Update README.md
Browse files
README.md
CHANGED
|
@@ -213,5 +213,106 @@ license: apache-2.0
|
|
| 213 |
</table>
|
| 214 |
|
| 215 |
|
| 216 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 217 |
|
|
|
|
| 213 |
</table>
|
| 214 |
|
| 215 |
|
| 216 |
+
### Accuracy
|
| 217 |
+
<table>
|
| 218 |
+
<thead>
|
| 219 |
+
<tr>
|
| 220 |
+
<th>Category</th>
|
| 221 |
+
<th>Metric</th>
|
| 222 |
+
<th>ibm-granite/granite-4.0-h-small</th>
|
| 223 |
+
<th>RedHatAI/granite-4.0-h-small-FP8-dynamic</th>
|
| 224 |
+
<th>Recovery (%)</th>
|
| 225 |
+
</tr>
|
| 226 |
+
</thead>
|
| 227 |
+
<tbody>
|
| 228 |
+
<!-- OpenLLM Leaderboard V1 -->
|
| 229 |
+
<tr>
|
| 230 |
+
<td rowspan="7"><b>OpenLLM V1</b></td>
|
| 231 |
+
<td>ARC-Challenge (Acc-Norm, 25-shot)</td>
|
| 232 |
+
<td>72.27</td>
|
| 233 |
+
<td>72.10</td>
|
| 234 |
+
<td>99.76</td>
|
| 235 |
+
</tr>
|
| 236 |
+
<tr>
|
| 237 |
+
<td>GSM8K (Strict-Match, 5-shot)</td>
|
| 238 |
+
<td>85.22</td>
|
| 239 |
+
<td>84.84</td>
|
| 240 |
+
<td>99.56</td>
|
| 241 |
+
</tr>
|
| 242 |
+
<tr>
|
| 243 |
+
<td>HellaSwag (Acc-Norm, 10-shot)</td>
|
| 244 |
+
<td>86.08</td>
|
| 245 |
+
<td>85.88</td>
|
| 246 |
+
<td>99.77</td>
|
| 247 |
+
</tr>
|
| 248 |
+
<tr>
|
| 249 |
+
<td>MMLU (Acc, 5-shot)</td>
|
| 250 |
+
<td>77.15</td>
|
| 251 |
+
<td>77.18</td>
|
| 252 |
+
<td>100.03</td>
|
| 253 |
+
</tr>
|
| 254 |
+
<tr>
|
| 255 |
+
<td>TruthfulQA (MC2, 0-shot)</td>
|
| 256 |
+
<td>57.64</td>
|
| 257 |
+
<td>57.63</td>
|
| 258 |
+
<td>100.00</td>
|
| 259 |
+
</tr>
|
| 260 |
+
<tr>
|
| 261 |
+
<td>Winogrande (Acc, 5-shot)</td>
|
| 262 |
+
<td>81.37</td>
|
| 263 |
+
<td>81.45</td>
|
| 264 |
+
<td>100.10</td>
|
| 265 |
+
</tr>
|
| 266 |
+
<tr>
|
| 267 |
+
<td><b>Average Score</b></td>
|
| 268 |
+
<td><b>76.62</b></td>
|
| 269 |
+
<td><b>76.51</b></td>
|
| 270 |
+
<td><b>99.86</b></td>
|
| 271 |
+
</tr>
|
| 272 |
+
<!-- OpenLLM Leaderboard V2 -->
|
| 273 |
+
<tr>
|
| 274 |
+
<td rowspan="7"><b>OpenLLM V2</b></td>
|
| 275 |
+
<td>IFEval (Inst Level Strict Acc, 0-shot)</td>
|
| 276 |
+
<td>87.53</td>
|
| 277 |
+
<td>87.41</td>
|
| 278 |
+
<td>99.86</td>
|
| 279 |
+
</tr>
|
| 280 |
+
<tr>
|
| 281 |
+
<td>BBH (Acc-Norm, 3-shot)</td>
|
| 282 |
+
<td>61.52</td>
|
| 283 |
+
<td>61.19</td>
|
| 284 |
+
<td>99.46</td>
|
| 285 |
+
</tr>
|
| 286 |
+
<tr>
|
| 287 |
+
<td>Math-Hard (Exact-Match, 4-shot)</td>
|
| 288 |
+
<td>46.22</td>
|
| 289 |
+
<td>41.77</td>
|
| 290 |
+
<td>90.36</td>
|
| 291 |
+
</tr>
|
| 292 |
+
<tr>
|
| 293 |
+
<td>GPQA (Acc-Norm, 0-shot)</td>
|
| 294 |
+
<td>35.23</td>
|
| 295 |
+
<td>34.23</td>
|
| 296 |
+
<td>97.14</td>
|
| 297 |
+
</tr>
|
| 298 |
+
<tr>
|
| 299 |
+
<td>MUSR (Acc-Norm, 0-shot)</td>
|
| 300 |
+
<td>46.69</td>
|
| 301 |
+
<td>45.77</td>
|
| 302 |
+
<td>98.02</td>
|
| 303 |
+
</tr>
|
| 304 |
+
<tr>
|
| 305 |
+
<td>MMLU-Pro (Acc, 5-shot)</td>
|
| 306 |
+
<td>47.99</td>
|
| 307 |
+
<td>47.58</td>
|
| 308 |
+
<td>99.15</td>
|
| 309 |
+
</tr>
|
| 310 |
+
<tr>
|
| 311 |
+
<td><b>Average Score</b></td>
|
| 312 |
+
<td><b>54.20</b></td>
|
| 313 |
+
<td><b>52.99</b></td>
|
| 314 |
+
<td><b>97.77</b></td>
|
| 315 |
+
</tr>
|
| 316 |
+
</tbody>
|
| 317 |
+
</table>
|
| 318 |
|