llama-cpp-sys-4 0.2.29

Low Level Bindings to llama.cpp
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
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135

[package]
name = "llama-cpp-sys-4"
description = "Low Level Bindings to llama.cpp"
version = "0.2.29"
edition = "2021"
license = "MIT OR Apache-2.0"
repository = "https://github.com/eugenehp/llama-cpp-rs"
keywords = ["llama", "llm", "ggml", "gguf", "inference"]
categories = ["api-bindings", "external-ffi-bindings"]
links = "llama"

include = [
    # Crate sources
    "build.rs",
    "wrapper.h",
    "/src",

    # Feature patches (applied at build time to the llama.cpp copy in OUT_DIR)
    "/patches",

    # test files touched by the q1 patch
    "/llama.cpp/tests/test-quantize-fns.cpp",

    # llama.cpp top-level
    "/llama.cpp/CMakeLists.txt",

    # llama.cpp cmake
    "/llama.cpp/cmake",

    # llama.cpp headers & sources
    "/llama.cpp/include/llama.h",
    "/llama.cpp/include/llama-cpp.h",
    "/llama.cpp/src/**/*",
    "/llama.cpp/common/**/*",

    # ggml cmake
    "/llama.cpp/ggml/CMakeLists.txt",
    "/llama.cpp/ggml/cmake",
    "/llama.cpp/ggml/src/CMakeLists.txt",

    # ggml headers & sources
    "/llama.cpp/ggml/include/*.h",
    "/llama.cpp/ggml/src/*.h",
    "/llama.cpp/ggml/src/*.c",
    "/llama.cpp/ggml/src/*.cpp",

    # ggml legacy top-level backend files
    "/llama.cpp/ggml/src/ggml-cuda.cu",
    "/llama.cpp/ggml/src/ggml-metal.m",
    "/llama.cpp/ggml/src/ggml-metal.metal",

    # ggml backends (ggml-kompute intentionally omitted — its only content is
    # an uninitialised nested submodule with no git-tracked files, which causes
    # `cargo publish` to abort with "uncommitted changes")
    "/llama.cpp/ggml/src/ggml-blas/**/*",
    "/llama.cpp/ggml/src/ggml-cann/**/*",
    "/llama.cpp/ggml/src/ggml-cpu/**/*",
    "/llama.cpp/ggml/src/ggml-cuda/**/*",
    "/llama.cpp/ggml/src/ggml-hexagon/**/*",
    "/llama.cpp/ggml/src/ggml-hip/**/*",
    "/llama.cpp/ggml/src/ggml-metal/**/*",
    "/llama.cpp/ggml/src/ggml-musa/**/*",
    "/llama.cpp/ggml/src/ggml-opencl/**/*",
    "/llama.cpp/ggml/src/ggml-rpc/**/*",
    "/llama.cpp/ggml/src/ggml-sycl/**/*",
    "/llama.cpp/ggml/src/ggml-virtgpu/**/*",
    "/llama.cpp/ggml/src/ggml-vulkan/**/*",
    "/llama.cpp/ggml/src/ggml-webgpu/**/*",
    "/llama.cpp/ggml/src/ggml-zdnn/**/*",
    "/llama.cpp/ggml/src/ggml-zendnn/**/*",

    # ggml misc
    "/llama.cpp/ggml/src/llamafile/sgemm.h",
    "/llama.cpp/ggml/src/llamafile/sgemm.cpp",
    "/llama.cpp/ggml/src/vulkan-shaders/**/*",

    # tools (tools/CMakeLists.txt unconditionally add_subdirectory()s all of
    # these, so every subdirectory must be present even when only mtmd is
    # needed; otherwise cmake exits 1 with "not an existing directory" errors)
    "/llama.cpp/tools/CMakeLists.txt",
    # mtmd tool — full sources needed for multimodal support
    "/llama.cpp/tools/mtmd/**/*",
    # quantize tool — needed by the q1 patch
    "/llama.cpp/tools/quantize/**/*",
    # Other tools — only CMakeLists.txt stubs so cmake doesn't error on
    # add_subdirectory(); their full sources are NOT shipped to save ~8 MB.
    "/llama.cpp/tools/batched-bench/CMakeLists.txt",
    "/llama.cpp/tools/cli/CMakeLists.txt",
    "/llama.cpp/tools/completion/CMakeLists.txt",
    "/llama.cpp/tools/cvector-generator/CMakeLists.txt",
    "/llama.cpp/tools/export-lora/CMakeLists.txt",
    "/llama.cpp/tools/fit-params/CMakeLists.txt",
    "/llama.cpp/tools/gguf-split/CMakeLists.txt",
    "/llama.cpp/tools/imatrix/CMakeLists.txt",
    "/llama.cpp/tools/llama-bench/CMakeLists.txt",
    "/llama.cpp/tools/parser/CMakeLists.txt",
    "/llama.cpp/tools/perplexity/CMakeLists.txt",
    "/llama.cpp/tools/results/CMakeLists.txt",
    "/llama.cpp/tools/rpc/CMakeLists.txt",
    "/llama.cpp/tools/server/CMakeLists.txt",
    "/llama.cpp/tools/tokenize/CMakeLists.txt",
    "/llama.cpp/tools/tts/CMakeLists.txt",

    # vendor
    "/llama.cpp/vendor/**/*",
]

# See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

[dependencies]

[build-dependencies]
bindgen = { workspace = true }
cc = { workspace = true, features = ["parallel"] }
cmake = "0.1"
glob = "0.3.3"

[target.'cfg(windows)'.build-dependencies]
winreg = "0.56"

[features]
cuda = []
metal = []
dynamic-link = []
vulkan = []
webgpu = []
blas = []
opencl = []
hip = []
native = []
openmp = []
mpi = []
rpc = []
mtmd = []
q1 = []