From 1123f7fbdfb8012e46f05e903e6f675922916378 Mon Sep 17 00:00:00 2001
From: slaren <slarengh@gmail.com>
Date: Tue, 22 Aug 2023 15:25:19 +0200
Subject: ggml-cuda : use graph allocator (#2684)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

use a different function for no_alloc to avoid breaking backwards compat, fixes lora

remove 512 n_batch limit

fixed 2048 batch size

cleanup

Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
---
 common/common.cpp | 1 -
 1 file changed, 1 deletion(-)

(limited to 'common/common.cpp')

diff --git a/common/common.cpp b/common/common.cpp
index d7e1a572..1623ba21 100644
--- a/common/common.cpp
+++ b/common/common.cpp
@@ -289,7 +289,6 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
                 break;
             }
             params.n_batch = std::stoi(argv[i]);
-            params.n_batch = std::min(512, params.n_batch);
         } else if (arg == "--keep") {
             if (++i >= argc) {
                 invalid_param = true;
-- 
cgit v1.2.3