From 1949c95598536d6e7cdcb3e9ed99862c191dc9a0 Mon Sep 17 00:00:00 2001
From: Georgi Gerganov <ggerganov@gmail.com>
Date: Sun, 24 Dec 2023 15:33:31 +0200
Subject: [PATCH] plamo : fix tensor names for correct GPU offload

---
 llama.cpp | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/llama.cpp b/llama.cpp
index 90dc1b11f..03d1a4b66 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -5704,13 +5704,14 @@ struct llm_build_context {
                         model.layers[il].ffn_gate, NULL,
                         model.layers[il].ffn_down, NULL,
                         LLM_FFN_SILU, LLM_FFN_PAR, cb, il);
-                cb(cur, "mlp_out", il);
+                cb(cur, "ffn_out", il);
             }
 
             cur = ggml_add(ctx0, cur, sa_out);
-            cb(cur, "mlp_out + sa_out", il);
+            cb(cur, "l_out", il);
+
             cur = ggml_add(ctx0, cur, inpL);
-            cb(cur, "mlp_out + sa_out + inpL", il);
+            cb(cur, "l_out", il);
 
             // input for next layer
             inpL = cur;