From: Georgi Gerganov <redacted>
Date: Mon, 24 Apr 2023 15:47:03 +0000 (+0300)
Subject: llama : increase scratch buffer size for 65B (ref #1152)
X-Git-Tag: gguf-v0.4.0~875
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=957c8ae21d1e7052ea45a40ee8c0407b909e90cc;p=pkg%2Fggml%2Fsources%2Fllama.cpp

llama : increase scratch buffer size for 65B (ref #1152)

Temporary solution
---

diff --git a/llama.cpp b/llama.cpp
index bc0ef128..28d27916 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -54,7 +54,7 @@ static const std::map<e_model, size_t> & MEM_REQ_SCRATCH0()
         { MODEL_7B,    512ull * MB },
         { MODEL_13B,   512ull * MB },
         { MODEL_30B,   512ull * MB },
-        { MODEL_65B,   512ull * MB },
+        { MODEL_65B,  1024ull * MB },
     };
     return _MEM_REQ_SCRATCH0;
 }
@@ -65,7 +65,7 @@ static const std::map<e_model, size_t> & MEM_REQ_SCRATCH1()
         { MODEL_7B,    512ull * MB },
         { MODEL_13B,   512ull * MB },
         { MODEL_30B,   512ull * MB },
-        { MODEL_65B,   512ull * MB },
+        { MODEL_65B,  1024ull * MB },
     };
     return _MEM_REQ_SCRATCH1;
 }