From 8c37a9478129f49fc8f8412edeaf5f85044b8644 Mon Sep 17 00:00:00 2001
From: Hansong Zhang <hsz@fb.com>
Date: Fri, 18 Apr 2025 17:19:36 -0700
Subject: [PATCH] [Android] Remove old onStats

---
 examples/demo-apps/android/LlamaDemo/README.md      |  4 ++--
 .../executorch/extension/llm/LlmCallback.java       | 13 +------------
 extension/android/jni/jni_layer_llama.cpp           |  8 --------
 3 files changed, 3 insertions(+), 22 deletions(-)

diff --git a/examples/demo-apps/android/LlamaDemo/README.md b/examples/demo-apps/android/LlamaDemo/README.md
index fae91857fbe..d4db18ec016 100644
--- a/examples/demo-apps/android/LlamaDemo/README.md
+++ b/examples/demo-apps/android/LlamaDemo/README.md
@@ -135,8 +135,8 @@ Ensure you have the following functions in your callback class that you provided
   }
 
   @Override
-  public void onStats(float tps) {
-    //...tps (tokens per second) stats is provided by framework
+  public void onStats(String stats) {
+    //... will be a json. See extension/llm/stats.h for the field definitions
   }
 
 ```
diff --git a/extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmCallback.java b/extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmCallback.java
index a829cf75dc4..639fd0812bd 100644
--- a/extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmCallback.java
+++ b/extension/android/executorch_android/src/main/java/org/pytorch/executorch/extension/llm/LlmCallback.java
@@ -28,17 +28,6 @@ public interface LlmCallback {
   @DoNotStrip
   public void onResult(String result);
 
-  /**
-   * Called when the statistics for the generate() is available.
-   *
-   * Note: This is a deprecated API and will be removed in the future. Please use onStats(String stats)
-   *
-   * @param tps Tokens/second for generated tokens.
-   */
-  @Deprecated
-  @DoNotStrip
-  default public void onStats(float tps) {}
-
   /**
    * Called when the statistics for the generate() is available.
    *
@@ -48,5 +37,5 @@ default public void onStats(float tps) {}
    * @param stats JSON string containing the statistics for the generate()
    */
   @DoNotStrip
-  default public void onStats(String stats) {}
+  default void onStats(String stats) {}
 }
diff --git a/extension/android/jni/jni_layer_llama.cpp b/extension/android/jni/jni_layer_llama.cpp
index 4a98ccb7c82..83ca1d898ed 100644
--- a/extension/android/jni/jni_layer_llama.cpp
+++ b/extension/android/jni/jni_layer_llama.cpp
@@ -100,14 +100,6 @@ class ExecuTorchLlmCallbackJni
 
   void onStats(const llm::Stats& result) const {
     static auto cls = ExecuTorchLlmCallbackJni::javaClassStatic();
-    static const auto tps_method = cls->getMethod<void(jfloat)>("onStats");
-    double eval_time =
-        (double)(result.inference_end_ms - result.prompt_eval_end_ms);
-
-    float tps = result.num_generated_tokens / eval_time *
-        result.SCALING_FACTOR_UNITS_PER_SECOND;
-    tps_method(self(), tps);
-
     static const auto on_stats_method =
         cls->getMethod<void(facebook::jni::local_ref<jstring>)>("onStats");
     on_stats_method(