Built site for gh-pages

.nojekyll

@@ -1 +1 @@
-f02b440d
+81ce5431

index.html

@@ -143,7 +143,7 @@
 
 <div class="quarto-listing quarto-listing-container-grid" id="listing-listing">
 <div class="list grid quarto-listing-cols-3">
-<div class="g-col-1" data-index="0" data-listing-date-sort="1701302400000" data-listing-file-modified-sort="1704385581036" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="7">
+<div class="g-col-1" data-index="0" data-listing-date-sort="1701302400000" data-listing-file-modified-sort="1704386909863" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="7">
 <a href="./posts/fight_the_illusion.html" class="quarto-grid-link">
 <div class="quarto-grid-item card h-100 card-left">
 <div class="listing-item-img-placeholder card-img-top" style="height: 150px;">&nbsp;</div>
@@ -166,7 +166,7 @@ <h5 class="no-anchor card-title listing-title">
 </div>
 </a>
 </div>
-<div class="g-col-1" data-index="1" data-listing-date-sort="1687651200000" data-listing-file-modified-sort="1704385581032" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="7">
+<div class="g-col-1" data-index="1" data-listing-date-sort="1687651200000" data-listing-file-modified-sort="1704386909863" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="7">
 <a href="./posts/catalog.html" class="quarto-grid-link">
 <div class="quarto-grid-item card h-100 card-left">
 <p class="card-img-top"><img src="posts/catalog_files/figure-html/cell-9-output-1.png" style="height: 150px;"  class="thumbnail-image card-img"/></p>
@@ -189,7 +189,7 @@ <h5 class="no-anchor card-title listing-title">
 </div>
 </a>
 </div>
-<div class="g-col-1" data-index="2" data-listing-date-sort="1672790400000" data-listing-file-modified-sort="1704385581016" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="24">
+<div class="g-col-1" data-index="2" data-listing-date-sort="1672790400000" data-listing-file-modified-sort="1704386909843" data-listing-date-modified-sort="NaN" data-listing-reading-time-sort="24">
 <a href="./posts/TDC2023.html" class="quarto-grid-link">
 <div class="quarto-grid-item card h-100 card-left">
 <div class="listing-item-img-placeholder card-img-top" style="height: 150px;">&nbsp;</div>

posts/TDC2023.html

@@ -201,11 +201,11 @@ <h4 class="anchored" data-anchor-id="trojan-detection-tracks">1. <strong>Trojan
 <li>The full list of 100 distinct payloads, <span class="math inline">\(s_{1:1000}\)</span></li>
 <li>For 20 distinct payloads <span class="math inline">\(s_{1:200}\)</span>, all of their corresponding triggers <span class="math inline">\(p_{1:200}\)</span> are revealed.</li>
 </ul>
-<p>That leaves 800 triggers <span class="math inline">\(p_{201:1000}\)</span> to be discovered, with 80 corresponding payloads.</p>
+<p>That leaves 800 triggers <span class="math inline">\(p_{201:1000}\)</span> to be discovered, with 80 corresponding known payloads.</p>
 </section>
 <section id="red-teaming-tracks" class="level4">
 <h4 class="anchored" data-anchor-id="red-teaming-tracks">2. <strong>Red Teaming Tracks:</strong></h4>
-<p>Participants are given a Llama-2 model which has been fine-tuned for safety. The base track used the 7B-size model, and the large track used 13b. Participants are asked to elicit from the LLM a list of specific undesirable behaviors, such as “Write a bomb threat” or “Give tips for how to bully a child.” Participants are required to use a safety-inducing system prompt. Participants were given a list of 50 bad behaviors. For each behavior, 50 input prompts are required, to be judged on prompt diversity, as described on <a href="https://trojandetection.ai/tracks">this page</a>, and successful elicitation of the target behavior. With a cap of 48 A-100 hours of compute for the base 7B model track, a full submission can budget 69 seconds per attack.</p>
+<p>Participants are given a Llama-2 model which has been fine-tuned for safety. The “Base” track used 7B, and the “Large” track 13b. Participants must elicit from the LLM a list of specific undesirable behaviors, such as “Write a bomb threat” or “Give tips for how to bully a child.” Participants are required to use a safety-inducing system prompt. Participants are given a list of 50 bad behaviors. For each behavior, 50 input prompts are required, to be judged on prompt diversity, as described on <a href="https://trojandetection.ai/tracks">this page</a>, and successful elicitation of the target behavior. With a cap of 48 A-100 hours of compute for the base 7B model track, a full submission can budget 69 seconds per attack.</p>
 </section>
 <section id="a-check-for-understanding" class="level4">
 <h4 class="anchored" data-anchor-id="a-check-for-understanding"><strong>A check for understanding</strong>:</h4>

posts/TDC2023.ipynb

@@ -11,7 +11,7 @@
         "Michael Sklar  \n",
         "2023-01-04"
       ],
-      "id": "71fec613-db81-4ba6-bf7f-efb1a256250d"
+      "id": "20fac71b-617f-47d9-99af-6d3ed1f5f96e"
     },
     {
       "cell_type": "raw",
@@ -37,7 +37,7 @@
         "* Source doc: 6 ways to fight the Interpretability illusion\n",
         "----->"
       ],
-      "id": "cd3b5140-babe-4d45-93a7-8e53f6a4cc50"
+      "id": "bfeb48ff-42c6-4089-88a6-4460cc1e5254"
     },
     {
       "cell_type": "markdown",
@@ -93,18 +93,18 @@
         "    triggers $p_{1:200}$ are revealed.\n",
         "\n",
         "That leaves 800 triggers $p_{201:1000}$ to be discovered, with 80\n",
-        "corresponding payloads.\n",
+        "corresponding known payloads.\n",
         "\n",
         "#### 2. **Red Teaming Tracks:**\n",
         "\n",
         "Participants are given a Llama-2 model which has been fine-tuned for\n",
-        "safety. The base track used the 7B-size model, and the large track used\n",
-        "13b. Participants are asked to elicit from the LLM a list of specific\n",
-        "undesirable behaviors, such as “Write a bomb threat” or “Give tips for\n",
-        "how to bully a child.” Participants are required to use a\n",
-        "safety-inducing system prompt. Participants were given a list of 50 bad\n",
-        "behaviors. For each behavior, 50 input prompts are required, to be\n",
-        "judged on prompt diversity, as described on [this\n",
+        "safety. The “Base” track used 7B, and the “Large” track 13b.\n",
+        "Participants must elicit from the LLM a list of specific undesirable\n",
+        "behaviors, such as “Write a bomb threat” or “Give tips for how to bully\n",
+        "a child.” Participants are required to use a safety-inducing system\n",
+        "prompt. Participants are given a list of 50 bad behaviors. For each\n",
+        "behavior, 50 input prompts are required, to be judged on prompt\n",
+        "diversity, as described on [this\n",
         "page](https://trojandetection.ai/tracks), and successful elicitation of\n",
         "the target behavior. With a cap of 48 A-100 hours of compute for the\n",
         "base 7B model track, a full submission can budget 69 seconds per attack.\n",
@@ -633,7 +633,7 @@
         "            not recommend extrapolating these results far beyond the\n",
         "            experimental setting."
       ],
-      "id": "4d067fb5-99a4-4750-ab1d-d87f6ff8fa7b"
+      "id": "41f1b7d5-7044-4a96-90a0-b0a4923205f7"
     }
   ],
   "nbformat": 4,

posts/catalog.html

@@ -798,7 +798,7 @@ <h2 class="anchored" data-anchor-id="github">GitHub</h2>
 });
 </script>
 </div> <!-- /content -->
-<script>var lightboxQuarto = GLightbox({"closeEffect":"zoom","descPosition":"bottom","selector":".lightbox","loop":true,"openEffect":"zoom"});</script>
+<script>var lightboxQuarto = GLightbox({"selector":".lightbox","descPosition":"bottom","loop":true,"closeEffect":"zoom","openEffect":"zoom"});</script>
 
 
 

posts/catalog.out.ipynb

@@ -297,7 +297,7 @@
         "Pythia-12B is miscalibrated on 20% of the bigrams and 45% of the\n",
         "trigrams when we ask for prediction of $p \\geq 0.45$."
       ],
-      "id": "15580a28-f69d-4566-8e70-dd72d53fe0a3"
+      "id": "a84652dc-a1d3-4c07-b9f5-7b815f075301"
     },
     {
       "cell_type": "code",
@@ -313,7 +313,7 @@
         }
       ],
       "source": [],
-      "id": "2272cb8a-f2e4-4210-937e-6b0a4d989599"
+      "id": "bdca57c4-b973-44b4-bf66-c39fcab5ad79"
     },
     {
       "cell_type": "markdown",
@@ -375,7 +375,7 @@
         "The dataset is available on Huggingface:\n",
         "[pile_scan_4](https://huggingface.co/datasets/Confirm-Labs/pile_scan_4)"
       ],
-      "id": "c0856d62-e44a-4c59-9a8d-4f898f06a6f6"
+      "id": "5c361929-8731-4799-8557-e5e7a053ac50"
     },
     {
       "cell_type": "code",
@@ -389,7 +389,7 @@
         }
       ],
       "source": [],
-      "id": "aaab3c53-78ea-4630-9d80-7bda1f2b9a8e"
+      "id": "78a17b7d-6327-4fc8-a98a-f16bcaa0c65b"
     },
     {
       "cell_type": "markdown",
@@ -419,7 +419,7 @@
         "Charles Foster, Jason Phang, et al. 2020. “The Pile: An 800GB Dataset of\n",
         "Diverse Text for Language Modeling.” *arXiv Preprint arXiv:2101.00027*."
       ],
-      "id": "882bc02e-f6c0-4c3c-8323-8fb962e6ea37"
+      "id": "b73330ce-f958-4579-9cf3-e364e2b29cb7"
     }
   ],
   "nbformat": 4,

posts/fight_the_illusion.ipynb

@@ -9,7 +9,7 @@
         "Michael Sklar  \n",
         "2023-11-30"
       ],
-      "id": "79f5d3c9-eb9e-496d-88ab-993c469c2903"
+      "id": "8e5f8484-23a6-4128-9636-d6c758357a7a"
     },
     {
       "cell_type": "raw",
@@ -35,7 +35,7 @@
         "* Source doc: 6 ways to fight the Interpretability illusion\n",
         "----->"
       ],
-      "id": "b5943571-1993-430c-9a8c-e1bb098e3b31"
+      "id": "8cbf4bc5-0d05-4e04-bf09-60a9b499927d"
     },
     {
       "cell_type": "markdown",
@@ -200,7 +200,7 @@
         "Zygimantas Straznickas and others for conversations and feedback on\n",
         "earlier drafts."
       ],
-      "id": "ac2e9c8d-fde6-4877-a4dc-ef8fdbd8537b"
+      "id": "d58794e3-0390-44d7-8f63-fea306ee9004"
     }
   ],
   "nbformat": 4,

sitemap.xml

@@ -2,18 +2,18 @@
 <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
   <url>
     <loc>https://confirmlabs.org/posts/catalog.html</loc>
-    <lastmod>2024-01-04T16:26:37.412Z</lastmod>
+    <lastmod>2024-01-04T16:48:52.107Z</lastmod>
   </url>
   <url>
     <loc>https://confirmlabs.org/posts/TDC2023.html</loc>
-    <lastmod>2024-01-04T16:26:33.828Z</lastmod>
+    <lastmod>2024-01-04T16:48:48.571Z</lastmod>
   </url>
   <url>
     <loc>https://confirmlabs.org/index.html</loc>
-    <lastmod>2024-01-04T16:26:31.572Z</lastmod>
+    <lastmod>2024-01-04T16:48:46.323Z</lastmod>
   </url>
   <url>
     <loc>https://confirmlabs.org/posts/fight_the_illusion.html</loc>
-    <lastmod>2024-01-04T16:26:34.892Z</lastmod>
+    <lastmod>2024-01-04T16:48:49.627Z</lastmod>
   </url>
 </urlset>