liqiang
b119135836
Build latest book artifacts / build (push) Canceled after 0s
dependency resolution / resolve (3.11) (push) Canceled after 0s
dependency resolution / resolve (3.13) (push) Canceled after 0s
deploy-pages / build (push) Canceled after 0s
deploy-pages / deploy (push) Canceled after 0s
i18n consistency check / check (push) Canceled after 0s
provider adoption tests / test (chapter2/context-compression) (push) Canceled after 0s
provider adoption tests / test (chapter2/prompt-injection) (push) Canceled after 0s
provider adoption tests / test (chapter2/system-hint) (push) Canceled after 0s
provider adoption tests / test (chapter3/log-sanitization) (push) Canceled after 0s
web-search-agent tests / test (push) Canceled after 0s
web-search-agent tests / agentbook (push) Canceled after 0s
40 lines
7.9 KiB
XML
40 lines
7.9 KiB
XML
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 40 780 360" width="780" height="360" style="background:#ffffff">
|
||
<defs><marker id="ah" markerWidth="12" markerHeight="8" refX="12" refY="4" orient="auto"><polygon points="0 0, 12 4, 0 8" fill="#333333"/></marker><marker id="ah-light" markerWidth="12" markerHeight="8" refX="12" refY="4" orient="auto"><polygon points="0 0, 12 4, 0 8" fill="#999999"/></marker></defs>
|
||
|
||
<rect x="60" y="160" width="180" height="160" rx="6" fill="#d0d0d0" stroke="#333333" stroke-width="2"/>
|
||
<text x="150" y="182" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="20" fill="#333333" text-anchor="middle" dominant-baseline="central" font-weight="bold">視覚エンコーダ</text>
|
||
<text x="150" y="202" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">(CLIP ViT)</text>
|
||
<text x="150" y="228" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="12" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">パラメータ: 凍結（学習しない）</text>
|
||
<text x="150" y="248" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">入力: 224×224 画像</text>
|
||
<text x="150" y="268" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">出力: 196 個の視覚トークン</text>
|
||
<text x="150" y="288" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">次元: 768</text>
|
||
<rect x="290" y="160" width="160" height="160" rx="6" fill="#f0f0f0" stroke="#333333" stroke-width="2"/>
|
||
<text x="370" y="182" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="20" fill="#333333" text-anchor="middle" dominant-baseline="central" font-weight="bold">射影層</text>
|
||
<text x="370" y="202" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">（全結合）</text>
|
||
<text x="370" y="228" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="10" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">唯一ゼロから学習する部分</text>
|
||
<text x="370" y="248" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">Linear(768, 512)</text>
|
||
<text x="370" y="268" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="11" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">視覚 → 言語空間へ整合</text>
|
||
<text x="370" y="288" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">パラメータ数: 約400K</text>
|
||
<rect x="500" y="160" width="220" height="160" rx="6" fill="#f0f0f0" stroke="#333333" stroke-width="2"/>
|
||
<text x="610" y="182" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="20" fill="#333333" text-anchor="middle" dominant-baseline="central" font-weight="bold">言語モデル</text>
|
||
<text x="610" y="202" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">(MiniMind 100M)</text>
|
||
<text x="610" y="228" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">SFT 段階で凍結解除</text>
|
||
<text x="610" y="248" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">hidden_dim=512</text>
|
||
<text x="610" y="268" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="12.5" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">自己回帰的な記述生成</text>
|
||
<text x="610" y="288" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">Muon オプティマイザで3倍高速化</text>
|
||
<line x1="242" y1="240" x2="288" y2="240" stroke="#333333" stroke-width="2" marker-end="url(#ah)"/>
|
||
<line x1="452" y1="240" x2="498" y2="240" stroke="#333333" stroke-width="2" marker-end="url(#ah)"/>
|
||
<text x="265" y="225" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">視覚</text>
|
||
<text x="265" y="255" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">トークン</text>
|
||
<text x="475" y="225" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">言語</text>
|
||
<text x="475" y="255" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">埋め込み</text>
|
||
<rect x="60" y="70" width="180" height="60" rx="4" fill="#f5f5f5" stroke="#333333" stroke-width="2"/>
|
||
<text x="150" y="90" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#333333" text-anchor="middle" dominant-baseline="central" font-weight="normal">入力画像</text>
|
||
<text x="150" y="110" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="14" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">（滝／都市／自然）</text>
|
||
<line x1="150" y1="132" x2="150" y2="158" stroke="#333333" stroke-width="2" marker-end="url(#ah)"/>
|
||
<rect x="500" y="70" width="220" height="60" rx="4" fill="#f5f5f5" stroke="#333333" stroke-width="2"/>
|
||
<text x="610" y="90" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#333333" text-anchor="middle" dominant-baseline="central" font-weight="normal">出力される記述</text>
|
||
<text x="610" y="110" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="7.5" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">「巨大な滝の脇に虹がかかっている」</text>
|
||
<line x1="610" y1="158" x2="610" y2="132" stroke="#333333" stroke-width="2" marker-end="url(#ah)"/>
|
||
<text x="390" y="378" font-family="Arial, 'Helvetica Neue', Helvetica, 'PingFang SC', 'Microsoft YaHei', sans-serif" font-size="16" fill="#666666" text-anchor="middle" dominant-baseline="central" font-weight="normal">戦略: LLM を凍結し射影層を学習 → SFT 中に LLM を凍結解除（破滅的忘却を回避）</text>
|
||
</svg> |