updated pubs and cv

content/publication/andukuri2024stargate.md

@@ -9,9 +9,9 @@
 title = "STaR-GATE: Teaching Language Models to Ask Clarifying Questions"
 date = "2024-03-31"
 authors = ["C. Andukuri","J. Fränken","T. Gerstenberg","N. D. Goodman"]
-publication_types = ["1"]
-publication_short = "_arXiv_"
-publication = "Andukuri, C., Fränken, J., Gerstenberg, T., Goodman, N. D. (2024). STaR-GATE: Teaching Language Models to Ask Clarifying Questions. _arXiv_."
+publication_types = ["3"]
+publication_short = "_Conference on Language Modeling (COLM)_"
+publication = "Andukuri, C., Fränken, J., Gerstenberg, T., Goodman, N. D. (2024). STaR-GATE: Teaching Language Models to Ask Clarifying Questions. _Conference on Language Modeling (COLM)_."
 abstract = "When prompting language models to complete a task, users often leave important aspects unsaid. While asking questions could resolve this ambiguity (GATE; Li et al., 2023), models often struggle to ask good questions. We explore a language model's ability to self-improve (STaR; Zelikman et al., 2022) by rewarding the model for generating useful questions---a simple method we dub STaR-GATE. We generate a synthetic dataset of 25,500 unique persona-task prompts to simulate conversations between a pretrained language model---the Questioner---and a Roleplayer whose preferences are unknown to the Questioner. By asking questions, the Questioner elicits preferences from the Roleplayer. The Questioner is iteratively finetuned on questions that increase the probability of high-quality responses to the task, which are generated by an Oracle with access to the Roleplayer's latent preferences. After two iterations of self-improvement, the Questioner asks better questions, allowing it to generate responses that are preferred over responses from the initial model on 72% of tasks. Our results indicate that teaching a language model to ask better questions leads to better personalized responses."
 image_preview = ""
 selected = false

docs/home/index.xml

@@ -15,7 +15,7 @@
     <item>
       <title>Causality in Cognition Lab</title>
       <link>https://cicl.stanford.edu/home/home/</link>
-      <pubDate>Sun, 15 Oct 2017 00:00:00 -0500</pubDate>
+      <pubDate>Sun, 15 Oct 2017 00:00:00 -0700</pubDate>
 
       <guid>https://cicl.stanford.edu/home/home/</guid>
       <description>The Causality in Cognition Lab at Stanford University studies the role of causality in our understanding of the world and of each other.</description>
@@ -53,7 +53,7 @@ We are looking to hire a lab manager to start in the fall of 2018. For informati
     <item>
       <title>Selected&lt;br&gt;publications</title>
       <link>https://cicl.stanford.edu/home/publications_selected/</link>
-      <pubDate>Wed, 20 Apr 2016 00:00:00 -0500</pubDate>
+      <pubDate>Wed, 20 Apr 2016 00:00:00 -0700</pubDate>
 
       <guid>https://cicl.stanford.edu/home/publications_selected/</guid>
       <description></description>

docs/member/tobias_gerstenberg/index.html

@@ -1153,7 +1153,7 @@ <h2 id="publications">Publications</h2>
     (2024).
 
   <a href="https://cicl.stanford.edu/publication/andukuri2024stargate/" itemprop="name">STaR-GATE: Teaching Language Models to Ask Clarifying Questions</a>.
-  <em>arXiv</em>.
+  <em>Conference on Language Modeling (COLM)</em>.
 
 
 

docs/publication/andukuri2024stargate/index.html

@@ -266,8 +266,8 @@ <h3>Abstract</h3>
           <div class="col-xs-12 col-sm-3 pub-row-heading">Type</div>
           <div class="col-xs-12 col-sm-9">
 
-            <a href="/publication/#1">
-              Preprint
+            <a href="/publication/#3">
+              Conference Proceedings
             </a>
 
           </div>
@@ -283,7 +283,7 @@ <h3>Abstract</h3>
       <div class="col-sm-10">
         <div class="row">
           <div class="col-xs-12 col-sm-3 pub-row-heading">Publication</div>
-          <div class="col-xs-12 col-sm-9">Andukuri, C., Fränken, J., Gerstenberg, T., Goodman, N. D. (2024). STaR-GATE: Teaching Language Models to Ask Clarifying Questions. <em>arXiv</em>.</div>
+          <div class="col-xs-12 col-sm-9">Andukuri, C., Fränken, J., Gerstenberg, T., Goodman, N. D. (2024). STaR-GATE: Teaching Language Models to Ask Clarifying Questions. <em>Conference on Language Modeling (COLM)</em>.</div>
         </div>
       </div>
       <div class="col-sm-1"></div>

docs/publication/index.html

@@ -2754,7 +2754,7 @@ <h1>Publications</h1>
 
 
 
-				<div class='grid-sizer col-md-12 isotope-item pubtype-1 year-2024 author-'>
+				<div class='grid-sizer col-md-12 isotope-item pubtype-3 year-2024 author-'>
 
 						<div class="pub-list-item" style="margin-bottom: 1rem" itemscope itemtype="http://schema.org/CreativeWork">
   <span itemprop="author">
@@ -2763,7 +2763,7 @@ <h1>Publications</h1>
     (2024).
 
   <a href="https://cicl.stanford.edu/publication/andukuri2024stargate/" itemprop="name">STaR-GATE: Teaching Language Models to Ask Clarifying Questions</a>.
-  <em>arXiv</em>.
+  <em>Conference on Language Modeling (COLM)</em>.
 
 
 

docs/publication_types/1/index.html

@@ -274,15 +274,6 @@ <h2><a href="https://cicl.stanford.edu/publication/du2024robotic/">To Err is Rob
     </div>
   </div>
 
-  <div>
-    <h2><a href="https://cicl.stanford.edu/publication/andukuri2024stargate/">STaR-GATE: Teaching Language Models to Ask Clarifying Questions</a></h2>
-    <div class="article-style">
-
-      When prompting language models to complete a task, users often leave important aspects unsaid. While asking questions could resolve this ambiguity (GATE; Li et al., 2023), models often struggle to ask good questions. We explore a language model's …
-
-    </div>
-  </div>
-
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/srivastava2022imitation/">Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models</a></h2>
     <div class="article-style">

docs/publication_types/1/index.xml

@@ -48,15 +48,6 @@
       <description></description>
     </item>
 
-    <item>
-      <title>STaR-GATE: Teaching Language Models to Ask Clarifying Questions</title>
-      <link>https://cicl.stanford.edu/publication/andukuri2024stargate/</link>
-      <pubDate>Sun, 31 Mar 2024 00:00:00 +0000</pubDate>
-
-      <guid>https://cicl.stanford.edu/publication/andukuri2024stargate/</guid>
-      <description></description>
-    </item>
-
     <item>
       <title>Beyond the Imitation Game: Quantifying and extrapolating the capabilities of language models</title>
       <link>https://cicl.stanford.edu/publication/srivastava2022imitation/</link>

docs/publication_types/3/index.xml

@@ -102,6 +102,15 @@
       <description></description>
     </item>
 
+    <item>
+      <title>STaR-GATE: Teaching Language Models to Ask Clarifying Questions</title>
+      <link>https://cicl.stanford.edu/publication/andukuri2024stargate/</link>
+      <pubDate>Sun, 31 Mar 2024 00:00:00 +0000</pubDate>
+
+      <guid>https://cicl.stanford.edu/publication/andukuri2024stargate/</guid>
+      <description></description>
+    </item>
+
     <item>
       <title>Anticipating the risks and benefits of counterfactual world simulation models</title>
       <link>https://cicl.stanford.edu/publication/kirfel2023anticipating/</link>

docs/publication_types/3/page/2/index.html

@@ -238,6 +238,15 @@ <h1 class="pt-3">3</h1>
 
 
 
+  <div>
+    <h2><a href="https://cicl.stanford.edu/publication/andukuri2024stargate/">STaR-GATE: Teaching Language Models to Ask Clarifying Questions</a></h2>
+    <div class="article-style">
+
+      When prompting language models to complete a task, users often leave important aspects unsaid. While asking questions could resolve this ambiguity (GATE; Li et al., 2023), models often struggle to ask good questions. We explore a language model's …
+
+    </div>
+  </div>
+
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/kirfel2023anticipating/">Anticipating the risks and benefits of counterfactual world simulation models</a></h2>
     <div class="article-style">
@@ -319,15 +328,6 @@ <h2><a href="https://cicl.stanford.edu/publication/nam2023show/">Show and tell:
     </div>
   </div>
 
-  <div>
-    <h2><a href="https://cicl.stanford.edu/publication/zhang2023llm/">You are what you&#39;re for: Essentialist categorization in large language models</a></h2>
-    <div class="article-style">
-
-      How do essentialist beliefs about categories arise? We hypothesize that such beliefs are transmitted via language. We subject large language models (LLMs) to vignettes from the literature on essentialist categorization and find that they align well …
-
-    </div>
-  </div>
-
 
 
 <nav>

docs/publication_types/3/page/3/index.html

@@ -238,6 +238,15 @@ <h1 class="pt-3">3</h1>
 
 
 
+  <div>
+    <h2><a href="https://cicl.stanford.edu/publication/zhang2023llm/">You are what you&#39;re for: Essentialist categorization in large language models</a></h2>
+    <div class="article-style">
+
+      How do essentialist beliefs about categories arise? We hypothesize that such beliefs are transmitted via language. We subject large language models (LLMs) to vignettes from the literature on essentialist categorization and find that they align well …
+
+    </div>
+  </div>
+
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/cao2023semantics/">A Semantics for Causing, Enabling, and Preventing Verbs Using Structural Causal Models</a></h2>
     <div class="article-style">
@@ -319,15 +328,6 @@ <h2><a href="https://cicl.stanford.edu/publication/vodrahalli2022humans/">Do hum
     </div>
   </div>
 
-  <div>
-    <h2><a href="https://cicl.stanford.edu/publication/vodrahalli2022uncalibrated/">Uncalibrated models can improve human-AI collaboration</a></h2>
-    <div class="article-style">
-
-      In many practical applications of AI, an AI model is used as a decision aid for human users. The AI provides advice that a human (sometimes) incorporates into their decision-making process. The AI advice is often presented with some measure of …
-
-    </div>
-  </div>
-
 
 
 <nav>

docs/publication_types/3/page/4/index.html

@@ -238,6 +238,15 @@ <h1 class="pt-3">3</h1>
 
 
 
+  <div>
+    <h2><a href="https://cicl.stanford.edu/publication/vodrahalli2022uncalibrated/">Uncalibrated models can improve human-AI collaboration</a></h2>
+    <div class="article-style">
+
+      In many practical applications of AI, an AI model is used as a decision aid for human users. The AI provides advice that a human (sometimes) incorporates into their decision-making process. The AI advice is often presented with some measure of …
+
+    </div>
+  </div>
+
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/davis2021fishing/">Who went fishing? Inferences from social evaluations</a></h2>
     <div class="article-style">
@@ -319,15 +328,6 @@ <h2><a href="https://cicl.stanford.edu/publication/stephan2017marbles/">Marbles
     </div>
   </div>
 
-  <div>
-    <h2><a href="https://cicl.stanford.edu/publication/yildirim2017problem/">Physical problem solving: Joint planning with symbolic, geometric, and dynamic constraints</a></h2>
-    <div class="article-style">
-
-      In this paper, we present a new task that investigates how people interact with and make judgments about towers of blocks. In Experiment 1, participants in the lab solved a series of problems in which they had to re-configure three blocks from an …
-
-    </div>
-  </div>
-
 
 
 <nav>

docs/publication_types/3/page/5/index.html

@@ -238,6 +238,15 @@ <h1 class="pt-3">3</h1>
 
 
 
+  <div>
+    <h2><a href="https://cicl.stanford.edu/publication/yildirim2017problem/">Physical problem solving: Joint planning with symbolic, geometric, and dynamic constraints</a></h2>
+    <div class="article-style">
+
+      In this paper, we present a new task that investigates how people interact with and make judgments about towers of blocks. In Experiment 1, participants in the lab solved a series of problems in which they had to re-configure three blocks from an …
+
+    </div>
+  </div>
+
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/niemi2016implicit/">Implicit measurement of motivated causal attribution</a></h2>
     <div class="article-style">
@@ -319,15 +328,6 @@ <h2><a href="https://cicl.stanford.edu/publication/gerstenberg2014simulation/">F
     </div>
   </div>
 
-  <div>
-    <h2><a href="https://cicl.stanford.edu/publication/bramley2014order/">The order of things: Inferring causal structure from temporal patterns</a></h2>
-    <div class="article-style">
-
-      The timing and order in which a set of events occur strongly influences whether people judge them to be causally related. But what do people think particular temporal patterns of events tell them about causal structure? And how do they integrate …
-
-    </div>
-  </div>
-
 
 
 <nav>

docs/publication_types/3/page/6/index.html

@@ -238,6 +238,15 @@ <h1 class="pt-3">3</h1>
 
 
 
+  <div>
+    <h2><a href="https://cicl.stanford.edu/publication/bramley2014order/">The order of things: Inferring causal structure from temporal patterns</a></h2>
+    <div class="article-style">
+
+      The timing and order in which a set of events occur strongly influences whether people judge them to be causally related. But what do people think particular temporal patterns of events tell them about causal structure? And how do they integrate …
+
+    </div>
+  </div>
+
   <div>
     <h2><a href="https://cicl.stanford.edu/publication/gerstenberg2014replacement/">Wins above replacement: Responsibility attributions as counterfactual replacements</a></h2>
     <div class="article-style">