Вариант B: абляция осей GF+A/PolyQ по единой downstream-BPT (SW-proxy) - #275
Open
gHashTag wants to merge 1 commit into
Open
Вариант B: абляция осей GF+A/PolyQ по единой downstream-BPT (SW-proxy)#275gHashTag wants to merge 1 commit into
gHashTag wants to merge 1 commit into
Conversation
…roxy) Micro-LM (4L d=128, seed=20260730), single downstream metric bits/token, budget-aligned at mean 4/6/8 bits. GF+A (per-row cross-format pocket choice) vs PolyQ-proxy (per-column activation-aware bit-width water-filling). Result: dBPT below Parameter Golf significance threshold (0.0195 BPT) on all budgets -> axes downstream-indistinguishable (orthogonal). No superiority claim.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Что
Прямая бит-выровненная абляция двух осей адаптивности по ЕДИНОЙ downstream-метрике (bits/token) на микро-LM (4L d=128, seed=20260730).
Результат
[измерено — SW proxy, CPU]Все значения ниже порога значимости Parameter Golf → оси downstream-неотличимы (ортогональны); ни одна не доминирует. БЕЗ заявлений «обходит».
Границы честности
(i) микро-масштаб = направление, НЕ величина; (ii) PolyQ-proxy без компиляторных кернелов (замер качества, не скорости); (iii) оверхед заголовков не вычтен; (iv) Hessian снят только с FFN (fused MHA не ловится хуком).