transpose and run cont

Author: Lorenzo Toniazzi

_BRANCH_SETUP.md

@@ -342,4 +342,7 @@ int main() {
   -n 128
   ```
 
-  make clean && make -j 8 LLAMA_DEBUG=1
+build for debug:
+```bash
+  make clean && make -j 8 LLAMA_DEBUG=1
+```

llama.cpp

@@ -9733,7 +9733,7 @@ struct llm_build_context {
     ggml_tensor * t_lora = ggml_mul_mat(ctx0,
                 loraA,
                 ggml_mul_mat(ctx0, 
-                    ggml_transpose(ctx0, loraB), 
+                    ggml_cont(ctx0, ggml_transpose(ctx0, loraB)),
                     cur
                 )
             );