update website

layer6ai-labs · Jul 16, 2024 · 2341418 · 2341418
1 parent e12232f
commit 2341418
Show file tree

Hide file tree

Showing 11 changed files with 263 additions and 1,201 deletions.
diff --git a/README.md b/README.md
@@ -1,6 +1,6 @@
 # DGM Geometry
 
-Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like OOD detection. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints.
+Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like the likelihood out-of-distribution paradox. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints.
 
 ## Installation
 
@@ -101,6 +101,11 @@ You may update the content of the website by changing the notebooks and `.qmd` f
 To see the updates in real-time, run the following command that will start a local server:
 ```bash
 # download and install Quarto from https://quarto.org/docs/get-started/
-cd docs
+cd docs_quarto
 quarto preview # opens up a local server on port 4200
 ```
+
+To publish the website, move everything to the docs directory:
+```bash
+cp -r docs_quarto/_output/* docs/
+```
diff --git a/docs/index.html b/docs/index.html
@@ -269,7 +269,7 @@ <h1 class="title">The Geometry of Deep Generative Models</h1>
 </header>
 
 
-<p>Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like memorization or be used for tasks like OOD detection. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints. Please navigate to our <a href="https://github.com/blross/dgm-geometry">repository</a> and run the following steps to get started.</p>
+<p>Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like the likelihood out-of-distribution paradox. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints. Please navigate to our <a href="https://github.com/blross/dgm-geometry">repository</a> and run the following steps to get started.</p>
 <section id="installation" class="level2" data-number="1">
 <h2 data-number="1" class="anchored" data-anchor-id="installation"><span class="header-section-number">1</span> Installation</h2>
 <p>We use a conda environment for this project. To create the environment, run:</p>

diff --git a/docs/search.json b/docs/search.json
diff --git a/docs/sections/lid.html b/docs/sections/lid.html
@@ -300,7 +300,7 @@ <h1 data-number="2"><span class="header-section-number">2</span> What is LID Use
 </section>
 <section id="useful-links" class="level1" data-number="3">
 <h1 data-number="3"><span class="header-section-number">3</span> Useful links</h1>
-<p>For a guide on how to use our LID estimators, check out <a href="../sections/lid/lid_guide.html">our notebook</a>. We are also planning to release our latest work on using the Fokker-Planck equation of diffusion models to estimate LID, which we call FLIPD. When posted, it will show up <a href="../sections/lid/flipd.html">here</a></p>
+<p>For a guide on how to use our LID estimators, check out <a href="../sections/lid/lid_guide.html">our notebook</a>. We are also planning to release our latest work on using the Fokker-Planck equation of diffusion models to estimate LID, which we call FLIPD. When posted, it will show up <a href="../sections/lid/flipd.html">here</a>.</p>
 <!-- 
 ## References -->
 <!-- 

diff --git a/docs/sections/lid/flipd.html b/docs/sections/lid/flipd.html
@@ -221,7 +221,7 @@ <h1 class="title">Fokker-Planck LID (FLIPD)</h1>
 </div>
 <section id="introduction" class="level2" data-number="1">
 <h2 data-number="1" class="anchored" data-anchor-id="introduction"><span class="header-section-number">1</span> Introduction</h2>
-<p>Check out our paper at <a href="https://arxiv.org/abs/2406.03537">arXiv</a> dsafor more details. The code will be available soon!</p>
+<p>Check out our paper at <a href="https://arxiv.org/abs/2406.03537">arXiv</a> for more details. The code will be available soon!</p>
 <!-- 
 ## References -->
 <!-- 

diff --git a/docs_quarto/_output/_readme_repository.qmd b/docs_quarto/_output/_readme_repository.qmd
diff --git a/docs_quarto/_output/index.html b/docs_quarto/_output/index.html
@@ -269,7 +269,7 @@ <h1 class="title">The Geometry of Deep Generative Models</h1>
 </header>
 
 
-<p>Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like memorization or be used for tasks like OOD detection. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints. Please navigate to our <a href="https://github.com/blross/dgm-geometry">repository</a> and run the following steps to get started.</p>
+<p>Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like the likelihood out-of-distribution paradox. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints. Please navigate to our <a href="https://github.com/blross/dgm-geometry">repository</a> and run the following steps to get started.</p>
 <section id="installation" class="level2" data-number="1">
 <h2 data-number="1" class="anchored" data-anchor-id="installation"><span class="header-section-number">1</span> Installation</h2>
 <p>We use a conda environment for this project. To create the environment, run:</p>

diff --git a/docs_quarto/_output/search.json b/docs_quarto/_output/search.json
diff --git a/docs_quarto/_output/sections/lid.html b/docs_quarto/_output/sections/lid.html
@@ -300,7 +300,7 @@ <h1 data-number="2"><span class="header-section-number">2</span> What is LID Use
 </section>
 <section id="useful-links" class="level1" data-number="3">
 <h1 data-number="3"><span class="header-section-number">3</span> Useful links</h1>
-<p>For a guide on how to use our LID estimators, check out <a href="../sections/lid/lid_guide.html">our notebook</a>. We are also planning to release our latest work on using the Fokker-Planck equation of diffusion models to estimate LID, which we call FLIPD. When posted, it will show up <a href="../sections/lid/flipd.html">here</a></p>
+<p>For a guide on how to use our LID estimators, check out <a href="../sections/lid/lid_guide.html">our notebook</a>. We are also planning to release our latest work on using the Fokker-Planck equation of diffusion models to estimate LID, which we call FLIPD. When posted, it will show up <a href="../sections/lid/flipd.html">here</a>.</p>
 <!-- 
 ## References -->
 <!-- 

diff --git a/docs_quarto/_output/sections/lid/flipd.html b/docs_quarto/_output/sections/lid/flipd.html
@@ -221,7 +221,7 @@ <h1 class="title">Fokker-Planck LID (FLIPD)</h1>
 </div>
 <section id="introduction" class="level2" data-number="1">
 <h2 data-number="1" class="anchored" data-anchor-id="introduction"><span class="header-section-number">1</span> Introduction</h2>
-<p>Check out our paper at <a href="https://arxiv.org/abs/2406.03537">arXiv</a> dsafor more details. The code will be available soon!</p>
+<p>Check out our paper at <a href="https://arxiv.org/abs/2406.03537">arXiv</a> for more details. The code will be available soon!</p>
 <!-- 
 ## References -->
 <!-- 

diff --git a/docs_quarto/index.qmd b/docs_quarto/index.qmd
@@ -15,8 +15,8 @@ jupyter: python3
 ---
 
 
-Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like memorization or be used for tasks like OOD detection. 
-In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints. Please navigate to our
+Here, we study how the geometry of deep generative models (DGMs) can inform our understanding of phenomena like the likelihood out-of-distribution paradox. In tandem and as a supplement to these topics, we also study algorithms for local intrinsic dimension (LID) estimation of datapoints.
+Please navigate to our
 [repository](https://github.com/blross/dgm-geometry) and run the following steps to get started.
 
 ## Installation