diff --git a/_layouts/post.html b/_layouts/post.html
index 404b388..a57027e 100644
--- a/_layouts/post.html
+++ b/_layouts/post.html
@@ -110,8 +110,7 @@ <h2 class="secondary-header mb-30">
                     target="_blank" rel="noopener noreferrer">
                     <img
                         src="/static/images/bluesky_logo_icon.svg" style="height: 15px;"
-                        alt="Bluesky"></a>    
-                    
+                        alt="Bluesky">
                 </a>
                 
             </div>
diff --git a/_posts/2020-06-21-ecmlpkdd-salha.md b/_posts/2020-06-21-ecmlpkdd-salha.md
index 17076a3..c9539a2 100644
--- a/_posts/2020-06-21-ecmlpkdd-salha.md
+++ b/_posts/2020-06-21-ecmlpkdd-salha.md
@@ -49,7 +49,7 @@ More precisely, our contribution is threefold:
 <ul>
     <li> We introduce and study simpler versions of graph AE and VAE, replacing multi-layer GCN encoders by linear models w.r.t. the direct neighborhood (one-hop) adjacency matrix of the graph, involving a unique weight matrix to tune, fewer operations and no activation function. </li>
     <li> Through an extensive empirical analysis on 17 real-world graphs with various sizes and characteristics, we show that these simplified models consistently reach competitive performances w.r.t. GCN-based graph AE and VAE on link prediction and node clustering tasks. We identify the settings where simple linear encoders appear as an effective alternative to GCNs, and as first relevant baseline to implement before diving into more complex models. We also question the relevance of current benchmark datasets (Cora, Citeseer, Pubmed) commonly used in the literature to evaluate graph AE and VAE. </li>
-    <li> We publicly release the code of these experiments, for reproducibility and easier future usages. </li>
+    <li> We publicly release the code of these experiments, for reproducibility and easier future use. </li>
 </ul>
 
 
diff --git a/_posts/2020-09-22-recsys-bendada.md b/_posts/2020-09-22-recsys-bendada.md
index ad582b1..a2ef129 100644
--- a/_posts/2020-09-22-recsys-bendada.md
+++ b/_posts/2020-09-22-recsys-bendada.md
@@ -28,7 +28,7 @@ promising research on these aspects to industrial-level applications.
 
 In particular, many global mobile apps and websites, notably from the music streaming industry, currently 
 leverage <i>swipeable carousels</i> to display recommended content on their homepages. These carousels, 
-also referred to as <i>sliders</i> or <i>shelves</i>, consist in ranked lists of items or <i>cards</i> (albums, 
+also referred to as <i>sliders</i> or <i>shelves</i>, consist of ranked lists of items or <i>cards</i> (albums, 
 artists, playlists...). A few cards are initially displayed to the users, who can click on them or swipe on the 
 screen to see some of the additional cards from the carousel.
 
diff --git a/_posts/2022-12-05-ismir-afchar.md b/_posts/2022-12-05-ismir-afchar.md
index f1f2178..b01f723 100644
--- a/_posts/2022-12-05-ismir-afchar.md
+++ b/_posts/2022-12-05-ismir-afchar.md
@@ -36,7 +36,7 @@ Music signals are difficult to interpret from their low-level features, perhaps
         src="{{ '/static/images/publis/afchar22ismir/spec_2.png' | prepend: site.url }}"
         alt="Spectrogram explanation"/>
     <br><span style="color: #AAA; font-size: 0.6em; width:50%;">
-	These two images of a spectrogram and generated explanation were shamefully stolen from <a href="https://arxiv.org/pdf/1905.11760.pdf"><i>"Two-level Explanations in Music Emotion Recognition" </i> V. Praher et al (2019)</a> for demonstration purpose.</span>
+	These two images of a spectrogram and generated explanation were shamefully stolen from <a href="https://arxiv.org/pdf/1905.11760.pdf"><i>"Two-level Explanations in Music Emotion Recognition" </i> V. Praher et al (2019)</a> for demonstration purposes.</span>
 </div>