Commit
•
677a2de
1
Parent(s):
c8d5bc8
Update README.md (#6)
Browse files- Update README.md (44e53a1bce02d9da2e8373f49ad10b711826e9a5)
Co-authored-by: William de Vazelhes <[email protected]>
README.md
CHANGED
@@ -149,7 +149,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
149 |
<th>Llama-3.2-1B</th>
|
150 |
<th>Qwen2.5-1.5B</th>
|
151 |
<th>SmolLM2-1.7B</th>
|
152 |
-
<th>gemma-2-2b</th>
|
153 |
<th>Falcon3-1B-Base</th>
|
154 |
</tr>
|
155 |
</thead>
|
@@ -160,7 +159,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
160 |
<td>31.1</td>
|
161 |
<td>61.0</td>
|
162 |
<td>50.1</td>
|
163 |
-
<td>53.0</td>
|
164 |
<td>42.5</td>
|
165 |
</tr>
|
166 |
<tr>
|
@@ -168,7 +166,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
168 |
<td>11.7</td>
|
169 |
<td>28.4</td>
|
170 |
<td>21.3</td>
|
171 |
-
<td>22.1</td>
|
172 |
<td>16.1</td>
|
173 |
</tr>
|
174 |
<tr>
|
@@ -176,7 +173,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
176 |
<td>14.8</td>
|
177 |
<td>26.0</td>
|
178 |
<td>24.2</td>
|
179 |
-
<td>20.3</td>
|
180 |
<td>25.2</td>
|
181 |
</tr>
|
182 |
<tr>
|
@@ -185,7 +181,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
185 |
<td>6.6</td>
|
186 |
<td>62.2</td>
|
187 |
<td>31.0</td>
|
188 |
-
<td>25.5</td>
|
189 |
<td>34.3</td>
|
190 |
</tr>
|
191 |
<tr>
|
@@ -193,7 +188,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
193 |
<td>0.2</td>
|
194 |
<td>6.7</td>
|
195 |
<td>1.4</td>
|
196 |
-
<td>2.6</td>
|
197 |
<td>2.2</td>
|
198 |
</tr>
|
199 |
<tr>
|
@@ -202,7 +196,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
202 |
<td>40.2</td>
|
203 |
<td>54.8</td>
|
204 |
<td>54.1</td>
|
205 |
-
<td>53.7</td>
|
206 |
<td>48.1</td>
|
207 |
</tr>
|
208 |
<tr>
|
@@ -210,7 +203,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
210 |
<td>24.2</td>
|
211 |
<td>28.1</td>
|
212 |
<td>28.9</td>
|
213 |
-
<td>25.5</td>
|
214 |
<td>28.1</td>
|
215 |
</tr>
|
216 |
<tr>
|
@@ -218,7 +210,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
218 |
<td>34.5</td>
|
219 |
<td>35.5</td>
|
220 |
<td>34.7</td>
|
221 |
-
<td>42.7</td>
|
222 |
<td>41.9</td>
|
223 |
</tr>
|
224 |
<tr>
|
@@ -226,7 +217,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
226 |
<td>31.2</td>
|
227 |
<td>41.1</td>
|
228 |
<td>34.2</td>
|
229 |
-
<td>36.8</td>
|
230 |
<td>36.0</td>
|
231 |
</tr>
|
232 |
<tr>
|
@@ -235,7 +225,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
235 |
<td>74.5</td>
|
236 |
<td>76.0</td>
|
237 |
<td>77.5</td>
|
238 |
-
<td>79.2</td>
|
239 |
<td>74.5</td>
|
240 |
</tr>
|
241 |
<tr>
|
@@ -243,7 +232,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
243 |
<td>88.5</td>
|
244 |
<td>93.1</td>
|
245 |
<td>90.8</td>
|
246 |
-
<td>95.7</td>
|
247 |
<td>91.1</td>
|
248 |
</tr>
|
249 |
<tr>
|
@@ -251,7 +239,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
251 |
<td>60.4</td>
|
252 |
<td>63.0</td>
|
253 |
<td>66.1</td>
|
254 |
-
<td>68.6</td>
|
255 |
<td>61.2</td>
|
256 |
</tr>
|
257 |
<tr>
|
@@ -259,7 +246,6 @@ Falcon3-7B is trained on 256 H100 nodes (world size 2048).
|
|
259 |
<td>37.4</td>
|
260 |
<td>40.4</td>
|
261 |
<td>44.0</td>
|
262 |
-
<td>41.8</td>
|
263 |
<td>41.0</td>
|
264 |
</tr>
|
265 |
</tbody>
|
|
|
149 |
<th>Llama-3.2-1B</th>
|
150 |
<th>Qwen2.5-1.5B</th>
|
151 |
<th>SmolLM2-1.7B</th>
|
|
|
152 |
<th>Falcon3-1B-Base</th>
|
153 |
</tr>
|
154 |
</thead>
|
|
|
159 |
<td>31.1</td>
|
160 |
<td>61.0</td>
|
161 |
<td>50.1</td>
|
|
|
162 |
<td>42.5</td>
|
163 |
</tr>
|
164 |
<tr>
|
|
|
166 |
<td>11.7</td>
|
167 |
<td>28.4</td>
|
168 |
<td>21.3</td>
|
|
|
169 |
<td>16.1</td>
|
170 |
</tr>
|
171 |
<tr>
|
|
|
173 |
<td>14.8</td>
|
174 |
<td>26.0</td>
|
175 |
<td>24.2</td>
|
|
|
176 |
<td>25.2</td>
|
177 |
</tr>
|
178 |
<tr>
|
|
|
181 |
<td>6.6</td>
|
182 |
<td>62.2</td>
|
183 |
<td>31.0</td>
|
|
|
184 |
<td>34.3</td>
|
185 |
</tr>
|
186 |
<tr>
|
|
|
188 |
<td>0.2</td>
|
189 |
<td>6.7</td>
|
190 |
<td>1.4</td>
|
|
|
191 |
<td>2.2</td>
|
192 |
</tr>
|
193 |
<tr>
|
|
|
196 |
<td>40.2</td>
|
197 |
<td>54.8</td>
|
198 |
<td>54.1</td>
|
|
|
199 |
<td>48.1</td>
|
200 |
</tr>
|
201 |
<tr>
|
|
|
203 |
<td>24.2</td>
|
204 |
<td>28.1</td>
|
205 |
<td>28.9</td>
|
|
|
206 |
<td>28.1</td>
|
207 |
</tr>
|
208 |
<tr>
|
|
|
210 |
<td>34.5</td>
|
211 |
<td>35.5</td>
|
212 |
<td>34.7</td>
|
|
|
213 |
<td>41.9</td>
|
214 |
</tr>
|
215 |
<tr>
|
|
|
217 |
<td>31.2</td>
|
218 |
<td>41.1</td>
|
219 |
<td>34.2</td>
|
|
|
220 |
<td>36.0</td>
|
221 |
</tr>
|
222 |
<tr>
|
|
|
225 |
<td>74.5</td>
|
226 |
<td>76.0</td>
|
227 |
<td>77.5</td>
|
|
|
228 |
<td>74.5</td>
|
229 |
</tr>
|
230 |
<tr>
|
|
|
232 |
<td>88.5</td>
|
233 |
<td>93.1</td>
|
234 |
<td>90.8</td>
|
|
|
235 |
<td>91.1</td>
|
236 |
</tr>
|
237 |
<tr>
|
|
|
239 |
<td>60.4</td>
|
240 |
<td>63.0</td>
|
241 |
<td>66.1</td>
|
|
|
242 |
<td>61.2</td>
|
243 |
</tr>
|
244 |
<tr>
|
|
|
246 |
<td>37.4</td>
|
247 |
<td>40.4</td>
|
248 |
<td>44.0</td>
|
|
|
249 |
<td>41.0</td>
|
250 |
</tr>
|
251 |
</tbody>
|