From 9c0aaacdfb7b1726f1f1e99af6a24d0ee35fb03f Mon Sep 17 00:00:00 2001 From: jgrusewski Date: Wed, 13 May 2026 08:26:46 +0200 Subject: [PATCH] =?UTF-8?q?docs(sp22):=20smoke=20scope=20=E2=80=94=20?= =?UTF-8?q?=CE=B1=20ACTIVE=20=CE=B2=20DEAD=20(B6=20not=20landed)?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Auditing the β reward shaping path revealed a producer gap: - experience_kernels.cu:3614-3628 correctly computes r_aux_align from scale_β = isv_signals[537] - But slot 537 has NO PRODUCER — SP11 controller still emits 6 components (N_COMPONENTS=6.0f), Phase B6 extension to 7 components hasn't landed - → scale_β stays at alloc_zeros 0 → r_aux_align = 0 → β no-op Current smoke validates α atom-shift mechanism ONLY. β remains dormant. Updated verdict outcome table: WR-shift implies α works alone; WR-stable with dist-shift implies α active but β needed (land B6 next, ~1-2hr). Phase B6 scope documented (single-kernel edit to SP11 controller + launcher update for N_COMPONENTS). Co-Authored-By: Claude Opus 4.7 (1M context) --- docs/dqn-wire-up-audit.md | 28 ++++++++++++++++++++++++++++ 1 file changed, 28 insertions(+) diff --git a/docs/dqn-wire-up-audit.md b/docs/dqn-wire-up-audit.md index 004a9bc99..d075ee053 100644 --- a/docs/dqn-wire-up-audit.md +++ b/docs/dqn-wire-up-audit.md @@ -17291,3 +17291,31 @@ Workflow: `train-qsltr` @ commit `5106e3b117e89ae5a7e6f0de9f09ed0fe600ad87`, 3 e **Diagnostic to add for next smoke** (deliberately deferred this run): - W[0..4] direct readback per epoch via mapped pinned [4] f32 + DtoD copy outside captured graph + HEALTH_DIAG log line. Would directly show Adam updates on W. + +#### Smoke scope clarification — α ACTIVE, β DEAD (2026-05-13) + +While preparing the verdict template, audited the β reward shaping path's producer chain. Finding: + +**β producer wired but UNFED**: `experience_kernels.cu:3614-3628` correctly computes `r_aux_align = scale_β × alignment × profit_pos` and writes to `reward_components_per_sample[out_off * 7 + 6]`. But `scale_β = isv_signals[SP22_AUX_ALIGN_SCALE_INDEX=537]` and slot 537 has NO PRODUCER — the SP11 `reward_subsystem_controller_kernel.cu` still treats `N_COMPONENTS = 6.0f` (popart/cf/trail/micro/opp_cost/bonus). Phase B6 (extend SP11 controller to 7 components with w_aux_align anchor) hasn't landed. + +**Net effect**: slot 537 stays at alloc_zeros sentinel 0 throughout training → `r_aux_align = 0` always → β reward shaping is a runtime no-op. + +**Implication for smoke verdict**: This smoke validates the **α atom-shift mechanism ONLY**. The β reward bonus (aux-aligned profitable trades earn extra reward) is dormant. If WR shifts off baseline, α is the cause; if it doesn't, the verdict is α alone is insufficient — β extension (Phase B6) becomes the next priority before investing in Phase D. + +**Phase B6 scope** (deferred): Extend `reward_subsystem_controller_kernel.cu`: +1. `N_COMPONENTS: 6.0 → 7.0`. +2. Read 7th component magnitude from `ISV[REWARD_AUX_ALIGN_EMA_INDEX=536]` (EMA already produced by reward_component_ema_kernel.cu's extended 0..7 loop — Phase B3 landed). +3. Compute w_aux_align via the same blend formula as other components (winner × improving + diversifier × stagnant); anchor target = ~10% of total reward magnitude; bounds `[0.05, 2.0]`. +4. Write to `ISV[SP22_AUX_ALIGN_SCALE_INDEX=537]`. + +Estimated ~1-2 hr engineering (single-kernel edit + launcher update if N_COMPONENTS is kernel-arg vs constant). + +**Updated outcome table** (replacing "Possible outcomes" in the verdict template): + +| Pattern | Interpretation | Next action | +|---------|----------------|-------------| +| WR ≥ 53% epoch 3, dist shifts | α atom-shift alone moves WR | Land B6 for β + Phase D for eval validation | +| WR 50-53%, dist shifts | α partial effect | Tune W magnitudes OR land B6 to add β reward signal | +| WR ~46%, dist shifts | Mechanism active but no WR translation | β probably necessary — land B6 next | +| WR ~46%, no dist shift | α inactive (state_121 not flowing) | Investigate aux head + state_121 wiring | +| Training collapse | Loss landscape unstable | Reduce W init magnitude or revert |