Gahahahahaha

index.html

@@ -34,11 +34,12 @@
   /* ====== Layout ====== */
   .wrap {
     max-width: 1200px;
-    margin: 32px auto 48px;
+    margin: 15px auto 48px;
     padding: 0 16px;
   }
+
   .title {
-    margin: 0 0 12px;
+    margin: 0 0 25px;
     font-size: 22px;
     font-weight: 800;
     letter-spacing: .2px;
@@ -113,9 +114,327 @@
   /* Make links subtle if you add them later */
   a { color: var(--accent); text-decoration: none; }
   a:hover { text-decoration: underline; }
+
+
+/* ===== Container aligns with your table width ===== */
+.lb-container, .lb-divider {
+  max-width: 1200px;
+  margin: 0 auto;
+  padding: 0 12px;
+}
+
+/* ----- Hero ----- */
+.lb-hero { margin: 28px 0 10px; }
+.lb-hero h1 {
+  margin: 0 0 6px;
+  font-size: 30px;
+  font-weight: 800;
+  letter-spacing: .2px;
+}
+
+.lb-tagline {
+  margin: 10px 0 2px;
+  max-width: 860px;
+  text-align: left;                 /* left-align instead of center */
+  color: rgba(255, 255, 255, 0.85);
+  font-size: 15px;
+  line-height: 1;
+}
+
+.lb-tagline p {
+  margin: 4px 0;
+}
+
+
+.lb-meta {
+  margin-top: 10px;
+  display: flex; align-items: center; gap: 8px; flex-wrap: wrap;
+}
+.badge {
+  font-size: 12px;
+  padding: 4px 8px;
+  border-radius: 999px;
+  border: 1px solid var(--border, rgba(255,255,255,.08));
+  background: var(--panel, rgba(255,255,255,.03));
+}
+.meta-spacer { flex: 1 1 auto; }
+.updated { font-size: 12px; opacity: .75; }
+
+/* ----- Cards grid ----- */
+.card-grid {
+  display: grid;
+  gap: 12px;
+  grid-template-columns: repeat(3, 1fr);
+  margin: 16px 0 12px;
+}
+@media (max-width: 980px) { .card-grid { grid-template-columns: 1fr; } }
+
+.card {
+  background: var(--panel, rgba(255,255,255,.03));
+  border: 1px solid var(--border, rgba(255,255,255,.08));
+  border-radius: 14px;
+  padding: 14px;
+  box-shadow: var(--shadow, 0 2px 12px rgba(0,0,0,.25));
+}
+.card h3 { margin: 0 0 2px; font-size: 16px; }
+.muted { opacity: .75; font-size: 13px; margin: 0 0 8px; }
+
+/* chips */
+.chips { display: flex; flex-wrap: wrap; gap: 8px; }
+.chip {
+  font-size: 12px;
+  padding: 4px 8px;
+  border-radius: 8px;
+  background: rgba(255,255,255,.06);
+  border: 1px solid rgba(255,255,255,.08);
+}
+
+/* ----- Definitions card ----- */
+.card-defs h3 { margin: 2px 0 8px; }
+.defs { display: grid; grid-template-columns: 220px 1fr; gap: 8px 16px; }
+@media (max-width: 800px) { .defs { grid-template-columns: 1fr; } }
+.defs dt { font-weight: 700; }
+.defs dd { margin: 0; opacity: .92; }
+.defs code {
+  background: rgba(255,255,255,.06);
+  padding: 1px 4px; border-radius: 4px;
+}
+
+/* ----- Links row ----- */
+.link-row {
+  display: flex; flex-wrap: wrap; gap: 10px;
+  margin: 12px 0 18px;
+}
+.btn {
+  display: inline-block;
+  padding: 8px 12px;
+  border-radius: 10px;
+  border: 1px solid var(--border, rgba(255,255,255,.10));
+  background: linear-gradient(180deg, rgba(255,255,255,.06), rgba(255,255,255,.02));
+  color: var(--accent, #8ab4ff); text-decoration: none;
+  font-weight: 600; font-size: 14px;
+}
+.btn:hover { text-decoration: underline; }
+.btn.ghost { background: transparent; }
+
+/* ----- Divider ----- */
+.lb-divider {
+  height: 1px; border: none;
+  background: linear-gradient(to right, transparent, rgba(255,255,255,.12), transparent);
+  margin: 18px 0 14px;
+}
+
+.card h3 {
+  margin-bottom: 10px; 
+}
+
+.equation-svg {
+  display: flex;           /* flex container */
+  align-items: center;     /* vertically aligns with bullets */
+  margin: 1.5em 2em 1em;  /* top/bottom spacing, tighter */
+}
+
+.equation-svg img {
+  max-width: 100%;
+  height: auto;
+  display: block;
+}
+
+/* Doc-style content card */
+.card-doc {
+  margin-top: 16px;
+}
+
+.card-doc h3 {
+  margin: 4px 0 10px;
+  font-size: 18px;
+  font-weight: 700;
+}
+
+.card-doc .lead {
+  margin: 6px 0 12px;
+  opacity: .92;
+  line-height: 1.6;
+}
+
+.contents-list {
+  margin: 0 0 12px 0;
+  padding-left: 22px;       /* neat bullets */
+}
+
+.contents-list li {
+  margin: 6px 0;
+  line-height: 1.6;
+}
+
+.ref-note {
+  margin-top: 12px;
+  padding-top: 10px;
+  border-top: 1px solid rgba(255,255,255,.08);
+  opacity: .9;
+}
+
+.ref-note a {
+  color: var(--accent, #8ab4ff);
+  text-decoration: underline;
+}
+.ref-note a:hover { text-decoration: none; }
+
 </style>
+
+<script src="https://polyfill.io/v3/polyfill.min.js?features=es6"></script>
+  <script id="MathJax-script" async
+          src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>
+
+
+
+          
+
 </head>
 <body>
+
+
+
+
+<!-- ===== Polished Info Header & Sections ===== -->
+<section class="lb-container">
+
+  <!-- Title + meta badges -->
+  <header class="lb-hero">
+    <h1>41 Open-Source LLMs Evaluated Locally on 19 Benchmarks ⚡</h1>
+    <p class="lb-tagline">
+      Evaluations run with the lm-evaluation-harness library on a local workstation.
+    </p>
+    <p class="lb-tagline">Benchmarks are grouped into three categories, with the corresponding tasks and metrics listed below.
+</p>
+  </header>
+
+
+
+  <!-- Three cards: task groups -->
+  <div class="card-grid">
+    <article class="card">
+      <h3>Reasoning & Math</h3>
+
+      <div class="chips">
+        <span class="chip">gsm8k(exact_match,strict-match)</span>
+        <span class="chip">bbh(exact_match,get-answer)</span>
+        <span class="chip">arc_challenge(acc_norm,none)</span>
+        <span class="chip">anli_r1(acc,none)</span>
+        <span class="chip">anli_r2(acc,none)</span>
+        <span class="chip">anli_r3(acc,none)</span>
+        <span class="chip">gqpa_main_zeroshot(acc_norm,none)</span>
+      </div>
+    </article>
+
+    <article class="card">
+      <h3>Commonsense & Natural Language Inference</h3>
+
+      <div class="chips">
+        <span class="chip">hellaswag(acc_norm,none)</span>
+        <span class="chip">piqa(acc_norm,none)</span>
+        <span class="chip">winogrande(acc,none)</span>
+        <span class="chip">boolq(acc,none)</span>
+        <span class="chip">openbookqa(acc_norm,none)</span>
+        <span class="chip">sciq(acc_norm,none)</span>
+        <span class="chip">qnli(acc,none)</span>
+      </div>
+    </article>
+
+    <article class="card">
+      <h3>Knowledge & Reading</h3>
+
+      <div class="chips">
+        <span class="chip">mmlu(acc,none)</span>
+        <span class="chip">nq_open(exact_match,remove_whitespace)</span>
+        <span class="chip">drop(f1,none)</span>
+        <span class="chip">truthfulqa_mc1(acc,none)</span>
+        <span class="chip">truthfulqa_mc2(acc,none)</span>
+        <span class="chip">triviaqa(exact_match,remove_whitespace)</span>
+      </div>
+    </article>
+  </div>
+
+  <!-- Table field definitions -->
+  <section class="card card-defs">
+    <h3>Table Fields</h3>
+    <dl class="defs">
+      <dt>Model Name</dt>
+      <dd>Listed as <code>Company_ModelName</code>. If quantized, <code>(8bit)</code> is appended.</dd>
+
+      <dt>Total Time</dt>
+      <dd>Wall-clock time for the full evaluation run per model.</dd>
+
+      <dt>GPU Util Time</dt>
+      <dd>
+        Equivalent RTX 5090 GPU time at 100% utilization. GPU utilization was logged once per minute,
+    and the effective utilization time was calculated using the following equation:
+
+        <div class="equation-svg">
+    <img src="equation.svg" alt="T_util-100% = Σ (u_i / 100) Δt" />
+  </div>
+        <ul>
+          <li><span class="muted">interval:</span> 1 minute</li>
+          <li><span class="muted">u<sub>i</sub>:</span> GPU utilization (%) during interval i</li>
+          <li><span class="muted">Δt:</span> duration of interval i (s)</li>
+        </ul>
+      </dd>
+
+      <dt>Mean Score</dt>
+      <dd>Arithmetic mean across the 19 tasks (0–1; higher is better). Unweighted.</dd>
+
+      <dt>Overall Rank</dt>
+      <dd>Rank by Mean Score (1 = best).</dd>
+    </dl>
+  </section>
+
+  <section class="card card-doc">
+  <h3>Repository Contents</h3>
+
+
+  <ul class="contents-list">
+    <li>
+      <strong>Scripts</strong> — scripts for running benchmarks and collecting metrics/logs.
+    </li>
+    <li>
+      <strong>Notebooks</strong> — Jupyter notebook for table generation and post-processing.
+    </li>
+    <li>
+      <strong>Results (CSV / Excel)</strong> — tables for each ranking category, overall rankings, and a master table with all scores and fields.
+    </li>
+    <li>
+      <strong>Raw JSON</strong> — per-run outputs from <code>lm-evaluation-harness</code>.
+    </li>
+    <li>
+      <strong>Stdout logs</strong> — per-run console logs.
+    </li>
+    <li>
+      <strong>GPU utilization logs</strong> — minute-level utilization log used to compute <em>GPU Util Time</em>.
+    </li>
+  </ul>
+
+  <div class="ref-note">
+    Benchmarks were referenced from
+    <a href="https://github.com/leobeeson/llm_benchmarks" target="_blank" rel="noopener">
+      leobeeson/llm_benchmarks
+    </a>.
+    Detailed descriptions of each benchmark can be found in this repository.
+  </div>
+</section>
+
+  <hr class="lb-divider">
+
+</section>
+
+<script>
+  // optional: stamp “last updated” from the page build time
+  const el = document.getElementById('last-updated');
+  if (el) el.textContent = new Date().toISOString().slice(0,10);
+</script>
+
+
+
+
   <div class="wrap">
     <h1 class="title">Overall Rank (Average Rank)</h1>
 
@@ -422,6 +741,943 @@
       <td>0.3799</td>
       <td>41</td>
     </tr>
+  </tbody>
+        </table>
+
+      </div>
+    </div>
+  </div>
+
+  <div class="wrap">
+    <h1 class="title">Reasoning & Math (Average Rank)</h1>
+
+    <div class="table-card">
+      <!-- 🔽 Paste your raw HTML table inside this div -->
+      <div id="table-slot">
+        <!-- Example: your pandas to_html dump goes here -->
+        <!-- <table> ... </table> -->
+        <table border="1" class="dataframe">
+  <thead>
+    <tr style="text-align: right;">
+      <th>Model Name</th>
+      <th>Total Time</th>
+      <th>GPU Util Time</th>
+      <th>Reasoning &amp; Math Mean Score</th>
+      <th>Reasoning &amp; Math Avg. Rank</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>google_gemma-3-12b-it</td>
+      <td>15h 45m</td>
+      <td>14h 8m</td>
+      <td>0.6266</td>
+      <td>1</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-8B</td>
+      <td>15h 31m</td>
+      <td>13h 44m</td>
+      <td>0.6214</td>
+      <td>2</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-14B (8bit)</td>
+      <td>29h 45m</td>
+      <td>17h 29m</td>
+      <td>0.5860</td>
+      <td>3</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-4B</td>
+      <td>5h 51m</td>
+      <td>5h 3m</td>
+      <td>0.5712</td>
+      <td>4</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct</td>
+      <td>9h 36m</td>
+      <td>8h 33m</td>
+      <td>0.5541</td>
+      <td>5</td>
+    </tr>
+    <tr>
+      <td>openchat_openchat-3.6-8b-20240522</td>
+      <td>7h 51m</td>
+      <td>6h 59m</td>
+      <td>0.5505</td>
+      <td>6</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-14B-Instruct (8bit)</td>
+      <td>52h 44m</td>
+      <td>29h 32m</td>
+      <td>0.5488</td>
+      <td>7</td>
+    </tr>
+    <tr>
+      <td>mistralai_Ministral-8B-Instruct-2410</td>
+      <td>10h 46m</td>
+      <td>9h 27m</td>
+      <td>0.5446</td>
+      <td>8</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-9B-Chat</td>
+      <td>13h 54m</td>
+      <td>12h 15m</td>
+      <td>0.5399</td>
+      <td>9</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-0528-Qwen3-8B</td>
+      <td>17h 57m</td>
+      <td>15h 30m</td>
+      <td>0.5387</td>
+      <td>10</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-4b-it</td>
+      <td>4h 51m</td>
+      <td>3h 50m</td>
+      <td>0.5374</td>
+      <td>11</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.1-8B-Instruct</td>
+      <td>12h 19m</td>
+      <td>10h 52m</td>
+      <td>0.5366</td>
+      <td>12</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Meta-Llama-3-8B-Instruct</td>
+      <td>6h 30m</td>
+      <td>5h 46m</td>
+      <td>0.5286</td>
+      <td>13</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2-7B-Instruct</td>
+      <td>11h 30m</td>
+      <td>10h 11m</td>
+      <td>0.5285</td>
+      <td>14</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct-1M</td>
+      <td>11h 17m</td>
+      <td>10h 10m</td>
+      <td>0.5245</td>
+      <td>15</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-9B</td>
+      <td>11h 43m</td>
+      <td>10h 26m</td>
+      <td>0.5206</td>
+      <td>16</td>
+    </tr>
+    <tr>
+      <td>NousResearch_Hermes-2-Pro-Mistral-7B</td>
+      <td>8h 27m</td>
+      <td>7h 28m</td>
+      <td>0.5184</td>
+      <td>17</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B</td>
+      <td>27h 21m</td>
+      <td>24h 38m</td>
+      <td>0.5010</td>
+      <td>18</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B-Chat</td>
+      <td>8h 4m</td>
+      <td>7h 1m</td>
+      <td>0.5006</td>
+      <td>19</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B-Instruct</td>
+      <td>5h 37m</td>
+      <td>4h 57m</td>
+      <td>0.4997</td>
+      <td>20</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-7B</td>
+      <td>6h 28m</td>
+      <td>5h 43m</td>
+      <td>0.4841</td>
+      <td>21</td>
+    </tr>
+    <tr>
+      <td>mistralai_Mistral-7B-Instruct-v0.3</td>
+      <td>8h 38m</td>
+      <td>7h 41m</td>
+      <td>0.4704</td>
+      <td>22</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-3B-Instruct</td>
+      <td>7h 12m</td>
+      <td>5h 57m</td>
+      <td>0.4688</td>
+      <td>23</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B</td>
+      <td>4h 28m</td>
+      <td>3h 54m</td>
+      <td>0.4495</td>
+      <td>24</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-1.7B</td>
+      <td>4h 25m</td>
+      <td>3h 36m</td>
+      <td>0.4493</td>
+      <td>25</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Llama-8B</td>
+      <td>11h 46m</td>
+      <td>10h 36m</td>
+      <td>0.4469</td>
+      <td>26</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-chat</td>
+      <td>10h 6m</td>
+      <td>9h 8m</td>
+      <td>0.4244</td>
+      <td>27</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-chat-hf</td>
+      <td>17h 8m</td>
+      <td>15h 37m</td>
+      <td>0.4143</td>
+      <td>28</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-1.5B-Instruct</td>
+      <td>3h 25m</td>
+      <td>2h 39m</td>
+      <td>0.4085</td>
+      <td>29</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-1.5B</td>
+      <td>3h 40m</td>
+      <td>2h 52m</td>
+      <td>0.4009</td>
+      <td>30</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-1.5B-Instruct</td>
+      <td>3h 20m</td>
+      <td>2h 36m</td>
+      <td>0.3874</td>
+      <td>31</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-3B-Instruct</td>
+      <td>7h 48m</td>
+      <td>6h 30m</td>
+      <td>0.3823</td>
+      <td>32</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-hf</td>
+      <td>19h 21m</td>
+      <td>17h 38m</td>
+      <td>0.3719</td>
+      <td>33</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-math-7b-rl</td>
+      <td>8h 2m</td>
+      <td>7h 12m</td>
+      <td>0.3702</td>
+      <td>34</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-chat-hf</td>
+      <td>6h 57m</td>
+      <td>6h 7m</td>
+      <td>0.3674</td>
+      <td>35</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-0.6B</td>
+      <td>3h 45m</td>
+      <td>2h 53m</td>
+      <td>0.3494</td>
+      <td>36</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-1B-Instruct</td>
+      <td>3h 30m</td>
+      <td>2h 35m</td>
+      <td>0.3450</td>
+      <td>37</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-base</td>
+      <td>7h 11m</td>
+      <td>6h 26m</td>
+      <td>0.3377</td>
+      <td>38</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-hf</td>
+      <td>5h 42m</td>
+      <td>4h 59m</td>
+      <td>0.3361</td>
+      <td>39</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-1b-it</td>
+      <td>6h 50m</td>
+      <td>4h 52m</td>
+      <td>0.3312</td>
+      <td>40</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-0.5B-Instruct</td>
+      <td>2h 34m</td>
+      <td>1h 48m</td>
+      <td>0.2914</td>
+      <td>41</td>
+    </tr>
+  </tbody>
+</table>
+
+      </div>
+    </div>
+  </div>
+  <div class="wrap">
+    <h1 class="title">Commonsense & NLI (Average Rank)</h1>
+
+    <div class="table-card">
+      <!-- 🔽 Paste your raw HTML table inside this div -->
+      <div id="table-slot">
+        <!-- Example: your pandas to_html dump goes here -->
+        <!-- <table> ... </table> -->
+        <table border="1" class="dataframe">
+  <thead>
+    <tr style="text-align: right;">
+      <th>Model Name</th>
+      <th>Total Time</th>
+      <th>GPU Util Time</th>
+      <th>Commonsense &amp; NLI Mean Score</th>
+      <th>Commonsense &amp; NLI Avg. Rank</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>Qwen_Qwen2.5-14B-Instruct (8bit)</td>
+      <td>52h 44m</td>
+      <td>29h 32m</td>
+      <td>0.7941</td>
+      <td>1</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-14B (8bit)</td>
+      <td>29h 45m</td>
+      <td>17h 29m</td>
+      <td>0.7807</td>
+      <td>2</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-12b-it</td>
+      <td>15h 45m</td>
+      <td>14h 8m</td>
+      <td>0.7737</td>
+      <td>3</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct</td>
+      <td>9h 36m</td>
+      <td>8h 33m</td>
+      <td>0.7730</td>
+      <td>4</td>
+    </tr>
+    <tr>
+      <td>openchat_openchat-3.6-8b-20240522</td>
+      <td>7h 51m</td>
+      <td>6h 59m</td>
+      <td>0.7726</td>
+      <td>5</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-9B-Chat</td>
+      <td>13h 54m</td>
+      <td>12h 15m</td>
+      <td>0.7691</td>
+      <td>6</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct-1M</td>
+      <td>11h 17m</td>
+      <td>10h 10m</td>
+      <td>0.7564</td>
+      <td>7</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-8B</td>
+      <td>15h 31m</td>
+      <td>13h 44m</td>
+      <td>0.7468</td>
+      <td>8</td>
+    </tr>
+    <tr>
+      <td>mistralai_Mistral-7B-Instruct-v0.3</td>
+      <td>8h 38m</td>
+      <td>7h 41m</td>
+      <td>0.7403</td>
+      <td>9</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B-Chat</td>
+      <td>8h 4m</td>
+      <td>7h 1m</td>
+      <td>0.7374</td>
+      <td>10</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-3B-Instruct</td>
+      <td>7h 48m</td>
+      <td>6h 30m</td>
+      <td>0.7367</td>
+      <td>11</td>
+    </tr>
+    <tr>
+      <td>mistralai_Ministral-8B-Instruct-2410</td>
+      <td>10h 46m</td>
+      <td>9h 27m</td>
+      <td>0.7328</td>
+      <td>12</td>
+    </tr>
+    <tr>
+      <td>NousResearch_Hermes-2-Pro-Mistral-7B</td>
+      <td>8h 27m</td>
+      <td>7h 28m</td>
+      <td>0.7284</td>
+      <td>13</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2-7B-Instruct</td>
+      <td>11h 30m</td>
+      <td>10h 11m</td>
+      <td>0.7274</td>
+      <td>14</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-9B</td>
+      <td>11h 43m</td>
+      <td>10h 26m</td>
+      <td>0.7266</td>
+      <td>15</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-4B</td>
+      <td>5h 51m</td>
+      <td>5h 3m</td>
+      <td>0.7266</td>
+      <td>16</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.1-8B-Instruct</td>
+      <td>12h 19m</td>
+      <td>10h 52m</td>
+      <td>0.7249</td>
+      <td>17</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B</td>
+      <td>4h 28m</td>
+      <td>3h 54m</td>
+      <td>0.7199</td>
+      <td>18</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-4b-it</td>
+      <td>4h 51m</td>
+      <td>3h 50m</td>
+      <td>0.7167</td>
+      <td>19</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-hf</td>
+      <td>19h 21m</td>
+      <td>17h 38m</td>
+      <td>0.7157</td>
+      <td>20</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-chat-hf</td>
+      <td>17h 8m</td>
+      <td>15h 37m</td>
+      <td>0.7153</td>
+      <td>21</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Meta-Llama-3-8B-Instruct</td>
+      <td>6h 30m</td>
+      <td>5h 46m</td>
+      <td>0.7147</td>
+      <td>22</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-0528-Qwen3-8B</td>
+      <td>17h 57m</td>
+      <td>15h 30m</td>
+      <td>0.7094</td>
+      <td>23</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-chat</td>
+      <td>10h 6m</td>
+      <td>9h 8m</td>
+      <td>0.7090</td>
+      <td>24</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-chat-hf</td>
+      <td>6h 57m</td>
+      <td>6h 7m</td>
+      <td>0.6978</td>
+      <td>25</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-hf</td>
+      <td>5h 42m</td>
+      <td>4h 59m</td>
+      <td>0.6956</td>
+      <td>26</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Llama-8B</td>
+      <td>11h 46m</td>
+      <td>10h 36m</td>
+      <td>0.6928</td>
+      <td>27</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-base</td>
+      <td>7h 11m</td>
+      <td>6h 26m</td>
+      <td>0.6886</td>
+      <td>28</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-1.5B-Instruct</td>
+      <td>3h 20m</td>
+      <td>2h 36m</td>
+      <td>0.6803</td>
+      <td>29</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-3B-Instruct</td>
+      <td>7h 12m</td>
+      <td>5h 57m</td>
+      <td>0.6788</td>
+      <td>30</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-math-7b-rl</td>
+      <td>8h 2m</td>
+      <td>7h 12m</td>
+      <td>0.6711</td>
+      <td>31</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B</td>
+      <td>27h 21m</td>
+      <td>24h 38m</td>
+      <td>0.6587</td>
+      <td>32</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-1.7B</td>
+      <td>4h 25m</td>
+      <td>3h 36m</td>
+      <td>0.6442</td>
+      <td>33</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-7B</td>
+      <td>6h 28m</td>
+      <td>5h 43m</td>
+      <td>0.6422</td>
+      <td>34</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-1b-it</td>
+      <td>6h 50m</td>
+      <td>4h 52m</td>
+      <td>0.6267</td>
+      <td>35</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-1B-Instruct</td>
+      <td>3h 30m</td>
+      <td>2h 35m</td>
+      <td>0.6264</td>
+      <td>36</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B-Instruct</td>
+      <td>5h 37m</td>
+      <td>4h 57m</td>
+      <td>0.6184</td>
+      <td>37</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-0.5B-Instruct</td>
+      <td>2h 34m</td>
+      <td>1h 48m</td>
+      <td>0.6039</td>
+      <td>38</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-1.5B</td>
+      <td>3h 40m</td>
+      <td>2h 52m</td>
+      <td>0.5703</td>
+      <td>39</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-0.6B</td>
+      <td>3h 45m</td>
+      <td>2h 53m</td>
+      <td>0.5696</td>
+      <td>40</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-1.5B-Instruct</td>
+      <td>3h 25m</td>
+      <td>2h 39m</td>
+      <td>0.5181</td>
+      <td>41</td>
+    </tr>
+  </tbody>
+</table>
+
+      </div>
+    </div>
+  </div>
+  <div class="wrap">
+    <h1 class="title">Knowledge & Reading (Average Rank)</h1>
+
+    <div class="table-card">
+      <!-- 🔽 Paste your raw HTML table inside this div -->
+      <div id="table-slot">
+        <!-- Example: your pandas to_html dump goes here -->
+        <!-- <table> ... </table> -->
+       <table border="1" class="dataframe">
+  <thead>
+    <tr style="text-align: right;">
+      <th>Model Name</th>
+      <th>Total Time</th>
+      <th>GPU Util Time</th>
+      <th>Knowledge &amp; Reading Mean Score</th>
+      <th>Knowledge &amp; Reading Avg. Rank</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>01-ai_Yi-1.5-9B</td>
+      <td>11h 43m</td>
+      <td>10h 26m</td>
+      <td>0.4369</td>
+      <td>1</td>
+    </tr>
+    <tr>
+      <td>openchat_openchat-3.6-8b-20240522</td>
+      <td>7h 51m</td>
+      <td>6h 59m</td>
+      <td>0.4136</td>
+      <td>2</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.1-8B-Instruct</td>
+      <td>12h 19m</td>
+      <td>10h 52m</td>
+      <td>0.4127</td>
+      <td>3</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B</td>
+      <td>4h 28m</td>
+      <td>3h 54m</td>
+      <td>0.4063</td>
+      <td>4</td>
+    </tr>
+    <tr>
+      <td>mistralai_Mistral-7B-Instruct-v0.3</td>
+      <td>8h 38m</td>
+      <td>7h 41m</td>
+      <td>0.4045</td>
+      <td>5</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct-1M</td>
+      <td>11h 17m</td>
+      <td>10h 10m</td>
+      <td>0.3963</td>
+      <td>6</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-14B (8bit)</td>
+      <td>29h 45m</td>
+      <td>17h 29m</td>
+      <td>0.3926</td>
+      <td>7</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Meta-Llama-3-8B-Instruct</td>
+      <td>6h 30m</td>
+      <td>5h 46m</td>
+      <td>0.3923</td>
+      <td>8</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-7B-Instruct</td>
+      <td>9h 36m</td>
+      <td>8h 33m</td>
+      <td>0.3810</td>
+      <td>9</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-12b-it</td>
+      <td>15h 45m</td>
+      <td>14h 8m</td>
+      <td>0.3791</td>
+      <td>10</td>
+    </tr>
+    <tr>
+      <td>NousResearch_Hermes-2-Pro-Mistral-7B</td>
+      <td>8h 27m</td>
+      <td>7h 28m</td>
+      <td>0.3719</td>
+      <td>11</td>
+    </tr>
+    <tr>
+      <td>mistralai_Ministral-8B-Instruct-2410</td>
+      <td>10h 46m</td>
+      <td>9h 27m</td>
+      <td>0.3683</td>
+      <td>12</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-14B-Instruct (8bit)</td>
+      <td>52h 44m</td>
+      <td>29h 32m</td>
+      <td>0.3581</td>
+      <td>13</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-8B</td>
+      <td>15h 31m</td>
+      <td>13h 44m</td>
+      <td>0.3566</td>
+      <td>14</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-9B-Chat</td>
+      <td>13h 54m</td>
+      <td>12h 15m</td>
+      <td>0.3467</td>
+      <td>15</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-3B-Instruct</td>
+      <td>7h 12m</td>
+      <td>5h 57m</td>
+      <td>0.3438</td>
+      <td>16</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-3B-Instruct</td>
+      <td>7h 48m</td>
+      <td>6h 30m</td>
+      <td>0.3406</td>
+      <td>17</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-hf</td>
+      <td>19h 21m</td>
+      <td>17h 38m</td>
+      <td>0.3374</td>
+      <td>18</td>
+    </tr>
+    <tr>
+      <td>01-ai_Yi-1.5-6B-Chat</td>
+      <td>8h 4m</td>
+      <td>7h 1m</td>
+      <td>0.3339</td>
+      <td>19</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-4b-it</td>
+      <td>4h 51m</td>
+      <td>3h 50m</td>
+      <td>0.3261</td>
+      <td>20</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-4B</td>
+      <td>5h 51m</td>
+      <td>5h 3m</td>
+      <td>0.3226</td>
+      <td>21</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-hf</td>
+      <td>5h 42m</td>
+      <td>4h 59m</td>
+      <td>0.3018</td>
+      <td>22</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-chat</td>
+      <td>10h 6m</td>
+      <td>9h 8m</td>
+      <td>0.3007</td>
+      <td>23</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2-7B-Instruct</td>
+      <td>11h 30m</td>
+      <td>10h 11m</td>
+      <td>0.2919</td>
+      <td>24</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-1.5B-Instruct</td>
+      <td>3h 20m</td>
+      <td>2h 36m</td>
+      <td>0.2903</td>
+      <td>25</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-13b-chat-hf</td>
+      <td>17h 8m</td>
+      <td>15h 37m</td>
+      <td>0.2864</td>
+      <td>26</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-llm-7b-base</td>
+      <td>7h 11m</td>
+      <td>6h 26m</td>
+      <td>0.2864</td>
+      <td>27</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-0528-Qwen3-8B</td>
+      <td>17h 57m</td>
+      <td>15h 30m</td>
+      <td>0.2834</td>
+      <td>28</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B</td>
+      <td>27h 21m</td>
+      <td>24h 38m</td>
+      <td>0.2827</td>
+      <td>29</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Llama-8B</td>
+      <td>11h 46m</td>
+      <td>10h 36m</td>
+      <td>0.2805</td>
+      <td>30</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-3.2-1B-Instruct</td>
+      <td>3h 30m</td>
+      <td>2h 35m</td>
+      <td>0.2731</td>
+      <td>31</td>
+    </tr>
+    <tr>
+      <td>meta-llama_Llama-2-7b-chat-hf</td>
+      <td>6h 57m</td>
+      <td>6h 7m</td>
+      <td>0.2656</td>
+      <td>32</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_deepseek-math-7b-rl</td>
+      <td>8h 2m</td>
+      <td>7h 12m</td>
+      <td>0.2581</td>
+      <td>33</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-1.7B</td>
+      <td>4h 25m</td>
+      <td>3h 36m</td>
+      <td>0.2567</td>
+      <td>34</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-7B</td>
+      <td>6h 28m</td>
+      <td>5h 43m</td>
+      <td>0.2340</td>
+      <td>35</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-7B-Instruct</td>
+      <td>5h 37m</td>
+      <td>4h 57m</td>
+      <td>0.2276</td>
+      <td>36</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-0.5B-Instruct</td>
+      <td>2h 34m</td>
+      <td>1h 48m</td>
+      <td>0.2218</td>
+      <td>37</td>
+    </tr>
+    <tr>
+      <td>google_gemma-3-1b-it</td>
+      <td>6h 50m</td>
+      <td>4h 52m</td>
+      <td>0.2202</td>
+      <td>38</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen3-0.6B</td>
+      <td>3h 45m</td>
+      <td>2h 53m</td>
+      <td>0.2000</td>
+      <td>39</td>
+    </tr>
+    <tr>
+      <td>Qwen_Qwen2.5-Math-1.5B-Instruct</td>
+      <td>3h 25m</td>
+      <td>2h 39m</td>
+      <td>0.1983</td>
+      <td>40</td>
+    </tr>
+    <tr>
+      <td>deepseek-ai_DeepSeek-R1-Distill-Qwen-1.5B</td>
+      <td>3h 40m</td>
+      <td>2h 52m</td>
+      <td>0.1954</td>
+      <td>41</td>
+    </tr>
   </tbody>
 </table>
 

style.css

@@ -1,28 +0,0 @@
-body {
-	padding: 2rem;
-	font-family: -apple-system, BlinkMacSystemFont, "Arial", sans-serif;
-}
-
-h1 {
-	font-size: 16px;
-	margin-top: 0;
-}
-
-p {
-	color: rgb(107, 114, 128);
-	font-size: 15px;
-	margin-bottom: 10px;
-	margin-top: 5px;
-}
-
-.card {
-	max-width: 620px;
-	margin: 0 auto;
-	padding: 16px;
-	border: 1px solid lightgray;
-	border-radius: 16px;
-}
-
-.card p:last-child {
-	margin-bottom: 0;
-}