update and align model names in model section

index.html

@@ -137,15 +137,15 @@ <h1>Our models</h1>
                 <section class="pictures-section">
                     <a href="#k2" class="picture-link">
                         <img src="images/k2_logo.png" alt="" class="picture">
-                        <h2>K2 65B</h2>
+                        <h2>K2-65B</h2>
                     </a>
                     <a href="#crystal" class="picture-link">
                         <img src="images/cc_logo.png" alt="Image 2" class="picture">
-                        <h2>Crystal 7B</h2>
+                        <h2>CrystalCoder-7B</h2>
                     </a>
                     <a href="#amber" class="picture-link">
                         <img src="images/amber_logo.png" alt="Image 3" class="picture">
-                        <h2>Amber 7B</h2>
+                        <h2>Amber-7B</h2>
                     </a>
                 </section>
                 <section id="k2">
@@ -154,7 +154,7 @@ <h2>Amber 7B</h2>
                     </span>
                     <div class="content">
                         <div class="inner">
-                            <h2>K2</h2>
+                            <h2>K2-65B</h2>
                             <p>A <strong>65B parameter</strong> language model trained on <strong>1.4T tokens</strong>. It outperforms <strong>Llama 2 70B</strong>, but uses approximately <strong>35% less</strong> compute to train.</p>
                             <section>
                                 <ul class="actions fit"  >
@@ -173,7 +173,7 @@ <h2>K2</h2>
                     </span>
                     <div class="content">
                         <div class="inner">
-                            <h2>CrystalCoder</h2>
+                            <h2>CrystalCoder-7B</h2>
                             <p> A 7B parameter language model, distinctively trained on the SlimPajama and StarCoder datasets,
                                 eclipsing the <strong>Llama 2</strong> frontier, skillfully <strong>balances</strong> language and coding.
                                 Its instruction-following variant, <a href="https://huggingface.co/LLM360/CrystalChat" target="_blank">CrystalChat</a>, stands out as a <strong>top-scoring</strong> 7B chat model, trained on a carefully selected mix publicly available language and code datasets.</p>
@@ -192,7 +192,7 @@ <h2>CrystalCoder</h2>
                     </span>
                     <div class="content">
                         <div class="inner">
-                            <h2>Amber</h2>
+                            <h2>Amber-7B</h2>
                             <p>A 7B parameter English language model based on the <strong>LLaMA</strong> architecture has two fine-tuned instruction-following models named <a href="https://huggingface.co/LLM360/AmberChat" target="_blank">AmberChat</a> and <a href="https://huggingface.co/LLM360/AmberSafe" target="_blank">AmberSafe</a>.</p>
                             <ul class="actions fit">
                                 <li><a href="https://huggingface.co/collections/LLM360/amber-65e7333ff73c7bbb014f2f2f" target="_blank" class="button style3 fit">Model</a></li>