blob: d2e10b29799f4316cd159a77eb057e96ff7da0c0 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
|
pkgbase = llama.cpp-cuda-f16
pkgdesc = Port of Facebook's LLaMA model in C/C++ (with NVIDIA CUDA optimizations and F16)
pkgver = b5929
pkgrel = 1
url = https://github.com/ggerganov/llama.cpp
arch = x86_64
arch = armv7h
arch = aarch64
license = MIT
makedepends = cmake
makedepends = git
depends = cuda
depends = curl
depends = gcc-libs
depends = glibc
depends = nvidia-utils
depends = python
depends = python-numpy
depends = python-sentencepiece
optdepends = python-pytorch
provides = llama.cpp
conflicts = llama.cpp
conflicts = libggml
conflicts = ggml
options = lto
options = !debug
source = git+https://github.com/ggerganov/llama.cpp#tag=b5929
source = llama.cpp.conf
source = llama.cpp.service
sha256sums = 0c24d9ca1f9370e34ee451d407a8d2b2beb6872b9caf1471dd7ac1d772677d23
sha256sums = 53fa70cfe40cb8a3ca432590e4f76561df0f129a31b121c9b4b34af0da7c4d87
sha256sums = 0377d08a07bda056785981d3352ccd2dbc0387c4836f91fb73e6b790d836620d
pkgname = llama.cpp-cuda-f16
|