限部分梯度恒为 0。 … 改按同侪定标取 **−0.5**(稳态 ≈ −0.095,与 `action_rate_l2` 的 −0.097 等量)。”