From bde56bc2bc3b02a7c19cbbcd8c55ed7885dd78f9 Mon Sep 17 00:00:00 2001 From: N0zoM1z0 <161784452+N0zoM1z0@users.noreply.github.com> Date: Fri, 16 Oct 2026 23:34:49 +0000 Subject: [PATCH] docs(experiment): update capability coverage after return fixes Co-authored-by: gensokyo-dev <342580529+gensokyo-dev@users.noreply.github.com> Co-authored-by: Kaoru0822 <208281322+kaoru0822-kitauji@users.noreply.github.com> --- experiments/atomic-codemode/README.md | 11 +++++++++-- experiments/atomic-codemode/capability-map.md | 13 ++++++++++--- 2 files changed, 19 insertions(+), 5 deletions(-) diff --git a/experiments/atomic-codemode/README.md b/experiments/atomic-codemode/README.md index e355c2b823..74f733b615 100644 --- a/experiments/atomic-codemode/README.md +++ b/experiments/atomic-codemode/README.md @@ -26,8 +26,15 @@ evidence; the shared bounded producer preserves space/width and uncertainty. All three real models now complete the behavior-level task. Code Mode takes 151.3 seconds / 388,365 cumulative input tokens, Direct 164.9 / 515,929 and traditional 345.7 / 681,613. Code has more uncached input than traditional and -more memory than Direct. A native BTI/PLT identity/no-return boundary remains -under investigation. +more memory than Direct. These are the earlier pinned measurements. + +The [return-coverage study](aarch64-return-study.md) subsequently fixes two +independent producer defects: BTI import-entry identity and false no-return +promotion that erased the post-warning path. All three fresh model answers pass +the same selected task. Direct uses 351,157 cumulative input tokens / 139.0 +seconds, Code Mode 371,571 / 166.8, and traditional 662,683 / 263.1. Code Mode +reduces MCP requests but adds model responses in this sample; correct primitives +and fewer RPCs do not establish a stable adapter advantage. The [architecture storage study](storage-study.md) adds exact native register-name and address-space queries, sharing one bounded immutable capture per lease. diff --git a/experiments/atomic-codemode/capability-map.md b/experiments/atomic-codemode/capability-map.md index 98c435cb89..abb231c78e 100644 --- a/experiments/atomic-codemode/capability-map.md +++ b/experiments/atomic-codemode/capability-map.md @@ -104,6 +104,13 @@ SSH checks verify both warning owners, callers, selected machine words and 77 GOT-slot reads. Coverage stays bounded and does not propagate live GOT contents. Traditional C/assembly preflight can answer the same new behavior-level question, avoiding a baseline missing necessary facts. All three real models -complete it, while token/time/resource tradeoffs remain mixed. All preserve the -recovered no-return disagreement and use machine facts for the post-call path; -a native BTI/veneer identity split is a concrete next investigation. +complete it, while token/time/resource tradeoffs remain mixed. That pinned run +preserves the recovered no-return disagreement and uses machine facts for the +post-call path. + +The [return-coverage study](aarch64-return-study.md) fixes the BTI entry split +and a separate false no-return promotion through the native owners. The fresh +three-way task passes with the strict call restored in both atomic captures; +their complete selected function documents agree. Direct has lower input, +time and CPU than Code Mode in this sample. Memory, analysis-profile differences, +the common client executor and n=1 sampling prevent a universal adapter claim.