clm_2e7fe3e736c84d9fb7ad7783eb7242d1_r1

NIEK2-0127

① SA Source

• Source: 開啟完整 SA 文章
• Section: Speculative Decoding
• Line hint: 137

Context Before

Speculative Decoding

A different way LPUs could improve decode phase latencies is by accelerating a speculative decoding setup, where we deploy draft models or Multi-Token Prediction (MTP) layers onto LPUs.

Evidence

To verify the draft tokens, the main model only needs one warm prefill of k new tokens, at the latency cost of roughly a single decode step

Context After

Source: SemiAnalysis

② Atomic Claim

在驗證 draft tokens 時，主模型只需要對這 k 個新 token 執行一次 warm prefill，其延遲成本約等同一次 decode step。

• Epistemic Mode: ASSERTED
• Mapping Status: COMPLETE

③ Semantic Frame

{
  "attribute": "LATENCY",
  "context_nodes": [
    {
      "id": "04_knowledge_base/Decode",
      "label": "decode"
    }
  ],
  "entity": {
    "id": "04_knowledge_base/Prefill",
    "label": "prefill"
  },
  "frame_type": "ATTRIBUTE",
  "qualifiers": {
    "condition_text": null,
    "numeric_mentions": [],
    "temporal_mentions": []
  },
  "value": {
    "numeric_mentions": [],
    "value_text": "在驗證 draft tokens 時，主模型只需要對這 k 個新 token 執行一次 warm prefill，其延遲成本約等同一次 decode step。"
  }
}

④ Canonical Entity Mapping

Role	Surface Label	Canonical Target
entity	prefill	Prefill
context_0	decode	Decode

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。