VR2R-0068_r1

VR2R-0068

① SA Source

• Source: 開啟完整 SA 文章
• Section: Rubin
• Line hint: 61

Context Before

Notably, the Tensor core width doubling only applies to FP4 and FP8, with BF16 and TF32 remaining the same as Blackwell, resulting in performance scaling only 1.6x of Blackwell.

Evidence

This architectural decision reflects NVIDIA’s belief that most training

Context After

and inference workloads will move away from TF32 and BF16 and onto FP8 and FP4.

② Atomic Claim

這項 architectural decision 反映 NVIDIA 對大多數 training……

• Epistemic Mode: ASSERTED
• Mapping Status: PARTIAL

③ Semantic Frame

{
  "frame_type": "CLAIM_ONLY",
  "qualifiers": {
    "condition_text": "句子在 training 後切斷，後半段位於 VR2R-0069",
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "reason": "EXTRACTION_FRAGMENT_CONTINUES_IN_NEXT_CLAIM",
  "statement": "這項 architectural decision 反映 NVIDIA 對大多數 training……"
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
—	—	—

⑤ Human Review

請在 Properties 逐項確認：

• 原文 → Atomic Claim 是否忠實
• Atomic Claim → Semantic Frame 是否忠實
• Canonical Entity mapping 是否正確
• Epistemic mode 是否保留原文語氣
• 最後選擇 review_action

Review state

Markdown 內文不是正式 approval。只有 Apply bridge 寫入的 Decision Ledger event 才是正式決策。