fix links

docs/docs.html

@@ -202,7 +202,7 @@ <h4>Key Features</h4>
                     <li><strong>37</strong> traditional FL (tFL) or personalized FL (pFL) algorithms,  <strong>3</strong> scenarios, and <strong>24</strong> datasets.</li>
                     <li>Some experimental results are avalible in the <a href="https://arxiv.org/abs/2312.04992"><strong>PFLlib paper</strong></a> and <a href="benchmark.html"><strong>Benchmark Results</strong></a>.</li>
                     <li>The benchmark platform can simulate scenarios using the 4-layer CNN on Cifar100 for <strong>500 clients</strong> on one NVIDIA GeForce RTX 3090 GPU card with <strong>only 5.08GB GPU memory cost</strong>.</li>
-                    <li>We provide <a href="#privacy-evaluation">privacy evaluation</a> and <a href="#systematical-research-supprot">systematical research support</a>.</li>
+                    <li>We provide <a href="features.html#privacy-evaluation">privacy evaluation</a> and <a href="features.html#systematical-research-supprot">systematical research support</a>.</li>
                     <li>You can now train on some clients and evaluate performance on new clients by setting <code>args.num_new_clients</code> in <code>./system/main.py</code>. Please note that not all tFL/pFL algorithms support this feature.</li>
                     <li>PFLlib primarily focuses on data (statistical) heterogeneity. For algorithms and a benchmark platform that address <strong>both data and model heterogeneity</strong>, please refer to our extended project <a href="https://github.com/TsingZ0/HtFLlib"><strong>HtFLlib</strong></a>.</li>
                     <li>As we strive to meet diverse user demands, frequent updates to the project may alter default settings and scenario creation codes, affecting experimental results.</li>

docs/features.html

@@ -175,7 +175,7 @@ <h1><a href="index.html"></a>PFLlib</a></h1>
         </div>
         <div class="content">
             <section id="features">
-                <h2>Privacy Evaluation</h2>
+                <h2 id="privacy-evaluation">Privacy Evaluation</h2>
                 <p>You can use the following privacy evaluation methods to assess the privacy-preserving capabilities of tFL/pFL algorithms in PFLlib. Please refer to <code>./system/flcore/servers/serveravg.py</code> for an example. Note that most of these evaluations are not typically considered in the original papers. <em>We encourage you to add more attacks and metrics for privacy evaluation.</em></p>
                 <h4>Currently supported attacks:</h4>
                 <ul>
@@ -186,7 +186,7 @@ <h4>Currently supported metrics:</h4>
                     <li><strong>PSNR (Peak Signal-to-Noise Ratio)</strong>: an objective metric for image evaluation, defined as the logarithm of the ratio of the squared maximum value of RGB image fluctuations to the Mean Squared Error (MSE) between two images. A lower PSNR score indicates better privacy-preserving capabilities.</li>
                 </ul>
 
-                <h2>Systematical research support</h2>
+                <h2 id="systematical-research-supprot">Systematical research support</h2>
                 <p>To simulate Federated Learning (FL) under practical conditions, such as <strong>client dropout</strong>, <strong>slow trainers</strong>, <strong>slow senders</strong>, and <strong>network TTL (Time-To-Live)</strong>, you can adjust the following parameters:</p>
                 <ul>
                 <li><code>-cdr</code>: Dropout rate for clients. Clients are randomly dropped at each training round based on this rate.</li>