NIEK2-0124
① SA Source
- Source: 開啟完整 SA 文章
- Section:
Speculative Decoding - Line hint:
135
Context Before
Source: SemiAnalysis
Speculative Decoding
Evidence
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
Context After
For a decoding step of context N tokens, adding k additional tokens during forward pass (a warm prefill of k new tokens) marginally increases the latency when k << N. Using this property, speculative decoding uses a small draft model or MTP layers to predict k new tokens, saving time since small models have lower latency per decode step. 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. Speculative decoding usually boosts output token per decode step by 1.5 to 2 tokens, depending on the draft model / MTP accuracy. With its low latency capabilities, LPUs can further increase the latency savings and improve throughput.
② Atomic Claim
LPU 改善 decode 延遲的另一種可能方式，是加速 speculative decoding，將 draft model 或 Multi-Token Prediction（MTP）layers 部署到 LPU 上。
- Epistemic Mode:
HYPOTHETICAL - Mapping Status:
PARTIAL
③ Semantic Frame
{
"frame_type": "NARY_RELATION",
"participants": [
{
"node": {
"id": "04_knowledge_base/Decode",
"label": "decode"
},
"role": "subject"
},
{
"node": {
"id": "04_knowledge_base/Speculative Decoding",
"label": "speculative decoding"
},
"role": "participant"
},
{
"node": {
"id": "04_knowledge_base/Multi-Token Prediction",
"label": "Multi-Token Prediction"
},
"role": "participant"
}
],
"qualifiers": {
"condition_text": null,
"numeric_mentions": [],
"temporal_mentions": []
},
"relation_type": "DEPLOYS"
}④ Canonical Entity Mapping
| Role | Surface Label | Canonical Target |
|---|---|---|
| subject | decode | Decode |
| participant | speculative decoding | 04_knowledge_base/Speculative Decoding |
| participant | Multi-Token Prediction | 04_knowledge_base/Multi-Token Prediction |
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。