gregH commited on
Commit
c9fcacb
1 Parent(s): 7979584

Update index.html

Browse files
Files changed (1) hide show
  1. index.html +3 -3
index.html CHANGED
@@ -134,15 +134,15 @@ in this demonstration. Multiple Bin numbers are also support</p>
134
  <div class="legend"><img src="images/demo-legend.png" alt="legend" /></div>
135
  </div>
136
  <div class="row align-items-center">
137
- <div class="attack-success-rate"><span class="jailbreak-metric">Average Malicious Refusal Rate</span><span class="attack-success-rate-value" id="asr-value">0.10731</span></div>
138
  </div>
139
  <div class="row align-items-center">
140
- <div class="benign-refusal-rate"><span class="jailbreak-metric">Benign Refusal Rate</span><span class="benign-refusal-rate-value" id="brr-value">0.10721</span></div>
141
  </div>
142
  </div>
143
  <div class="col-8">
144
  <figure class="figure">
145
- <img id="reliability-diagram" src="images/cifar100/resnet110/none/bin15.png" alt="CIFAR-100 Calibrated Reliability Diagram (Full)" />
146
  <div class="slider-container">
147
  <div class="slider-label"><span>Perplexity Threshold</span></div>
148
  <div class="slider-content" id="ppl-slider"><div id="ppl-threshold" class="ui-slider-handle"></div></div>
 
134
  <div class="legend"><img src="images/demo-legend.png" alt="legend" /></div>
135
  </div>
136
  <div class="row align-items-center">
137
+ <div class="attack-success-rate"><span class="jailbreak-metric">Average Malicious Refusal Rate</span><span class="attack-success-rate-value" id="asr-value">0.95875</span></div>
138
  </div>
139
  <div class="row align-items-center">
140
+ <div class="benign-refusal-rate"><span class="jailbreak-metric">Benign Refusal Rate</span><span class="benign-refusal-rate-value" id="brr-value">0.05000</span></div>
141
  </div>
142
  </div>
143
  <div class="col-8">
144
  <figure class="figure">
145
+ <img id="reliability-diagram" src="demo_results/gradient_cuff_llama2_7b_chat_threshold_100.png" alt="CIFAR-100 Calibrated Reliability Diagram (Full)" />
146
  <div class="slider-container">
147
  <div class="slider-label"><span>Perplexity Threshold</span></div>
148
  <div class="slider-content" id="ppl-slider"><div id="ppl-threshold" class="ui-slider-handle"></div></div>