Update index.html

index.html

@@ -100,32 +100,32 @@
         <h2 class="title is-3">Abstract</h2>
         <div class="content has-text-justified">
           <p>
-            We present the first method capable of photorealistically reconstructing a non-rigidly
-            deforming scene using photos/videos captured casually from mobile phones.
+            WOPR, Word Predictor, is a memory-based language model developed in 2006-2011.
           </p>
           <p>
-            Our approach augments neural radiance fields
-            (NeRF) by optimizing an
-            additional continuous volumetric deformation field that warps each observed point into a
-            canonical 5D NeRF.
-            We observe that these NeRF-like deformation fields are prone to local minima, and
-            propose a coarse-to-fine optimization method for coordinate-based models that allows for
-            more robust optimization.
-            By adapting principles from geometry processing and physical simulation to NeRF-like
-            models, we propose an elastic regularization of the deformation field that further
-            improves robustness.
+            A memory-based language model, in this case running on the TiMBL classifier,
+            is in the most basis sense a <i>k</i>-nearest neighbor classifier. However, on
+            tasks like next-word prediction, <i>k</i>-NN becomes inhibitively slow. Fortunately,
+            TiMBL offers a number of fast approximations of <i>k</i>-NN classification, all
+            partly based on decision-tree classification and many orders of magnitude faster.
           </p>
           <p>
-            We show that <span class="dnerf">Nerfies</span> can turn casually captured selfie
-            photos/videos into deformable NeRF
-            models that allow for photorealistic renderings of the subject from arbitrary
-            viewpoints, which we dub <i>"nerfies"</i>. We evaluate our method by collecting data
-            using a
-            rig with two mobile phones that take time-synchronized photos, yielding train/validation
-            images of the same pose at different viewpoints. We show that our method faithfully
-            reconstructs non-rigidly deforming scenes and reproduces unseen views with high
-            fidelity.
+            Compared to Transformer-based LLMs, on the plus side memory-based LLMs are
           </p>
+          <ul>
+            <li>very efficient in training. Training is essentially reading the data (in linear time)
+              and compressing it into a decision tree structure. This can be done on CPUs,
+                with sufficient RAM;</li>
+            <li>pretty efficient in generation when running with the fastest decision-tree
+              approximations of <i>k</i>-NN classification. This can be done on CPUs as well.
+              Accuracy is traded for speed, however.</li>
+          </ul>
+          <p>On the downside,</p>
+          <ul>
+            <li>Memory requirements during training are heavy with large datasets (>100 million words);</li>
+            <li>Memory-based LLMs are not efficient at generation time when running relatively
+            slower approximations of <i>k</i>-NN classifiers, trading speed for accuracy.</li>
+          </ul>
         </div>
       </div>
     </div>