ollama/llama/patches/0020-ollama-debug-tensor.patch
2025-03-11 14:49:19 -07:00

33 lines
1.1 KiB
Diff

From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
From: Michael Yang <mxyng@pm.me>
Date: Sun, 9 Mar 2025 14:44:16 -0700
Subject: [PATCH] ollama debug tensor
---
ggml/src/ggml-cpu/ggml-cpu.c | 6 ++++++
1 file changed, 6 insertions(+)
diff --git a/ggml/src/ggml-cpu/ggml-cpu.c b/ggml/src/ggml-cpu/ggml-cpu.c
index 2f606d82..ec60e8fc 100644
--- a/ggml/src/ggml-cpu/ggml-cpu.c
+++ b/ggml/src/ggml-cpu/ggml-cpu.c
@@ -11,6 +11,8 @@
#include "ggml-threading.h"
#include "ggml.h"
+#include "ollama-debug.h"
+
#if defined(_MSC_VER) || defined(__MINGW32__)
#include <malloc.h> // using malloc.h with MSC/MINGW
#elif !defined(__FreeBSD__) && !defined(__NetBSD__) && !defined(__OpenBSD__)
@@ -14103,6 +14105,10 @@ static thread_ret_t ggml_graph_compute_thread(void * data) {
ggml_compute_forward(&params, node);
+#ifdef OLLAMA_DEBUG
+ ollama_debug(node, true);
+#endif
+
if (state->ith == 0 && cplan->abort_callback &&
cplan->abort_callback(cplan->abort_callback_data)) {
atomic_store_explicit(&tp->abort, node_n + 1, memory_order_relaxed);