Fixed incorrectly applying RMS norm twice (#1925)

Add steps for using termux on android devices to prevent common errors.

README.md

@@ -617,7 +617,12 @@ And after 4.45 hours, you will have the final perplexity.
 
 #### Building the Project using Android NDK
 You can easily run `llama.cpp` on Android device with [termux](https://termux.dev/).
-First, obtain the [Android NDK](https://developer.android.com/ndk) and then build with CMake:
+
+First, install the essential packages for termux:
+```
+pkg install clang wget git cmake
+```
+Second, obtain the [Android NDK](https://developer.android.com/ndk) and then build with CMake:
 ```
 $ mkdir build-android
 $ cd build-android

ggml.c

@@ -7918,7 +7918,7 @@ static void ggml_compute_forward_add_q_f32(
 
         void  * src0_row = (void *) ((char *) src0->data + (i01*nb01 + i02*nb02 + i03*nb03));
         float * src1_row = (float *)((char *) src1->data + (i11*nb11 + i12*nb12 + i13*nb13));
-        void  * dst_row  = (void *) ((char *)  dst->data + ( i1*nb1  +  i2*nb2  +  i3*nb0));
+        void  * dst_row  = (void *) ((char *)  dst->data + ( i1*nb1  +  i2*nb2  +  i3*nb3));
 
         assert(ne00 % 32 == 0);
 

llama.cpp

@@ -1657,11 +1657,7 @@ static bool llama_eval_internal(
     {
         cur = ggml_rms_norm(ctx0, inpL);
         offload_func_nr(cur);
-        ggml_set_name(cur, "rms_norm_inpL");
-
-        cur = ggml_rms_norm(ctx0, cur);
-        offload_func_nr(cur);
-        ggml_set_name(cur, "rms_norm_after");
+        ggml_set_name(cur, "rms_norm_2");
 
         // cur = cur*norm(broadcasted)
         cur = ggml_mul(ctx0, cur, model.norm);