Adding support to demo prompt classification with Llama Guard (#5595)

Adding support to demo prompt classification with Llama Guard (#5553)

Summary:
Pull Request resolved: #5553

Adding support to load Llama Guard model and run prompt classification task

Reviewed By: cmodi-meta, kirklandsign

Differential Revision: D63148252

fbshipit-source-id: 482559e694da05bdec75b9a2dbd76163c686e47d
(cherry picked from commit 61cb5b0)

Co-authored-by: Riandy Riandy <[email protected]>

Author: pytorchbot

examples/demo-apps/android/LlamaDemo/app/src/main/java/com/example/executorchllamademo/MainActivity.java

@@ -704,6 +704,16 @@ public void run() {
                         startPos,
                         MainActivity.this,
                         false);
+                  } else if (mCurrentSettingsFields.getModelType() == ModelType.LLAMA_GUARD_3) {
+                    String llamaGuardPromptForClassification =
+                        PromptFormat.getFormattedLlamaGuardPrompt(rawPrompt);
+                    ETLogging.getInstance()
+                        .log("Running inference.. prompt=" + llamaGuardPromptForClassification);
+                    mModule.generate(
+                        llamaGuardPromptForClassification,
+                        llamaGuardPromptForClassification.length() + 64,
+                        MainActivity.this,
+                        false);
                   } else {
                     ETLogging.getInstance().log("Running inference.. prompt=" + finalPrompt);
                     mModule.generate(

examples/demo-apps/android/LlamaDemo/app/src/main/java/com/example/executorchllamademo/ModelType.java

@@ -12,4 +12,5 @@ public enum ModelType {
   LLAMA_3,
   LLAMA_3_1,
   LLAVA_1_5,
+  LLAMA_GUARD_3,
 }

examples/demo-apps/android/LlamaDemo/app/src/main/java/com/example/executorchllamademo/PromptFormat.java

@@ -33,6 +33,7 @@ public static String getUserPromptTemplate(ModelType modelType) {
     switch (modelType) {
       case LLAMA_3:
       case LLAMA_3_1:
+      case LLAMA_GUARD_3:
         return "<|start_header_id|>user<|end_header_id|>\n"
             + USER_PLACEHOLDER
             + "<|eot_id|>"
@@ -60,6 +61,7 @@ public static String getStopToken(ModelType modelType) {
     switch (modelType) {
       case LLAMA_3:
       case LLAMA_3_1:
+      case LLAMA_GUARD_3:
         return "<|eot_id|>";
       case LLAVA_1_5:
         return "</s>";
@@ -72,4 +74,44 @@ public static String getLlavaPresetPrompt() {
     return "A chat between a curious human and an artificial intelligence assistant. The assistant"
         + " gives helpful, detailed, and polite answers to the human's questions. USER: ";
   }
+
+  public static String getFormattedLlamaGuardPrompt(String userPrompt) {
+    return getUserPromptTemplate(ModelType.LLAMA_GUARD_3)
+        .replace(
+            USER_PLACEHOLDER, getLlamaGuardPresetPrompt().replace(USER_PLACEHOLDER, userPrompt));
+  }
+
+  public static String getLlamaGuardPresetPrompt() {
+    return "Task: Check if there is unsafe content in 'User' messages in conversations"
+        + " according our safety policy with the below categories.\n"
+        + "\n"
+        + "<BEGIN UNSAFE CONTENT CATEGORIES>\n"
+        + "S1: Violent Crimes. \n"
+        + "S2: Non-Violent Crimes. \n"
+        + "S3: Sex Crimes. \n"
+        + "S4: Child Exploitation. \n"
+        + "S5: Defamation. \n"
+        + "S6: Specialized Advice. \n"
+        + "S7: Privacy. \n"
+        + "S8: Intellectual Property. \n"
+        + "S9: Indiscriminate Weapons. \n"
+        + "S10: Hate. \n"
+        + "S11: Self-Harm. \n"
+        + "S12: Sexual Content. \n"
+        + "S13: Elections. \n"
+        + "S14: Code Interpreter Abuse.\n"
+        + "<END UNSAFE CONTENT CATEGORIES>\n"
+        + "\n"
+        + "<BEGIN CONVERSATION>\n"
+        + "\n"
+        + "User: "
+        + USER_PLACEHOLDER
+        + "\n"
+        + "<END CONVERSATION>\n"
+        + "Provide your safety assessment for ONLY THE LAST User message in the above"
+        + " conversation:\n"
+        + " - First line must read 'safe' or 'unsafe'.\n"
+        + " - If unsafe, a second line must include a comma-separated list of violated"
+        + " categories.";
+  }
 }