Spaces:
Running on Zero
Running on Zero
Ledger tab: link the public experiment-log bucket
Browse files- index.html +4 -0
index.html
CHANGED
|
@@ -313,6 +313,10 @@
|
|
| 313 |
<b>Relative CER reduction</b>: how much of that damage the model repaired.
|
| 314 |
<b>Over‑correction</b>: how much text that was already right the model needlessly changed.
|
| 315 |
<b>Fix rate</b>: how much of what was actually wrong it fixed.</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 316 |
</div>
|
| 317 |
<div id="results-table"><p style="text-align:center;font-style:italic">Fetching the ledger…</p></div>
|
| 318 |
<div class="footnotes" id="results-notes"></div>
|
|
|
|
| 313 |
<b>Relative CER reduction</b>: how much of that damage the model repaired.
|
| 314 |
<b>Over‑correction</b>: how much text that was already right the model needlessly changed.
|
| 315 |
<b>Fix rate</b>: how much of what was actually wrong it fixed.</p>
|
| 316 |
+
<p><b>The full record.</b> Every experiment behind these numbers — scripts, configs, findings
|
| 317 |
+
(including the negative results) — is logged in a public
|
| 318 |
+
<a href="https://huggingface.co/buckets/davanstrien/diffusiongemma-ocr-bench">experiment-log bucket</a>;
|
| 319 |
+
all runs executed on <a href="https://huggingface.co/docs/huggingface_hub/en/guides/jobs">Hugging Face Jobs</a>.</p>
|
| 320 |
</div>
|
| 321 |
<div id="results-table"><p style="text-align:center;font-style:italic">Fetching the ledger…</p></div>
|
| 322 |
<div class="footnotes" id="results-notes"></div>
|