Myric commited on
Commit
058729c
Β·
verified Β·
1 Parent(s): d69035a

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +28 -7
README.md CHANGED
@@ -196,7 +196,10 @@ spans 2.5–14.5 h while the 12 B-active 120–124 B MoEs finish in 1.8–3.4 h.
196
  <details>
197
  <summary><strong>Wall clock vs active parameters</strong> β€” tap to expand</summary>
198
 
199
- <img src="plots/wall_vs_active.png" alt="Wall clock vs active parameters" style="max-width:100%; height:auto;" />
 
 
 
200
 
201
  </details>
202
 
@@ -212,7 +215,10 @@ fill) and a gray edge labelled **Abliterated**, because its Heretic abliteration
212
  <details>
213
  <summary><strong>Wall clock vs active parameters, perfect-score runs by country</strong> β€” tap to expand</summary>
214
 
215
- <img src="plots/wall_vs_active_perfect_country.png" alt="Wall clock vs active parameters, perfect-score runs by country" style="max-width:100%; height:auto;" />
 
 
 
216
 
217
  </details>
218
 
@@ -222,7 +228,10 @@ finish in 1.75 h while a 4 B model takes 14.5 h.
222
  <details>
223
  <summary><strong>Wall clock vs total parameters</strong> β€” tap to expand</summary>
224
 
225
- <img src="plots/wall_vs_total.png" alt="Wall clock vs total parameters" style="max-width:100%; height:auto;" />
 
 
 
226
 
227
  </details>
228
 
@@ -232,7 +241,10 @@ omitted); wall time still includes voided tasks.
232
  <details>
233
  <summary><strong>Wall clock vs total parameters, perfect-score runs by country</strong> β€” tap to expand</summary>
234
 
235
- <img src="plots/wall_vs_total_perfect_country.png" alt="Wall clock vs total parameters, perfect-score runs by country" style="max-width:100%; height:auto;" />
 
 
 
236
 
237
  </details>
238
 
@@ -242,7 +254,10 @@ tokens take longer. The Spark outliers are the high-token runs, not the large mo
242
  <details>
243
  <summary><strong>Wall clock vs output tokens</strong> β€” tap to expand</summary>
244
 
245
- <img src="plots/wall_vs_tokens.png" alt="Wall clock vs output tokens" style="max-width:100%; height:auto;" />
 
 
 
246
 
247
  </details>
248
 
@@ -253,7 +268,10 @@ outlier at ~18 k tokens/task.
253
  <details>
254
  <summary><strong>Verbosity vs active parameters</strong> β€” tap to expand</summary>
255
 
256
- <img src="plots/tokens_per_task_vs_active.png" alt="Verbosity vs active parameters" style="max-width:100%; height:auto;" />
 
 
 
257
 
258
  </details>
259
 
@@ -266,7 +284,10 @@ raw decode speed from the custom ternary kernels.
266
  <details>
267
  <summary><strong>Performance per byte</strong> β€” tap to expand</summary>
268
 
269
- <img src="plots/perf_per_byte.png" alt="Performance per byte" style="max-width:100%; height:auto;" />
 
 
 
270
 
271
  </details>
272
 
 
196
  <details>
197
  <summary><strong>Wall clock vs active parameters</strong> β€” tap to expand</summary>
198
 
199
+ <a href="plots/wall_vs_active.png" target="_blank">
200
+ <img src="plots/wall_vs_active.png" alt="Wall clock vs active parameters" style="width:100%; height:auto; max-width:100%;" />
201
+ </a>
202
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
203
 
204
  </details>
205
 
 
215
  <details>
216
  <summary><strong>Wall clock vs active parameters, perfect-score runs by country</strong> β€” tap to expand</summary>
217
 
218
+ <a href="plots/wall_vs_active_perfect_country.png" target="_blank">
219
+ <img src="plots/wall_vs_active_perfect_country.png" alt="Wall clock vs active parameters, perfect-score runs by country" style="width:100%; height:auto; max-width:100%;" />
220
+ </a>
221
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
222
 
223
  </details>
224
 
 
228
  <details>
229
  <summary><strong>Wall clock vs total parameters</strong> β€” tap to expand</summary>
230
 
231
+ <a href="plots/wall_vs_total.png" target="_blank">
232
+ <img src="plots/wall_vs_total.png" alt="Wall clock vs total parameters" style="width:100%; height:auto; max-width:100%;" />
233
+ </a>
234
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
235
 
236
  </details>
237
 
 
241
  <details>
242
  <summary><strong>Wall clock vs total parameters, perfect-score runs by country</strong> β€” tap to expand</summary>
243
 
244
+ <a href="plots/wall_vs_total_perfect_country.png" target="_blank">
245
+ <img src="plots/wall_vs_total_perfect_country.png" alt="Wall clock vs total parameters, perfect-score runs by country" style="width:100%; height:auto; max-width:100%;" />
246
+ </a>
247
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
248
 
249
  </details>
250
 
 
254
  <details>
255
  <summary><strong>Wall clock vs output tokens</strong> β€” tap to expand</summary>
256
 
257
+ <a href="plots/wall_vs_tokens.png" target="_blank">
258
+ <img src="plots/wall_vs_tokens.png" alt="Wall clock vs output tokens" style="width:100%; height:auto; max-width:100%;" />
259
+ </a>
260
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
261
 
262
  </details>
263
 
 
268
  <details>
269
  <summary><strong>Verbosity vs active parameters</strong> β€” tap to expand</summary>
270
 
271
+ <a href="plots/tokens_per_task_vs_active.png" target="_blank">
272
+ <img src="plots/tokens_per_task_vs_active.png" alt="Verbosity vs active parameters" style="width:100%; height:auto; max-width:100%;" />
273
+ </a>
274
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
275
 
276
  </details>
277
 
 
284
  <details>
285
  <summary><strong>Performance per byte</strong> β€” tap to expand</summary>
286
 
287
+ <a href="plots/perf_per_byte.png" target="_blank">
288
+ <img src="plots/perf_per_byte.png" alt="Performance per byte" style="width:100%; height:auto; max-width:100%;" />
289
+ </a>
290
+ <p><em>Tap chart to open full-resolution β€” pinch-zoom works in the image viewer.</em></p>
291
 
292
  </details>
293