Update

MBZUAI-Paris · Jan 24, 2025 · d815967 · d815967
1 parent a310713
commit d815967
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 15 deletions.
diff --git a/index.html b/index.html
@@ -89,10 +89,10 @@ <h2>Program Highlights & Poster Registration</h2>
     <h2>Organizers</h2>
     <p>
         <b>Organizing committee:</b><br>
-        <a href="https://mbzuai.ac.ae/study/faculty/professor-eric-moulines/">Eric Moulines</a>, Ecole Polytechnique & MBZUAI<br>
-        Guokan Shang, MBZUAI France Lab<br>
-        <a href="https://mbzuai.ac.ae/study/faculty/michalis-vazirgiannis/">Michalis Vazirgiannis</a>, Ecole Polytechnique & MBZUAI<br>
-        <a href="https://mbzuai.ac.ae/study/faculty/kun-zhang/">Kun Zhang</a>, MBZUAI
+        <a target="_blank" href="https://mbzuai.ac.ae/study/faculty/professor-eric-moulines/">Eric Moulines</a>, Ecole Polytechnique & MBZUAI<br>
+        <a target="_blank" href="https://www.linkedin.com/in/guokan-shang">Guokan Shang</a>, MBZUAI France Lab<br>
+        <a target="_blank" href="https://mbzuai.ac.ae/study/faculty/michalis-vazirgiannis/">Michalis Vazirgiannis</a>, Ecole Polytechnique & MBZUAI<br>
+        <a target="_blank" href="https://mbzuai.ac.ae/study/faculty/kun-zhang/">Kun Zhang</a>, MBZUAI
     </p>
     <p>
         <b>Logistics support:</b><br>

diff --git a/program-day-1/index.html b/program-day-1/index.html
@@ -57,7 +57,7 @@ <h2>Program on Wednesday, February 12</h2>
     <table>
         <tr>
             <td class="date" rowspan="2">
-                9:00am
+                09:00am
             </td>
             <td class="title-special">
                 Registration and Coffee &amp; Tea!
@@ -173,7 +173,7 @@ <h2>Program on Wednesday, February 12</h2>
     <table id="PUTSPEAKERNAMEHERE">
         <tr>
             <td class="date" rowspan="3">
-                12:20am
+                12:20pm
             </td>
             <td class="title">
                 Exploiting Knowledge for Model-based Deep Music Generation
@@ -194,7 +194,7 @@ <h2>Program on Wednesday, February 12</h2>
     <table>
         <tr>
             <td class="date" rowspan="2">
-                12:20pm
+                13:00pm
             </td>
             <td class="title-special">
                 Lunch
@@ -212,7 +212,7 @@ <h2>Program on Wednesday, February 12</h2>
     <table id="PUTSPEAKERNAMEHERE">
         <tr>
             <td class="date" rowspan="3">
-                14:00am
+                14:00pm
             </td>
             <td class="title">
                 TBD
@@ -237,7 +237,7 @@ <h2>Program on Wednesday, February 12</h2>
                 14:40pm
             </td>
             <td class="title">
-                TBD
+                Intricacies of Game-theoretical LLM Alignment
             </td>
         </tr>
         <tr>
@@ -247,7 +247,7 @@ <h2>Program on Wednesday, February 12</h2>
         </tr>
         <tr>
             <td class="abstract">
-                TBD
+                Ensuring alignment of language models' outputs with human preferences is critical to guarantee a useful, safe, and pleasant user experience. Thus, human alignment has been extensively studied recently and several methods such as Reinforcement Learning from Human Feedback (RLHF), Direct Policy Optimisation (DPO) and Sequence Likelihood Calibration (SLiC) have emerged. In this paper, our contribution is two-fold. First, we show the equivalence between two recent alignment methods, namely Identity Policy Optimisation (IPO) and Nash Mirror Descent (Nash-MD). Second, we introduce a generalisation of IPO, named IPO-MD, that leverages the regularised sampling approach proposed by Nash-MD. @This equivalence may seem surprising at first sight, since IPO is an offline method whereas Nash-MD is an online method using a preference model. However, this equivalence can be proven when we consider the online version of IPO, that is when both generations are sampled by the online policy and annotated by a trained preference model. Optimising the IPO loss with such a stream of data becomes equivalent to finding the Nash equilibrium of the preference model through self-play. Building on this equivalence, we introduce the IPO-MD algorithm that generates data with a mixture policy (between the online and reference policy) similarly as the general Nash-MD algorithm. We compare online-IPO and IPO-MD to different online versions of existing losses on preference data such as DPO and SLiC on a summarisation task.
             </td>
         </tr>
     </table>
@@ -257,7 +257,7 @@ <h2>Program on Wednesday, February 12</h2>
     <table>
         <tr>
             <td class="date" rowspan="2">
-                14:50pm
+                15:20pm
             </td>
             <td class="title-special">
                 Coffee &amp; Tea Break
@@ -339,7 +339,7 @@ <h2>Program on Wednesday, February 12</h2>
                 18:00pm
             </td>
             <td class="title">
-                Poster Session at MBZUAI France Lab
+                Poster Session with Buffet at MBZUAI France Lab
             </td>
         </tr>
         <tr>

diff --git a/program-day-2/index.html b/program-day-2/index.html
@@ -57,7 +57,7 @@ <h2>Program on Thursday, February 13</h2>
     <table>
         <tr>
             <td class="date" rowspan="2">
-                9:00am
+                09:00am
             </td>
             <td class="title-special">
                 Registration and Coffee &amp; Tea!
@@ -173,7 +173,7 @@ <h2>Program on Thursday, February 13</h2>
     <table id="PUTSPEAKERNAMEHERE">
         <tr>
             <td class="date" rowspan="3">
-                12:20am
+                12:20pm
             </td>
             <td class="title">
                 GFlowNets: A Novel Framework for Diverse Generation in Combinatorial and Continuous Spaces
@@ -194,7 +194,7 @@ <h2>Program on Thursday, February 13</h2>
     <table>
         <tr>
             <td class="date" rowspan="2">
-                13:10pm
+                13:00pm
             </td>
             <td class="title-special">
                 Lunch