llama_cpp 0.15.4 → 0.16.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- checksums.yaml +4 -4
- data/CHANGELOG.md +16 -0
- data/ext/llama_cpp/extconf.rb +3 -2
- data/ext/llama_cpp/llama_cpp.cpp +17 -3
- data/lib/llama_cpp/version.rb +2 -2
- data/sig/llama_cpp.rbs +15 -1
- data/vendor/tmp/llama.cpp/Makefile +166 -82
- data/vendor/tmp/llama.cpp/ggml-alloc.c +82 -26
- data/vendor/tmp/llama.cpp/ggml-backend-impl.h +20 -8
- data/vendor/tmp/llama.cpp/ggml-backend.c +183 -69
- data/vendor/tmp/llama.cpp/ggml-backend.h +4 -4
- data/vendor/tmp/llama.cpp/ggml-blas.cpp +363 -0
- data/vendor/tmp/llama.cpp/ggml-blas.h +23 -0
- data/vendor/tmp/llama.cpp/ggml-common.h +6 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/acc.cu +47 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/arange.cu +34 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/argsort.cu +104 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/binbcast.cu +280 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/clamp.cu +34 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/concat.cu +196 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/convert.cu +686 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/cpy.cu +490 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/diagmask.cu +40 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/dmmv.cu +674 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/fattn-tile-f16.cu +319 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/fattn-tile-f32.cu +312 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/fattn.cu +345 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/getrows.cu +178 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/im2col.cu +104 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/mmq.cu +88 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/mmvq.cu +419 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/norm.cu +221 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/pad.cu +49 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/pool2d.cu +94 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/quantize.cu +112 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/rope.cu +271 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/scale.cu +31 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/softmax.cu +206 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/sumrows.cu +40 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb16.cu +10 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqfloat-cpb32.cu +9 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb16.cu +10 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb32.cu +10 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/fattn-wmma-f16-instance-kqhalf-cpb8.cu +8 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q2_k.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q3_k.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q4_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q4_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q4_k.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q5_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q5_1.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q5_k.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q6_k.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/template-instances/mmq-instance-q8_0.cu +5 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/tsembd.cu +47 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/unary.cu +286 -0
- data/vendor/tmp/llama.cpp/ggml-cuda/upscale.cu +51 -0
- data/vendor/tmp/llama.cpp/ggml-cuda.cu +103 -135
- data/vendor/tmp/llama.cpp/ggml-kompute.cpp +29 -13
- data/vendor/tmp/llama.cpp/ggml-metal.h +1 -1
- data/vendor/tmp/llama.cpp/ggml-metal.m +45 -33
- data/vendor/tmp/llama.cpp/ggml-metal.metal +83 -59
- data/vendor/tmp/llama.cpp/ggml-rpc.cpp +15 -14
- data/vendor/tmp/llama.cpp/ggml-sycl.cpp +26 -90
- data/vendor/tmp/llama.cpp/ggml-vulkan-shaders.hpp +74522 -14913
- data/vendor/tmp/llama.cpp/ggml-vulkan.cpp +631 -471
- data/vendor/tmp/llama.cpp/ggml.c +278 -603
- data/vendor/tmp/llama.cpp/ggml.h +9 -28
- data/vendor/tmp/llama.cpp/llama.cpp +345 -473
- data/vendor/tmp/llama.cpp/llama.h +21 -43
- metadata +134 -7
- data/vendor/tmp/llama.cpp/ggml-mpi.c +0 -216
- data/vendor/tmp/llama.cpp/ggml-mpi.h +0 -39
- data/vendor/tmp/llama.cpp/ggml-opencl.cpp +0 -2305
- data/vendor/tmp/llama.cpp/ggml-opencl.h +0 -36
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: '0408c7d579262d0ba34013671a446a76513f6f4564270ef8ba471203fba75c59'
|
|
4
|
+
data.tar.gz: a8085d9678999bb79ba788f7ce856c3f4fe1c6e131af569eaa54aa71fc9ae773
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 32e8f294a7f88db05abba3d1e11c951a38f366cac83712f89aa68ed95c581d8eaa4df3d5473f1af3cee965d7a66ea2bf5ccb00222337c59d97fca32ba5e9cade
|
|
7
|
+
data.tar.gz: 2c5f66e2902eb1d72e45261e80f1a2599534e02d0e54e83ff432414d355fab67f6bda9eee095f17904a4fb2d3bf5cb4a1509a346fadff13d999c2f057db972c6
|
data/CHANGELOG.md
CHANGED
|
@@ -1,3 +1,19 @@
|
|
|
1
|
+
## [[0.16.1](https://github.com/yoshoku/llama_cpp.rb/compare/v0.16.0...v0.16.1)] - 2024-06-15
|
|
2
|
+
|
|
3
|
+
- Bump llama.cpp from b3091 to b3151.
|
|
4
|
+
- Add `--with-openblas641` and `--with-no-llamafile` config options.
|
|
5
|
+
- Add `LLAMA_VOCAB_PRE_TYPE_PORO` and `LLAMA_GRETYPE_CHAR_ANY` constants.
|
|
6
|
+
|
|
7
|
+
## [[0.16.0](https://github.com/yoshoku/llama_cpp.rb/compare/v0.15.4...v0.16.0)] - 2024-06-08
|
|
8
|
+
|
|
9
|
+
**Breaking Changes**
|
|
10
|
+
|
|
11
|
+
- Bump llama.cpp from b3056 to b3091.
|
|
12
|
+
- Rename `type` method to `token_attr` in `Model`.
|
|
13
|
+
- Add constants for token attribute types.
|
|
14
|
+
- Remove `--with-clblast` and `--with-mpi` config options.
|
|
15
|
+
- Add `--with-no-openmp` config option.
|
|
16
|
+
|
|
1
17
|
## [[0.15.4](https://github.com/yoshoku/llama_cpp.rb/compare/v0.15.3...v0.15.4)] - 2024-06-01
|
|
2
18
|
|
|
3
19
|
- Bump llama.cpp from b2988 to b3056.
|
data/ext/llama_cpp/extconf.rb
CHANGED
|
@@ -14,13 +14,14 @@ make_envs << ' LLAMA_DEBUG=1' if with_config('debug')
|
|
|
14
14
|
make_envs << ' LLAMA_QKK_64=1' if with_config('qkk-64')
|
|
15
15
|
make_envs << ' LLAMA_NO_ACCELERATE=1' if with_config('no-accelerate')
|
|
16
16
|
make_envs << ' LLAMA_OPENBLAS=1' if with_config('openblas')
|
|
17
|
+
make_envs << ' LLAMA_OPENBLAS64=1' if with_config('openblas64')
|
|
17
18
|
make_envs << ' LLAMA_BLIS=1' if with_config('blis')
|
|
18
19
|
make_envs << ' LLAMA_CUBLAS=1' if with_config('cublas') # Deprecated, use --with-cuda instead
|
|
19
20
|
make_envs << ' LLAMA_CUDA=1' if with_config('cuda')
|
|
20
|
-
make_envs << ' LLAMA_CLBLAST=1' if with_config('clblast')
|
|
21
21
|
make_envs << ' LLAMA_HIPBLAS=1' if with_config('hipblas')
|
|
22
|
-
make_envs << ' LLAMA_MPI=1' if with_config('mpi')
|
|
23
22
|
make_envs << ' LLAMA_VULKAN=1' if with_config('vulkan')
|
|
23
|
+
make_envs << ' LLAMA_NO_OPENMP=1' if with_config('no-openmp')
|
|
24
|
+
make_envs << ' LLAMA_NO_LLAMAFILE=1' if with_config('no-llamafile')
|
|
24
25
|
|
|
25
26
|
make_envs << ' LLAMA_METAL_EMBED_LIBRARY=1' if RUBY_PLATFORM.match?(/darwin/)
|
|
26
27
|
|
data/ext/llama_cpp/llama_cpp.cpp
CHANGED
|
@@ -1523,7 +1523,7 @@ public:
|
|
|
1523
1523
|
rb_define_method(rb_cLLaMAModel, "n_params", RUBY_METHOD_FUNC(_llama_model_get_model_n_params), 0);
|
|
1524
1524
|
rb_define_method(rb_cLLaMAModel, "text", RUBY_METHOD_FUNC(_llama_model_get_text), 1);
|
|
1525
1525
|
rb_define_method(rb_cLLaMAModel, "score", RUBY_METHOD_FUNC(_llama_model_get_score), 1);
|
|
1526
|
-
rb_define_method(rb_cLLaMAModel, "
|
|
1526
|
+
rb_define_method(rb_cLLaMAModel, "token_attr", RUBY_METHOD_FUNC(_llama_model_get_token_attr), 1);
|
|
1527
1527
|
rb_define_method(rb_cLLaMAModel, "token_bos", RUBY_METHOD_FUNC(_llama_model_token_bos), 0);
|
|
1528
1528
|
rb_define_method(rb_cLLaMAModel, "token_eos", RUBY_METHOD_FUNC(_llama_model_token_eos), 0);
|
|
1529
1529
|
rb_define_method(rb_cLLaMAModel, "token_cls", RUBY_METHOD_FUNC(_llama_model_token_cls), 0);
|
|
@@ -1778,10 +1778,10 @@ private:
|
|
|
1778
1778
|
return DBL2NUM(score);
|
|
1779
1779
|
}
|
|
1780
1780
|
|
|
1781
|
-
static VALUE
|
|
1781
|
+
static VALUE _llama_model_get_token_attr(VALUE self, VALUE token_) {
|
|
1782
1782
|
LLaMAModelWrapper* ptr = get_llama_model(self);
|
|
1783
1783
|
const llama_token token = NUM2INT(token_);
|
|
1784
|
-
const
|
|
1784
|
+
const llama_token_attr type = llama_token_get_attr(ptr->model, token);
|
|
1785
1785
|
return INT2NUM(type);
|
|
1786
1786
|
}
|
|
1787
1787
|
|
|
@@ -3494,6 +3494,7 @@ extern "C" void Init_llama_cpp(void) {
|
|
|
3494
3494
|
rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_OLMO", INT2NUM(LLAMA_VOCAB_PRE_TYPE_OLMO));
|
|
3495
3495
|
rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_DBRX", INT2NUM(LLAMA_VOCAB_PRE_TYPE_DBRX));
|
|
3496
3496
|
rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_SMAUG", INT2NUM(LLAMA_VOCAB_PRE_TYPE_SMAUG));
|
|
3497
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_VOCAB_PRE_TYPE_PORO", INT2NUM(LLAMA_VOCAB_PRE_TYPE_PORO));
|
|
3497
3498
|
|
|
3498
3499
|
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_TYPE_UNDEFINED", INT2NUM(LLAMA_TOKEN_TYPE_UNDEFINED));
|
|
3499
3500
|
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_TYPE_NORMAL", INT2NUM(LLAMA_TOKEN_TYPE_NORMAL));
|
|
@@ -3503,6 +3504,18 @@ extern "C" void Init_llama_cpp(void) {
|
|
|
3503
3504
|
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_TYPE_UNUSED", INT2NUM(LLAMA_TOKEN_TYPE_UNUSED));
|
|
3504
3505
|
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_TYPE_BYTE", INT2NUM(LLAMA_TOKEN_TYPE_BYTE));
|
|
3505
3506
|
|
|
3507
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_UNDEFINED", INT2NUM(LLAMA_TOKEN_ATTR_UNDEFINED));
|
|
3508
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_UNKNOWN", INT2NUM(LLAMA_TOKEN_ATTR_UNKNOWN));
|
|
3509
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_UNUSED", INT2NUM(LLAMA_TOKEN_ATTR_UNUSED));
|
|
3510
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_NORMAL", INT2NUM(LLAMA_TOKEN_ATTR_NORMAL));
|
|
3511
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_CONTROL", INT2NUM(LLAMA_TOKEN_ATTR_CONTROL));
|
|
3512
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_USER_DEFINED", INT2NUM(LLAMA_TOKEN_ATTR_USER_DEFINED));
|
|
3513
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_BYTE", INT2NUM(LLAMA_TOKEN_ATTR_BYTE));
|
|
3514
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_NORMALIZED", INT2NUM(LLAMA_TOKEN_ATTR_NORMALIZED));
|
|
3515
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_LSTRIP", INT2NUM(LLAMA_TOKEN_ATTR_LSTRIP));
|
|
3516
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_RSTRIP", INT2NUM(LLAMA_TOKEN_ATTR_RSTRIP));
|
|
3517
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_TOKEN_ATTR_SINGLE_WORD", INT2NUM(LLAMA_TOKEN_ATTR_SINGLE_WORD));
|
|
3518
|
+
|
|
3506
3519
|
rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_ALL_F32", INT2NUM(LLAMA_FTYPE_ALL_F32));
|
|
3507
3520
|
rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_F16", INT2NUM(LLAMA_FTYPE_MOSTLY_F16));
|
|
3508
3521
|
rb_define_const(rb_mLLaMACpp, "LLAMA_FTYPE_MOSTLY_Q4_0", INT2NUM(LLAMA_FTYPE_MOSTLY_Q4_0));
|
|
@@ -3547,6 +3560,7 @@ extern "C" void Init_llama_cpp(void) {
|
|
|
3547
3560
|
rb_define_const(rb_mLLaMACpp, "LLAMA_GRETYPE_CHAR_NOT", INT2NUM(LLAMA_GRETYPE_CHAR_NOT));
|
|
3548
3561
|
rb_define_const(rb_mLLaMACpp, "LLAMA_GRETYPE_CHAR_RNG_UPPER", INT2NUM(LLAMA_GRETYPE_CHAR_RNG_UPPER));
|
|
3549
3562
|
rb_define_const(rb_mLLaMACpp, "LLAMA_GRETYPE_CHAR_ALT", INT2NUM(LLAMA_GRETYPE_CHAR_ALT));
|
|
3563
|
+
rb_define_const(rb_mLLaMACpp, "LLAMA_GRETYPE_CHAR_ANY", INT2NUM(LLAMA_GRETYPE_CHAR_ANY));
|
|
3550
3564
|
|
|
3551
3565
|
rb_define_const(rb_mLLaMACpp, "LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED", INT2NUM(LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED));
|
|
3552
3566
|
rb_define_const(rb_mLLaMACpp, "LLAMA_ROPE_SCALING_TYPE_NONE", INT2NUM(LLAMA_ROPE_SCALING_TYPE_NONE));
|
data/lib/llama_cpp/version.rb
CHANGED
|
@@ -3,8 +3,8 @@
|
|
|
3
3
|
# llama_cpp.rb provides Ruby bindings for the llama.cpp.
|
|
4
4
|
module LLaMACpp
|
|
5
5
|
# The version of llama_cpp.rb you install.
|
|
6
|
-
VERSION = '0.
|
|
6
|
+
VERSION = '0.16.1'
|
|
7
7
|
|
|
8
8
|
# The version of llama.cpp bundled with llama_cpp.rb.
|
|
9
|
-
LLAMA_CPP_VERSION = '
|
|
9
|
+
LLAMA_CPP_VERSION = 'b3151'
|
|
10
10
|
end
|
data/sig/llama_cpp.rbs
CHANGED
|
@@ -31,6 +31,19 @@ module LLaMACpp
|
|
|
31
31
|
LLAMA_VOCAB_PRE_TYPE_OLMO: Integer
|
|
32
32
|
LLAMA_VOCAB_PRE_TYPE_DBRX: Integer
|
|
33
33
|
LLAMA_VOCAB_PRE_TYPE_SMAUG: Integer
|
|
34
|
+
LLAMA_VOCAB_PRE_TYPE_PORO: Integer
|
|
35
|
+
|
|
36
|
+
LLAMA_TOKEN_ATTR_UNDEFINED: Integer
|
|
37
|
+
LLAMA_TOKEN_ATTR_UNKNOWN: Integer
|
|
38
|
+
LLAMA_TOKEN_ATTR_UNUSED: Integer
|
|
39
|
+
LLAMA_TOKEN_ATTR_NORMAL: Integer
|
|
40
|
+
LLAMA_TOKEN_ATTR_CONTROL: Integer
|
|
41
|
+
LLAMA_TOKEN_ATTR_USER_DEFINED: Integer
|
|
42
|
+
LLAMA_TOKEN_ATTR_BYTE: Integer
|
|
43
|
+
LLAMA_TOKEN_ATTR_NORMALIZED: Integer
|
|
44
|
+
LLAMA_TOKEN_ATTR_LSTRIP: Integer
|
|
45
|
+
LLAMA_TOKEN_ATTR_RSTRIP: Integer
|
|
46
|
+
LLAMA_TOKEN_ATTR_SINGLE_WORD: Integer
|
|
34
47
|
|
|
35
48
|
LLAMA_FTYPE_ALL_F32: Integer
|
|
36
49
|
LLAMA_FTYPE_MOSTLY_F16: Integer
|
|
@@ -74,6 +87,7 @@ module LLaMACpp
|
|
|
74
87
|
LLAMA_GRETYPE_CHAR_NOT: Integer
|
|
75
88
|
LLAMA_GRETYPE_CHAR_RNG_UPPER: Integer
|
|
76
89
|
LLAMA_GRETYPE_CHAR_ALT: Integer
|
|
90
|
+
LLAMA_GRETYPE_CHAR_ANY: Integer
|
|
77
91
|
|
|
78
92
|
LLAMA_ROPE_SCALING_TYPE_UNSPECIFIED: Integer
|
|
79
93
|
LLAMA_ROPE_SCALING_TYPE_NONE: Integer
|
|
@@ -147,7 +161,7 @@ module LLaMACpp
|
|
|
147
161
|
def n_params: () -> Integer
|
|
148
162
|
def text: (Integer) -> String
|
|
149
163
|
def score: (Integer) -> Float
|
|
150
|
-
def
|
|
164
|
+
def token_attr: (Integer) -> Integer
|
|
151
165
|
def token_bos: () -> Integer
|
|
152
166
|
def token_eos: () -> Integer
|
|
153
167
|
def token_cls: () -> Integer
|