From: Georgi Gerganov <redacted>
Date: Wed, 2 Apr 2025 13:38:54 +0000 (+0300)
Subject: model : print tensor size during load (#12711)
X-Git-Tag: gguf-v0.16.1~125
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=833e2b7409211a07df97716998c5002526642652;p=pkg%2Fggml%2Fsources%2Fllama.cpp

model : print tensor size during load (#12711)

* model : print tensor size during load

* cont : fix units MB -> MiB

Co-authored-by: Diego Devesa <redacted>

---------

Co-authored-by: Diego Devesa <redacted>
---

diff --git a/src/llama-model-loader.cpp b/src/llama-model-loader.cpp
index ec1d78e3..ea73a8a7 100644
--- a/src/llama-model-loader.cpp
+++ b/src/llama-model-loader.cpp
@@ -603,7 +603,9 @@ llama_model_loader::llama_model_loader(
 
             if (trace > 0) {
                 const uint16_t sid = w.idx;
-                LLAMA_LOG_INFO("%s: - tensor split %2d: %32s %-8s [ %s ]\n", __func__, sid, ggml_get_name(tensor), ggml_type_name(type), llama_format_tensor_shape(tensor).c_str());
+                LLAMA_LOG_INFO("%s: - tensor split %2d: %32s %-8s [ %s ] %8.2f MiB\n", __func__,
+                        sid, ggml_get_name(tensor), ggml_type_name(type), llama_format_tensor_shape(tensor).c_str(),
+                        ggml_nbytes(tensor)/1024.0f/1024.0f);
             }
         }