update llama-cpp

This commit is contained in:
2025-12-03 21:06:50 -05:00
parent 0ed577e815
commit 50b03508e1

View File

@@ -23,12 +23,12 @@ in
metalSupport = false; metalSupport = false;
}).overrideAttrs }).overrideAttrs
(oldAttrs: rec { (oldAttrs: rec {
version = "7213"; version = "7253";
src = pkgs.fetchFromGitHub { src = pkgs.fetchFromGitHub {
owner = "ggml-org"; owner = "ggml-org";
repo = "llama.cpp"; repo = "llama.cpp";
tag = "b${version}"; tag = "b${version}";
hash = "sha256-DVP1eGuSbKd0KxmrDCVAdOe16r1wJ9+RonbrUTgVS7U="; hash = "sha256-Gx8c00mwh/ySHDbjqCPu7nKymb24gCB/NHMGjo4FS08=";
leaveDotGit = true; leaveDotGit = true;
postFetch = '' postFetch = ''
git -C "$out" rev-parse --short HEAD > $out/COMMIT git -C "$out" rev-parse --short HEAD > $out/COMMIT
@@ -146,7 +146,7 @@ in
# https://huggingface.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/tree/main # https://huggingface.co/unsloth/Qwen3-Coder-30B-A3B-Instruct-GGUF/tree/main
"qwen3-coder-30b-instruct" = { "qwen3-coder-30b-instruct" = {
name = "Qwen3 Coder (30B) - Instruct"; name = "Qwen3 Coder (30B) - Instruct";
cmd = "${pkgs.llama-cpp}/bin/llama-server --port \${PORT} -m /mnt/ssd/Models/Qwen3-Coder-30B-A3B-Instruct-Q4_K_M.gguf --ctx-size 16384 --temp 0.7 --min-p 0.0 --top-p 0.8 --top-k 20 --cache-type-k q4_0 --cache-type-v q4_0"; cmd = "${pkgs.llama-cpp}/bin/llama-server --port \${PORT} -m /mnt/ssd/Models/Qwen3-Coder-30B-A3B-Instruct-UD-Q4_K_XL.gguf --ctx-size 16384 --temp 0.7 --min-p 0.0 --top-p 0.8 --top-k 20";
}; };
# https://huggingface.co/unsloth/Qwen3-30B-A3B-Instruct-2507-GGUF/tree/main # https://huggingface.co/unsloth/Qwen3-30B-A3B-Instruct-2507-GGUF/tree/main