├── host-index.bat ├── host-index.sh ├── .github ├── dependabot.yml └── workflows │ ├── build-wheels-oobabooga-rocm.yml │ ├── build-everything-tgw.yml │ ├── build-wheels-cpu.yml │ ├── build-wheels-macos.yml │ └── build-wheels-oobabooga.yml ├── index ├── AVX │ ├── rocm5.4.2 │ │ └── index.html │ ├── rocm5.5.1 │ │ ├── index.html │ │ └── llama-cpp-python │ │ │ └── index.html │ ├── rocm5.5 │ │ └── index.html │ ├── rocm5.6.1 │ │ └── index.html │ ├── cpu │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu116 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu117 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu118 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu120 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu121 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu122 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ └── index.html ├── AVX2 │ ├── rocm5.4.2 │ │ └── index.html │ ├── rocm5.5.1 │ │ └── index.html │ ├── rocm5.5 │ │ └── index.html │ ├── rocm5.6.1 │ │ └── index.html │ ├── cpu │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu116 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu117 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu118 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu120 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu121 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu122 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ └── index.html ├── basic │ ├── rocm5.5 │ │ └── index.html │ ├── rocm5.4.2 │ │ └── index.html │ ├── rocm5.5.1 │ │ ├── index.html │ │ └── llama-cpp-python │ │ │ └── index.html │ ├── rocm5.6.1 │ │ └── index.html │ ├── cpu │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu116 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu117 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu118 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu120 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu121 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu122 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ └── index.html ├── textgen │ ├── AVX │ │ ├── rocm5.5 │ │ │ └── index.html │ │ ├── rocm5.4.2 │ │ │ └── index.html │ │ ├── rocm5.5.1 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-cuda │ │ │ │ └── index.html │ │ ├── rocm5.6.1 │ │ │ └── index.html │ │ ├── cu117 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu118 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu120 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu121 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu122 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ └── index.html │ ├── AVX2 │ │ ├── rocm5.4.2 │ │ │ └── index.html │ │ ├── rocm5.5.1 │ │ │ └── index.html │ │ ├── rocm5.5 │ │ │ └── index.html │ │ ├── rocm5.6.1 │ │ │ └── index.html │ │ ├── cu117 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu118 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu120 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu121 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu122 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ └── index.html │ ├── basic │ │ ├── rocm5.4.2 │ │ │ └── index.html │ │ ├── rocm5.5.1 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-cuda │ │ │ │ └── index.html │ │ ├── rocm5.5 │ │ │ └── index.html │ │ ├── rocm5.6.1 │ │ │ └── index.html │ │ ├── cu117 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu118 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu120 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu121 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ ├── cu122 │ │ │ ├── index.html │ │ │ └── llama-cpp-python-ggml-cuda │ │ │ │ └── index.html │ │ └── index.html │ └── index.html ├── AVX512 │ ├── cpu │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu116 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu117 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu118 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu120 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu121 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ ├── cu122 │ │ ├── index.html │ │ └── llama-cpp-python-ggml │ │ │ └── index.html │ └── index.html └── index.html ├── LICENSE ├── README.md ├── workflows.md ├── generate-textgen-html.ps1 └── generate-html.ps1 /host-index.bat: -------------------------------------------------------------------------------- 1 | @echo off 2 | 3 | cd index 4 | python -m http.server 7860 5 | -------------------------------------------------------------------------------- /host-index.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | cd index 4 | python -m http.server 7860 5 | -------------------------------------------------------------------------------- /.github/dependabot.yml: -------------------------------------------------------------------------------- 1 | version: 2 2 | updates: 3 | - package-ecosystem: "github-actions" 4 | directory: "/" 5 | schedule: 6 | interval: "daily" 7 | -------------------------------------------------------------------------------- /index/AVX/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX2/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX2/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX2/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/AVX2/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/basic/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/basic/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/basic/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/basic/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX2/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX2/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX2/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/AVX2/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/basic/rocm5.4.2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/basic/rocm5.5.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/basic/rocm5.5/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/basic/rocm5.6.1/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | 6 | 7 | -------------------------------------------------------------------------------- /index/textgen/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | AVX
5 | AVX2
6 | basic
7 | 8 | 9 | -------------------------------------------------------------------------------- /index/AVX/cpu/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu116/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cpu/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu116/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX2/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cpu/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu116/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cpu/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu116/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/basic/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python
5 | llama_cpp_python_ggml 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | AVX
5 | AVX2
6 | AVX512
7 | basic
8 | 9 | 10 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/basic/cu117/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/basic/cu118/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/basic/cu120/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/basic/cu121/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/textgen/basic/cu122/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda
5 | llama_cpp_python_ggml_cuda 6 | 7 | 8 | -------------------------------------------------------------------------------- /index/AVX512/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.6
5 | CUDA 11.7
6 | CUDA 11.8
7 | CUDA 12.0
8 | CUDA 12.1
9 | CUDA 12.2
10 | cpu
11 | 12 | 13 | -------------------------------------------------------------------------------- /index/textgen/AVX/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.7
5 | CUDA 11.8
6 | CUDA 12.0
7 | CUDA 12.1
8 | CUDA 12.2
9 | ROCm 5.4.2
10 | ROCm 5.5
11 | ROCm 5.5.1
12 | ROCm 5.6.1
13 | 14 | 15 | -------------------------------------------------------------------------------- /index/textgen/AVX2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.7
5 | CUDA 11.8
6 | CUDA 12.0
7 | CUDA 12.1
8 | CUDA 12.2
9 | ROCm 5.4.2
10 | ROCm 5.5
11 | ROCm 5.5.1
12 | ROCm 5.6.1
13 | 14 | 15 | -------------------------------------------------------------------------------- /index/textgen/basic/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.7
5 | CUDA 11.8
6 | CUDA 12.0
7 | CUDA 12.1
8 | CUDA 12.2
9 | ROCm 5.4.2
10 | ROCm 5.5
11 | ROCm 5.5.1
12 | ROCm 5.6.1
13 | 14 | 15 | -------------------------------------------------------------------------------- /index/AVX/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.6
5 | CUDA 11.7
6 | CUDA 11.8
7 | CUDA 12.0
8 | CUDA 12.1
9 | CUDA 12.2
10 | ROCm 5.4.2
11 | ROCm 5.5
12 | ROCm 5.5.1
13 | ROCm 5.6.1
14 | cpu
15 | 16 | 17 | -------------------------------------------------------------------------------- /index/AVX2/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.6
5 | CUDA 11.7
6 | CUDA 11.8
7 | CUDA 12.0
8 | CUDA 12.1
9 | CUDA 12.2
10 | ROCm 5.4.2
11 | ROCm 5.5
12 | ROCm 5.5.1
13 | ROCm 5.6.1
14 | cpu
15 | 16 | 17 | -------------------------------------------------------------------------------- /index/basic/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | CUDA 11.6
5 | CUDA 11.7
6 | CUDA 11.8
7 | CUDA 12.0
8 | CUDA 12.1
9 | CUDA 12.2
10 | ROCm 5.4.2
11 | ROCm 5.5
12 | ROCm 5.5.1
13 | ROCm 5.6.1
14 | cpu
15 | 16 | 17 | -------------------------------------------------------------------------------- /.github/workflows/build-wheels-oobabooga-rocm.yml: -------------------------------------------------------------------------------- 1 | name: Build ROCm Wheels 2 | 3 | on: 4 | workflow_dispatch: 5 | inputs: 6 | version: 7 | description: 'Version tag of llama-cpp-python to build: v0.2.9' 8 | default: 'v0.2.9' 9 | required: true 10 | type: string 11 | workflow_call: 12 | inputs: 13 | version: 14 | description: 'Version tag of llama-cpp-python to build: v0.2.9' 15 | default: 'v0.2.9' 16 | required: true 17 | type: string 18 | 19 | permissions: 20 | contents: write 21 | 22 | jobs: 23 | build_wheels_rocm: 24 | name: ROCm Wheels 25 | uses: ./.github/workflows/build-wheels-rocm-full.yml 26 | with: 27 | version: ${{ inputs.version }} 28 | config: 'rename:1' 29 | -------------------------------------------------------------------------------- /LICENSE: -------------------------------------------------------------------------------- 1 | This is free and unencumbered software released into the public domain. 2 | 3 | Anyone is free to copy, modify, publish, use, compile, sell, or 4 | distribute this software, either in source code form or as a compiled 5 | binary, for any purpose, commercial or non-commercial, and by any 6 | means. 7 | 8 | In jurisdictions that recognize copyright laws, the author or authors 9 | of this software dedicate any and all copyright interest in the 10 | software to the public domain. We make this dedication for the benefit 11 | of the public at large and to the detriment of our heirs and 12 | successors. We intend this dedication to be an overt act of 13 | relinquishment in perpetuity of all present and future rights to this 14 | software under copyright law. 15 | 16 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 17 | EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 18 | MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. 19 | IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR 20 | OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 21 | ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR 22 | OTHER DEALINGS IN THE SOFTWARE. 23 | 24 | For more information, please refer to 25 | -------------------------------------------------------------------------------- /.github/workflows/build-everything-tgw.yml: -------------------------------------------------------------------------------- 1 | name: Build Everything TGW 2 | 3 | on: 4 | workflow_dispatch: 5 | inputs: 6 | version: 7 | description: 'Version tag of llama-cpp-python to build: v0.2.14' 8 | default: 'v0.2.14' 9 | required: true 10 | type: string 11 | 12 | permissions: 13 | contents: write 14 | 15 | jobs: 16 | build_textgen_wheels_windows: 17 | name: CUDA Wheels Windows 18 | uses: ./.github/workflows/build-wheels-oobabooga.yml 19 | with: 20 | version: ${{ inputs.version }} 21 | config: 'os:windows-2019' 22 | 23 | build_textgen_wheels_linux: 24 | name: CUDA Wheels Linux 25 | uses: ./.github/workflows/build-wheels-oobabooga.yml 26 | with: 27 | version: ${{ inputs.version }} 28 | config: 'os:ubuntu-22.04' 29 | 30 | build_wheels_tensorcores_windows: 31 | name: Tensor Core Windows 32 | uses: ./.github/workflows/build-wheels-tensorcores.yml 33 | with: 34 | version: ${{ inputs.version }} 35 | config: 'os:windows-2019' 36 | 37 | build_wheels_tensorcores_linux: 38 | name: Tensor Core Linux 39 | uses: ./.github/workflows/build-wheels-tensorcores.yml 40 | with: 41 | version: ${{ inputs.version }} 42 | config: 'os:ubuntu-22.04' 43 | 44 | build_wheels_cpu: 45 | name: CPU-only Wheels 46 | uses: ./.github/workflows/build-wheels-cpu.yml 47 | with: 48 | version: ${{ inputs.version }} 49 | 50 | build_wheels_macos: 51 | name: MacOS Metal Wheels 52 | uses: ./.github/workflows/build-wheels-macos.yml 53 | with: 54 | version: ${{ inputs.version }} 55 | 56 | build_wheels_rocm: 57 | name: ROCm Wheels 58 | uses: ./.github/workflows/build-wheels-oobabooga-rocm.yml 59 | with: 60 | version: ${{ inputs.version }} 61 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu117/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu117-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu118/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu118-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu120/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu120-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu121/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu121-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX2/cu122/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu122-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu117/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu117avx-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu118/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu118avx-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu120/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu120avx-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu121/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu121avx-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/AVX/cu122/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu122avx-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/basic/cu117/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu117basic-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/basic/cu118/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu118basic-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/basic/cu120/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu120basic-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/basic/cu121/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu121basic-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/textgen/basic/cu122/llama-cpp-python-ggml-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp38-cp38-linux_x86_64.whl
5 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp38-cp38-win_amd64.whl
6 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp39-cp39-linux_x86_64.whl
7 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp39-cp39-win_amd64.whl
8 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp310-cp310-linux_x86_64.whl
9 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp310-cp310-win_amd64.whl
10 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp311-cp311-linux_x86_64.whl
11 | llama_cpp_python_ggml_cuda-0.1.78+cu122basic-cp311-cp311-win_amd64.whl
12 | 13 | 14 | -------------------------------------------------------------------------------- /index/AVX/cu116/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cu117/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cu118/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cu120/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cu121/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cu122/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX/cpu/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cpuavx-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cpuavx-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cpuavx-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cpuavx-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cpuavx-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cpuavx-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cpuavx-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cpuavx-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cpuavx-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cpuavx-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu116/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu117/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu118/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu120/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu121/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cu122/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu116/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu117/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu118/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu120/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu121/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cu122/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu116/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu116-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu116-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu116-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu116-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu116-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu117/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu117-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu117-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu117-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu117-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu117-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu118/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu118-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu118-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu118-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu118-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu118-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu120/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu120-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu120-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu120-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu120-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu120-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu121/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu121-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu121-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu121-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu121-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu121-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cu122/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cu122-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cu122-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cu122-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cu122-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cu122-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX2/cpu/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cpuavx2-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/basic/cpu/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cpubasic-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cpubasic-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cpubasic-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cpubasic-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cpubasic-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cpubasic-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cpubasic-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cpubasic-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cpubasic-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cpubasic-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /index/AVX512/cpu/llama-cpp-python-ggml/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp37-cp37m-linux_x86_64.whl
5 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp37-cp37m-win_amd64.whl
6 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp38-cp38-linux_x86_64.whl
7 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp38-cp38-win_amd64.whl
8 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp39-cp39-linux_x86_64.whl
9 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp39-cp39-win_amd64.whl
10 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp310-cp310-linux_x86_64.whl
11 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp311-cp311-linux_x86_64.whl
13 | llama_cpp_python_ggml-0.1.78+cpuavx512-cp311-cp311-win_amd64.whl
14 | 15 | 16 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # llama-cpp-python cuBLAS wheels 2 | Wheels for [llama-cpp-python](https://github.com/abetlen/llama-cpp-python) compiled with cuBLAS support. 3 | 4 | Requirements: 5 | - Windows x64, Linux x64, or MacOS 11.0+ 6 | - CUDA 11.6 - 12.2 7 | - CPython 3.8 - 3.11 8 | 9 | > [!WARNING] 10 | > MacOS 11 and Windows ROCm wheels are unavailable for 0.2.22+. 11 | > This is due to build issues with llama.cpp that are not yet resolved. 12 | 13 | ROCm builds for AMD GPUs: https://github.com/jllllll/llama-cpp-python-cuBLAS-wheels/releases/tag/rocm 14 | Metal builds for MacOS 11.0+: https://github.com/jllllll/llama-cpp-python-cuBLAS-wheels/releases/tag/metal 15 | 16 | Installation instructions: 17 | --- 18 | To install, you can use this command: 19 | ``` 20 | python -m pip install llama-cpp-python --prefer-binary --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 21 | ``` 22 | This will install the latest llama-cpp-python version available from here for CUDA 11.7. You can change `cu117` to change the CUDA version. 23 | You can also change `AVX2` to `AVX`, `AVX512` or `basic` based on what your CPU supports. 24 | `basic` is a build without `AVX`, `FMA` and `F16C` instructions for old or basic CPUs. 25 | CPU-only builds are also available by changing `cu117` to `cpu`. 26 | 27 | You can install a specific version with: 28 | ``` 29 | python -m pip install llama-cpp-python== --prefer-binary --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 30 | ``` 31 | An example for installing 0.1.62 for CUDA 12.1 on a CPU without AVX2 support: 32 | ``` 33 | python -m pip install llama-cpp-python==0.1.62 --prefer-binary --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX/cu121 34 | ``` 35 | List of available versions: 36 | ``` 37 | python -m pip index versions llama-cpp-python --index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 38 | ``` 39 | 40 | If you are replacing an already existing installation, you may need to uninstall that version before running the command above. 41 | You can also replace the existing version in one command like so: 42 | ``` 43 | python -m pip install llama-cpp-python --force-reinstall --no-deps --index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 44 | -OR- 45 | python -m pip install llama-cpp-python==0.1.66 --force-reinstall --no-deps --index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 46 | -OR- 47 | python -m pip install llama-cpp-python --prefer-binary --upgrade --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 48 | ``` 49 | 50 | Wheels can be manually downloaded from: https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels 51 | 52 | --- 53 | I have renamed llama-cpp-python packages available to ease the transition to GGUF. 54 | This is accomplished by installing the renamed package alongside the main llama-cpp-python package. 55 | This should allow applications to maintain GGML support while still supporting GGUF. 56 | ``` 57 | python -m pip install llama-cpp-python-ggml --prefer-binary --extra-index-url=https://jllllll.github.io/llama-cpp-python-cuBLAS-wheels/AVX2/cu117 58 | ``` 59 | 60 | --- 61 | ### All wheels are compiled using GitHub Actions 62 | -------------------------------------------------------------------------------- /workflows.md: -------------------------------------------------------------------------------- 1 | All workflows are configured to accept a llama-cpp-python release tag to build a specific version of the package. 2 | For the most part, they are written to account for changes in every version since 0.1.62. 3 | 4 | Primary workflows used for new llama-cpp-python releases 5 | ---- 6 | - `build-wheels.yml` 7 | - This workflow will build around 192 wheels for various CUDA, Python and CPU configurations. After this, it will call the `build-wheels-cpu.yml` workflow. 8 | - `build-wheels-full-release.yml` 9 | - This workflow calls these workflows in order: `build-wheels.yml build-wheels-oobabooga.yml build-wheels-rocm-full.yml build-wheels-macos.yml` 10 | - Somewhere around 370 wheels are produced in total, last I checked. This number will likely increase as additional builds, such as MacOS Metal, are eventually included. 11 | - `build-wheels-prioritized-release.yml` 12 | - This workflow is much like `build-wheels-full-release.yml`, except `build-wheels.yml` and `build-wheels-oobabooga.yml` are incorporated into the workflow instead of being called due to minor modifications. 13 | - This workflow is configured to build the wheels used by [text-generation-webui](https://github.com/oobabooga/text-generation-webui) first. This is because the long runtime of the workflow (currently 5 - 6 hours) was causing significant delays in updating the project. 14 | - `build-wheels-cpu.yml` 15 | - This workflow builds CPU-only wheels for all of the CPU configurations supported by the other workflows. 16 | - It was made because the wheels in the main repo are only built to support the default configuration of `AVX2`. 17 | 18 | ~~These workflows, and their dependents, were recently optimized to significantly reduce run times from 6 hours for the longest down to around 2 hours.~~ 19 | Workflow optimizations have been made incompatible with llama-cpp-python 0.2.X+ due to abetlen switching the build backend to one that does not support modifications of the build process. 20 | Copies of the optimized workflows can be found in the `old_workflows` directory. 21 | 22 | Renamed package workflows 23 | ---- 24 | These workflows produced renamed packages under different namespaces to allow for simultaneous installation with the main package. 25 | - `build-wheels-oobabooga*.yml` 26 | - This workflow builds wheels with packages renamed to `llama_cpp_python_cuda`. 27 | - These wheels are to allow applications to simultaneously support both CPU and CUDA builds of llama-cpp-python. 28 | - As the name implies, this was made for text-generation-webui. 29 | - `build-wheels-ggml*.yml` 30 | - This workflow was made to produce wheels for llama-cpp-python 0.1.78 under the name of `llama_cpp_python_ggml`. 31 | - This allows applications to maintain support for GGML while updating to newer versions of llama-cpp-python. 32 | - Intended to be a temporary measure until more models are converted to GGUF. 33 | 34 | Configuration-specific workflows 35 | ---- 36 | - `*-basic.yml` 37 | - `*-avx.yml` 38 | 39 | These are copies of other workflows with build matrices limited to specific configurations. 40 | For the most part, I made these to rebuild previous versions of llama-cpp-python as needed to support new configurations that were added to the main workflows. 41 | 42 | Batch build workflows 43 | ---- 44 | - `build-wheels-batch-*.yml` 45 | 46 | These workflows accept a comma-separated string of llama-cpp-python release tags. 47 | They then use Powershell to parse the input and construct a JSON string that is used to form a job matrix. 48 | Associated workflows are then called as needed to build each version. 49 | Only one workflow is executed at a time due to the large number of jobs that can be generated. 50 | 51 | Experimental workflows used for more specialized builds 52 | ---- 53 | - `build-wheel-rocm.yml` 54 | - This workflow builds Linux and Windows wheels for AMD GPUs using ROCm. 55 | - Linux wheels are built using these ROCm versions: `5.4.2 5.5 5.6.1` 56 | - Currently considered experimental until someone with an AMD GPU can confirm if the resulting wheels work. 57 | - `build-wheels-oobabooga-rocm.yml` 58 | - This workflow is much like the previous. It additionally builds `llama_cpp_python_cuda` wheels. 59 | - `build-wheels-rocm-full.yml` 60 | - This workflow is essentially a combination of the previous 2. 61 | - `build-wheels-macos.yml` 62 | - This workflow builds wheels with MacOS Metal support for MacOS 11, 12 and 13. 63 | - Builds separate wheels for Intel and Apple Silicon CPU support. 64 | - Is currently experimental and may not produce functional Metal wheels. I do not have a Mac to test with, so I can only go off of build logs. 65 | 66 | Utility workflows 67 | ---- 68 | - `deploy-index.yml` 69 | - This workflow is for deploying the package index to GitHub Pages. 70 | - It is configured to run automatically when the index's html files are altered. 71 | - `build-wheels-test.yml` 72 | - This workflow is entirely for testing new workflow code and is changed frequently as needed. 73 | -------------------------------------------------------------------------------- /.github/workflows/build-wheels-cpu.yml: -------------------------------------------------------------------------------- 1 | name: Build CPU-only Wheels 2 | 3 | on: 4 | workflow_dispatch: 5 | inputs: 6 | version: 7 | description: 'Version tag of llama-cpp-python to build: v0.2.14' 8 | default: 'v0.2.14' 9 | required: true 10 | type: string 11 | config: 12 | description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 13 | default: 'Default' 14 | required: false 15 | type: string 16 | exclude: 17 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 18 | default: 'None' 19 | required: false 20 | type: string 21 | workflow_call: 22 | inputs: 23 | version: 24 | description: 'Version tag of llama-cpp-python to build: v0.2.14' 25 | default: 'v0.2.14' 26 | required: true 27 | type: string 28 | config: 29 | description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 30 | default: 'Default' 31 | required: false 32 | type: string 33 | exclude: 34 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 35 | default: 'None' 36 | required: false 37 | type: string 38 | 39 | permissions: 40 | contents: write 41 | 42 | jobs: 43 | define_matrix: 44 | name: Define Build Matrix 45 | runs-on: ubuntu-latest 46 | outputs: 47 | matrix: ${{ steps.set-matrix.outputs.matrix }} 48 | defaults: 49 | run: 50 | shell: pwsh 51 | env: 52 | CONFIGIN: ${{ inputs.config }} 53 | EXCLUDEIN: ${{ inputs.exclude }} 54 | 55 | steps: 56 | - name: Define Job Output 57 | id: set-matrix 58 | run: | 59 | $matrix = @{ 60 | 'os' = @('ubuntu-22.04', 'windows-2019') 61 | 'pyver' = @("3.11") 62 | 'avx' = @("AVX", "AVX2") 63 | } 64 | 65 | if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})} 66 | 67 | if ($env:EXCLUDEIN -ne 'None') { 68 | $exclusions = @() 69 | $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData 70 | $matrix['exclude'] = $exclusions 71 | } 72 | 73 | $matrixOut = ConvertTo-Json $matrix -Compress 74 | Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT 75 | 76 | build_wheels: 77 | name: ${{ matrix.os }} ${{ matrix.pyver }} CPU ${{ matrix.avx }} 78 | needs: define_matrix 79 | runs-on: ${{ matrix.os }} 80 | strategy: 81 | matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }} 82 | defaults: 83 | run: 84 | shell: pwsh 85 | env: 86 | AVXVER: ${{ matrix.avx }} 87 | PCKGVER: ${{ inputs.version }} 88 | 89 | steps: 90 | - uses: actions/checkout@v4 91 | with: 92 | repository: 'abetlen/llama-cpp-python' 93 | ref: ${{ inputs.version }} 94 | submodules: 'recursive' 95 | 96 | - uses: actions/setup-python@v4 97 | with: 98 | python-version: ${{ matrix.pyver }} 99 | 100 | - name: Setup Mamba 101 | uses: conda-incubator/setup-miniconda@v3.1.0 102 | with: 103 | activate-environment: "TGW" 104 | python-version: ${{ matrix.pyver }} 105 | miniforge-version: latest 106 | add-pip-as-python-dependency: true 107 | auto-activate-base: false 108 | 109 | - name: Install Dependencies 110 | run: | 111 | python -m pip install build wheel 112 | 113 | - name: Build Wheel 114 | run: | 115 | $packageVersion = [version]$env:PCKGVER.TrimStart('v') 116 | $env:VERBOSE = '1' 117 | if ($env:AVXVER -eq 'AVX') {$env:CMAKE_ARGS = '-DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off'} 118 | if ($env:AVXVER -eq 'AVX512') {$env:CMAKE_ARGS = '-DGGML_AVX512=on'} 119 | if ($env:AVXVER -eq 'basic') {$env:CMAKE_ARGS = '-DGGML_AVX=off -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off'} 120 | if ($packageVersion -gt [version]'0.2.13') {$env:CMAKE_ARGS = "-DGGML_NATIVE=off $env:CMAKE_ARGS"} 121 | $buildtag = "+cpu$env:AVXVER" 122 | if ($packageVersion -lt [version]'0.2.0') { 123 | $env:FORCE_CMAKE = '1' 124 | python -m build --wheel -C--build-option=egg_info "-C--build-option=--tag-build=$buildtag" 125 | } else { 126 | $initpath = Join-Path '.' 'llama_cpp' '__init__.py' -resolve 127 | $initcontent = Get-Content $initpath -raw 128 | $regexstr = '(?s)(?<=__version__ \= ")\d+(?:\.\d+)*(?=")' 129 | $regexmatch = [Regex]::Matches($initcontent,$regexstr) 130 | if (!($regexmatch[0].Success)) {throw '__init__.py parsing failed'} 131 | $newinit = $regexmatch[0].Result(('$`' + '$&' + $buildtag + '$''')) 132 | New-Item $initpath -itemType File -value $newinit -force 133 | python -m build --wheel 134 | } 135 | 136 | - name: Upload files to a GitHub release 137 | id: upload-release 138 | uses: svenstaro/upload-release-action@2.7.0 139 | continue-on-error: true 140 | with: 141 | repo_token: ${{ secrets.GITHUB_TOKEN }} 142 | file: ./dist/*.whl 143 | tag: 'cpu' 144 | file_glob: true 145 | make_latest: false 146 | overwrite: true 147 | 148 | - uses: actions/upload-artifact@v4 149 | if: steps.upload-release.outcome == 'failure' 150 | with: 151 | name: cpu 152 | path: ./dist/*.whl 153 | -------------------------------------------------------------------------------- /.github/workflows/build-wheels-macos.yml: -------------------------------------------------------------------------------- 1 | name: Build MacOS Wheels 2 | 3 | on: 4 | workflow_dispatch: 5 | inputs: 6 | version: 7 | description: 'Version tag of llama-cpp-python to build: v0.2.20' 8 | default: 'v0.2.20' 9 | required: true 10 | type: string 11 | config: 12 | description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 13 | default: 'Default' 14 | required: false 15 | type: string 16 | exclude: 17 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 18 | default: 'None' 19 | required: false 20 | type: string 21 | workflow_call: 22 | inputs: 23 | version: 24 | description: 'Version tag of llama-cpp-python to build: v0.2.20' 25 | default: 'v0.2.20' 26 | required: true 27 | type: string 28 | config: 29 | description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 30 | default: 'Default' 31 | required: false 32 | type: string 33 | exclude: 34 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 35 | default: 'None' 36 | required: false 37 | type: string 38 | 39 | permissions: 40 | contents: write 41 | 42 | jobs: 43 | define_matrix: 44 | name: Define Build Matrix 45 | runs-on: ubuntu-latest 46 | outputs: 47 | matrix: ${{ steps.set-matrix.outputs.matrix }} 48 | defaults: 49 | run: 50 | shell: pwsh 51 | env: 52 | CONFIGIN: ${{ inputs.config }} 53 | EXCLUDEIN: ${{ inputs.exclude }} 54 | 55 | steps: 56 | - name: Define Job Output 57 | id: set-matrix 58 | run: | 59 | $matrix = @{ 60 | 'os' = @("macos-13", "macos-14", "macos-15") 61 | 'pyver' = @("3.11") 62 | } 63 | 64 | if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})} 65 | 66 | if ($env:EXCLUDEIN -ne 'None') { 67 | $exclusions = @() 68 | $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData 69 | $matrix['exclude'] = $exclusions 70 | } 71 | 72 | $matrixOut = ConvertTo-Json $matrix -Compress 73 | Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT 74 | 75 | build_wheels: 76 | name: ${{ matrix.os }} Python ${{ matrix.pyver }} 77 | needs: define_matrix 78 | runs-on: ${{ matrix.os }} 79 | strategy: 80 | matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }} 81 | env: 82 | OSVER: ${{ matrix.os }} 83 | 84 | steps: 85 | - uses: actions/checkout@v4 86 | with: 87 | repository: 'abetlen/llama-cpp-python' 88 | ref: ${{ inputs.version }} 89 | submodules: 'recursive' 90 | 91 | - uses: actions/setup-python@v4 92 | with: 93 | python-version: ${{ matrix.pyver }} 94 | 95 | - name: Install Dependencies 96 | run: | 97 | python -m pip install build wheel cmake 98 | 99 | - name: Build Wheel 100 | run: | 101 | # Find the actual Xcode path once 102 | XCODE_PATH=$(ls -d /Applications/Xcode*.app | head -1) 103 | echo "Using Xcode at: ${XCODE_PATH}" 104 | XCODEDEV="${XCODE_PATH}/Contents/Developer" 105 | XCODEBINPATH="${XCODEDEV}/Toolchains/XcodeDefault.xctoolchain/usr/bin" 106 | 107 | # Set common arguments 108 | export CMAKE_ARGS="-DGGML_NATIVE=off -DGGML_METAL=on" 109 | 110 | # Always set the SDK root to avoid 'System' library not found errors 111 | export SDKROOT="${XCODEDEV}/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk" 112 | 113 | # Set correct compiler paths 114 | export CC="${XCODEBINPATH}/cc" 115 | export CXX="${XCODEBINPATH}/c++" 116 | 117 | # Set deployment target based on OS version 118 | [[ "$OSVER" == "macos-15" ]] && export MACOSX_DEPLOYMENT_TARGET="15.0" 119 | [[ "$OSVER" == "macos-14" ]] && export MACOSX_DEPLOYMENT_TARGET="14.0" 120 | [[ "$OSVER" == "macos-13" ]] && export MACOSX_DEPLOYMENT_TARGET="13.0" 121 | [[ "$OSVER" == "macos-12" ]] && export MACOSX_DEPLOYMENT_TARGET="12.0" 122 | [[ "$OSVER" == "macos-11" ]] && export MACOSX_DEPLOYMENT_TARGET="11.0" 123 | 124 | # ARM64 build 125 | export CMAKE_OSX_ARCHITECTURES="arm64" && export ARCHFLAGS="-arch arm64" 126 | VERBOSE=1 python -m build --wheel 127 | 128 | # Create aarch64 copies 129 | for file in ./dist/*.whl; do cp "$file" "${file/arm64.whl/aarch64.whl}"; done 130 | 131 | # Explicitly reset the SDK path for the second build 132 | export SDKROOT="${XCODEDEV}/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk" 133 | 134 | # x86_64 build with specific flags 135 | export CMAKE_OSX_ARCHITECTURES="x86_64" 136 | export CMAKE_ARGS="-DGGML_NATIVE=off -DGGML_AVX=off -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off -DGGML_METAL=on" 137 | export ARCHFLAGS="-arch x86_64" 138 | VERBOSE=1 python -m build --wheel 139 | 140 | - name: Upload files to a GitHub release 141 | id: upload-release 142 | uses: svenstaro/upload-release-action@2.7.0 143 | continue-on-error: true 144 | with: 145 | repo_token: ${{ secrets.GITHUB_TOKEN }} 146 | file: ./dist/*.whl 147 | tag: 'metal' 148 | file_glob: true 149 | make_latest: false 150 | overwrite: true 151 | 152 | - uses: actions/upload-artifact@v4 153 | if: steps.upload-release.outcome == 'failure' 154 | with: 155 | name: macos-wheels 156 | path: ./dist/*.whl 157 | -------------------------------------------------------------------------------- /generate-textgen-html.ps1: -------------------------------------------------------------------------------- 1 | Set-Location $PSScriptRoot 2 | 3 | $destinationDir = if (Test-Path $(Join-Path $(Resolve-Path '.') 'index')) {Join-Path '.' 'index' -resolve} else {(New-Item 'index' -ItemType 'Directory').fullname} 4 | $destinationDir = if (Test-Path $(Join-Path $destinationDir 'textgen')) {Join-Path $destinationDir 'textgen'} else {(New-Item $(Join-Path $destinationDir 'textgen') -ItemType 'Directory').fullname} 5 | $avxVersions = "AVX","AVX2","basic" 6 | $cudaVersions = "11.7","11.8","12.0","12.1","12.2","rocm5.4.2","rocm5.5","rocm5.5.1","rocm5.6.1" 7 | $packageVersions = (73..74+76..85).foreach({"$_".Insert(0,'0.1.')}) + (0..11+14..20+@(23)).foreach({"$_".Insert(0,'0.2.')}) 8 | $pythonVersions = "3.8","3.9","3.10","3.11" 9 | $supportedSystems = 'linux_x86_64','win_amd64' 10 | $wheelSource = 'https://github.com/jllllll/llama-cpp-python-cuBLAS-wheels/releases/download' 11 | $packageName = 'llama_cpp_python_cuda' 12 | $packageNameNormalized = 'llama-cpp-python-cuda' 13 | $packageNameAlt = 'llama_cpp_python_ggml_cuda' 14 | $packageNameAltNormalized = 'llama-cpp-python-ggml-cuda' 15 | $packageAltVersions = @("0.1.78") 16 | 17 | $avxVersions.foreach({Set-Variable "$_`Dir" $(if (Test-Path $(Join-Path $destinationDir $_)) {Join-Path $destinationDir $_} else {(New-Item $(Join-Path $destinationDir $_) -ItemType 'Directory').fullname})}) 18 | 19 | $indexContent = "`n`n `n " 20 | Foreach ($avxVersion in $avxVersions) 21 | { 22 | $wheelURL = $wheelSource.TrimEnd('/') + '/textgen-webui' 23 | $subIndexContent = "`n`n `n " 24 | ForEach ($cudaVersion in $cudaVersions) 25 | { 26 | if ($cudaVersion.StartsWith('rocm')) {$wheelURL = $wheelSource.TrimEnd('/') + '/rocm'} 27 | $cu = if ($cudaVersion.StartsWith('rocm')) {$cudaVersion} else {'cu' + $cudaVersion.replace('.','')} 28 | $cuContent = "`n`n `n " 29 | $cuContentAlt = "`n`n `n " 30 | ForEach ($packageVersion in $packageVersions) 31 | { 32 | if ($avxVersion -eq 'basic' -and $packageVersion -eq '0.1.73') {continue} 33 | if ($cudaVersion.StartsWith('rocm') -and [version]$packageVersion -lt [version]"0.1.80") {continue} 34 | if ($cudaVersion.StartsWith('rocm') -and $avxVersion -ne 'AVX2' -and [version]$packageVersion -lt [version]"0.2.7") {continue} 35 | ForEach ($pythonVersion in $pythonVersions) 36 | { 37 | $pyVer = $pythonVersion.replace('.','') 38 | ForEach ($supportedSystem in $supportedSystems) 39 | { 40 | if ($cudaVersion.StartsWith('rocm') -and [version]$packageVersion -gt [version]"0.2.21" -and $supportedSystem -eq 'win_amd64') {continue} 41 | if ($cudaVersion.StartsWith('rocm') -and $cudaVersion.Split('rocm')[-1] -ne '5.5.1' -and $supportedSystem -eq 'win_amd64') {continue} 42 | if ($cudaVersion.StartsWith('rocm') -and $cudaVersion.Split('rocm')[-1] -eq '5.5.1' -and $supportedSystem -eq 'linux_x86_64') {continue} 43 | if ([version]$packageVersion -gt [version]"0.1.85" -and $supportedSystem -eq 'linux_x86_64') {$supportedSystem = 'manylinux_2_31_x86_64'} 44 | $wheel = if ($avxVersion -eq 'AVX') { "$packageName-$packageVersion+$cu$('avx')-cp$pyVer-cp$pyVer-$supportedSystem.whl" 45 | } elseif ($avxVersion -eq 'basic') { "$packageName-$packageVersion+$cu$('basic')-cp$pyVer-cp$pyVer-$supportedSystem.whl" 46 | } else {"$packageName-$packageVersion+$cu-cp$pyVer-cp$pyVer-$supportedSystem.whl"} 47 | $wheelAlt = if ($avxVersion -eq 'AVX') { "$packageNameAlt-$packageVersion+$cu$('avx')-cp$pyVer-cp$pyVer-$supportedSystem.whl" 48 | } elseif ($avxVersion -eq 'basic') { "$packageNameAlt-$packageVersion+$cu$('basic')-cp$pyVer-cp$pyVer-$supportedSystem.whl" 49 | } else {"$packageNameAlt-$packageVersion+$cu-cp$pyVer-cp$pyVer-$supportedSystem.whl"} 50 | if (!($packageVersion -eq '0.1.73' -and $avxVersion -eq 'AVX')) {$cuContent += "$wheel
`n "} 51 | if ($packageVersion -in $packageAltVersions -and !$cudaVersion.StartsWith('rocm')) {$cuContentAlt += "$wheelAlt
`n "} 52 | } 53 | } 54 | if (!($packageVersion -eq '0.1.73' -and $avxVersion -eq 'AVX')) {$cuContent += "`n "} 55 | if ($packageVersion -in $packageAltVersions) {$cuContentAlt += "`n "} 56 | } 57 | $cuDir = if (Test-Path $(Join-Path $(Get-Variable "$avxVersion`Dir").Value $cu)) {Join-Path $(Get-Variable "$avxVersion`Dir").Value $cu} else {(New-Item $(Join-Path $(Get-Variable "$avxVersion`Dir").Value $cu) -ItemType 'Directory').fullname} 58 | $packageDir = if (Test-Path $(Join-Path $cuDir $packageNameNormalized)) {Join-Path $cuDir $packageNameNormalized} else {(New-Item $(Join-Path $cuDir $packageNameNormalized) -ItemType 'Directory').fullname} 59 | $packageAltDir = if (Test-Path $(Join-Path $cuDir $packageNameAltNormalized)) {Join-Path $cuDir $packageNameAltNormalized} elseif (!$cudaVersion.StartsWith('rocm')) {(New-Item $(Join-Path $cuDir $packageNameAltNormalized) -ItemType 'Directory').fullname} 60 | $cuLabel = if ($cudaVersion.StartsWith('rocm')) {'ROCm'+' '+$cudaVersion.Split('rocm')[-1]} else {"CUDA $cudaVersion"} 61 | $subIndexContent += "$cuLabel
`n " 62 | New-Item $(Join-Path $packageDir "index.html") -itemType File -value $($cuContent.TrimEnd() + "`n `n`n") -force > $null 63 | if (!$cudaVersion.StartsWith('rocm')) {New-Item $(Join-Path $packageAltDir "index.html") -itemType File -value $($cuContentAlt.TrimEnd() + "`n `n`n") -force > $null} 64 | $packageIndexContent = "`n`n `n $packageName
`n" 65 | if (!$cudaVersion.StartsWith('rocm')) {$packageIndexContent += " $packageNameAlt`n"} 66 | New-Item $(Join-Path $cuDir "index.html") -itemType File -value $($packageIndexContent + " `n`n") -force > $null 67 | } 68 | $indexContent += "$avxVersion
`n " 69 | New-Item $(Join-Path $(Get-Variable "$avxVersion`Dir").Value "index.html") -itemType File -value $($subIndexContent.TrimEnd() + "`n `n`n") -force > $null 70 | } 71 | New-Item $(Join-Path $destinationDir "index.html") -itemType File -value $($indexContent.TrimEnd() + "`n `n`n") -force > $null 72 | #"`n`n `n `n `n `n CUDA $cudaVersion
`n `n" 73 | 74 | pause 75 | -------------------------------------------------------------------------------- /generate-html.ps1: -------------------------------------------------------------------------------- 1 | Set-Location $PSScriptRoot 2 | 3 | $destinationDir = if (Test-Path $(Join-Path $(Resolve-Path '.') 'index')) {Join-Path '.' 'index' -resolve} else {(New-Item 'index' -ItemType 'Directory').fullname} 4 | $avxVersions = "AVX","AVX2","AVX512","basic" 5 | $cudaVersions = "11.6","11.7","11.8","12.0","12.1","12.2","rocm5.4.2","rocm5.5","rocm5.5.1","rocm5.6.1","cpu" 6 | $packageVersions = (@(62)+66..74+76..85).foreach({"$_".Insert(0,'0.1.')}) + (0..11+14..20+@(23)).foreach({"$_".Insert(0,'0.2.')}) 7 | $pythonVersions = "3.7","3.8","3.9","3.10","3.11" 8 | $supportedSystems = 'linux_x86_64','win_amd64','macosx_11_0_x86_64','macosx_12_0_x86_64','macosx_13_0_x86_64','macosx_14_0_x86_64','macosx_11_0_arm64','macosx_12_0_arm64','macosx_13_0_arm64','macosx_14_0_arm64','macosx_11_0_aarch64','macosx_12_0_aarch64','macosx_13_0_aarch64','macosx_14_0_aarch64' 9 | $wheelSource = 'https://github.com/jllllll/llama-cpp-python-cuBLAS-wheels/releases/download' 10 | $packageName = 'llama_cpp_python' 11 | $packageNameNormalized = 'llama-cpp-python' 12 | $packageNameAlt = 'llama_cpp_python_ggml' 13 | $packageNameAltNormalized = 'llama-cpp-python-ggml' 14 | $packageAltVersions = @("0.1.78") 15 | 16 | $avxVersions.foreach({Set-Variable "$_`Dir" $(if (Test-Path $(Join-Path $destinationDir $_)) {Join-Path $destinationDir $_} else {(New-Item $(Join-Path $destinationDir $_) -ItemType 'Directory').fullname})}) 17 | 18 | $indexContent = "`n`n `n " 19 | Foreach ($avxVersion in $avxVersions) 20 | { 21 | if ($avxVersion -eq 'AVX2') {$wheelURL = $wheelSource.TrimEnd('/') + '/wheels'} else {$wheelURL = $wheelSource.TrimEnd('/') + "/$avxVersion"} 22 | $wheelMacosURL = $wheelSource.TrimEnd('/') + '/metal' 23 | $subIndexContent = "`n`n `n " 24 | ForEach ($cudaVersion in $cudaVersions) 25 | { 26 | if ($cudaVersion.StartsWith('rocm') -and $avxVersion -eq 'AVX512') {continue} 27 | $cu = if ($cudaVersion -in 'cpu' -or $cudaVersion.StartsWith('rocm')) {$cudaVersion} else {'cu' + $cudaVersion.replace('.','')} 28 | if ($cudaVersion -eq 'cpu') {$wheelURL = $wheelSource.TrimEnd('/') + '/cpu'} 29 | if ($cudaVersion.StartsWith('rocm')) {$wheelURL = $wheelSource.TrimEnd('/') + '/rocm'} 30 | $cuContent = "`n`n `n " 31 | $cuContentAlt = "`n`n `n " 32 | ForEach ($packageVersion in $packageVersions) 33 | { 34 | if (($avxVersion -eq 'basic' -or $cudaVersion -eq 'cpu') -and [version]$packageVersion -lt [version]"0.1.70") {continue} 35 | if ($cudaVersion.StartsWith('rocm') -and [version]$packageVersion -lt [version]"0.1.80") {continue} 36 | if ($cudaVersion.StartsWith('rocm') -and $avxVersion -ne 'AVX2' -and [version]$packageVersion -lt [version]"0.2.7") {continue} 37 | ForEach ($pythonVersion in $pythonVersions) 38 | { 39 | if ($cudaVersion.StartsWith('rocm') -or [version]$packageVersion -gt [version]"0.1.85" -and $pythonVersion -eq "3.7") {continue} 40 | $pyVer = $pythonVersion.replace('.','') 41 | ForEach ($supportedSystem in $supportedSystems) 42 | { 43 | $doMacos = $avxVersion -eq 'basic' -and $cudaVersion -eq 'cpu' -and $supportedSystem.contains('macosx') -and (($packageVersion -eq '0.1.85' -and !$supportedSystem.contains('macosx_14_0')) -or [version]$packageVersion -gt [version]'0.2.4') 44 | if ([version]$packageVersion -gt '0.2.20' -and $supportedSystem.contains('macosx_11_0')) {$doMacos = $false} 45 | if ($cudaVersion.StartsWith('rocm') -and [version]$packageVersion -gt [version]"0.2.21" -and $supportedSystem -eq 'win_amd64') {continue} 46 | if ($cudaVersion.StartsWith('rocm') -and $cudaVersion.Split('rocm')[-1] -ne '5.5.1' -and $supportedSystem -eq 'win_amd64') {continue} 47 | if ($cudaVersion.StartsWith('rocm') -and $cudaVersion.Split('rocm')[-1] -eq '5.5.1' -and $supportedSystem -eq 'linux_x86_64') {continue} 48 | if ([version]$packageVersion -gt [version]"0.1.85" -and $supportedSystem -eq 'linux_x86_64') {$supportedSystem = 'manylinux_2_31_x86_64'} 49 | $wheelTag = if ($cudaVersion -eq 'cpu' -and !$supportedSystem.contains('macosx')) {"+cpu$($avxVersion.ToLower())"} elseif ($cudaVersion.StartsWith('rocm') -and $avxVersion -ne 'AVX2') {"+$cu$($avxVersion.ToLower())"} elseif (!$supportedSystem.contains('macosx')) {"+$cu"} else {''} 50 | $wheel = if ($pyVer -eq '37') {"$packageName-$packageVersion$wheelTag-cp$pyVer-cp$pyVer`m-$supportedSystem.whl"} else {"$packageName-$packageVersion$wheelTag-cp$pyVer-cp$pyVer-$supportedSystem.whl"} 51 | $wheelAlt = if ($pyVer -eq '37') {"$packageNameAlt-$packageVersion$wheelTag-cp$pyVer-cp$pyVer`m-$supportedSystem.whl"} else {"$packageNameAlt-$packageVersion$wheelTag-cp$pyVer-cp$pyVer-$supportedSystem.whl"} 52 | if (!$supportedSystem.contains('macosx')) {$cuContent += "$wheel
`n "} elseif ($doMacos) {$cuContent += "$wheel
`n "} 53 | if ($packageVersion -in $packageAltVersions -and !$cudaVersion.StartsWith('rocm') -and !$supportedSystem.contains('macosx')) {$cuContentAlt += "$wheelAlt
`n "} 54 | } 55 | } 56 | $cuContent += "`n " 57 | if ($packageVersion -in $packageAltVersions) {$cuContentAlt += "`n "} 58 | } 59 | $cuDir = if (Test-Path $(Join-Path $(Get-Variable "$avxVersion`Dir").Value "$cu")) {Join-Path $(Get-Variable "$avxVersion`Dir").Value "$cu"} else {(New-Item $(Join-Path $(Get-Variable "$avxVersion`Dir").Value "$cu") -ItemType 'Directory').fullname} 60 | $packageDir = if (Test-Path $(Join-Path $cuDir $packageNameNormalized)) {Join-Path $cuDir $packageNameNormalized} else {(New-Item $(Join-Path $cuDir $packageNameNormalized) -ItemType 'Directory').fullname} 61 | $packageAltDir = if (Test-Path $(Join-Path $cuDir $packageNameAltNormalized)) {Join-Path $cuDir $packageNameAltNormalized} elseif (!$cudaVersion.StartsWith('rocm')) {(New-Item $(Join-Path $cuDir $packageNameAltNormalized) -ItemType 'Directory').fullname} 62 | $cuLabel = if ($cudaVersion -eq 'cpu') {$cudaVersion} elseif ($cudaVersion.StartsWith('rocm')) {'ROCm'+' '+$cudaVersion.Split('rocm')[-1]} else {"CUDA $cudaVersion"} 63 | $subIndexContent += "$cuLabel
`n " 64 | New-Item $(Join-Path $packageDir "index.html") -itemType File -value $($cuContent.TrimEnd() + "`n `n`n") -force > $null 65 | if (!$cudaVersion.StartsWith('rocm')) {New-Item $(Join-Path $packageAltDir "index.html") -itemType File -value $($cuContentAlt.TrimEnd() + "`n `n`n") -force > $null} 66 | $packageIndexContent = "`n`n `n $packageName
`n" 67 | if (!$cudaVersion.StartsWith('rocm')) {$packageIndexContent += " $packageNameAlt`n"} 68 | New-Item $(Join-Path $cuDir "index.html") -itemType File -value $($packageIndexContent + " `n`n") -force > $null 69 | if ($avxVersion -eq 'AVX2') {New-Item $(Join-Path $destinationDir "$cu.html") -itemType File -value $($cuContent.TrimEnd() + "`n `n`n") -force > $null} 70 | } 71 | $indexContent += "$avxVersion
`n " 72 | New-Item $(Join-Path $(Get-Variable "$avxVersion`Dir").Value "index.html") -itemType File -value $($subIndexContent.TrimEnd() + "`n `n`n") -force > $null 73 | } 74 | New-Item $(Join-Path $destinationDir "index.html") -itemType File -value $($indexContent.TrimEnd() + "`n `n`n") -force > $null 75 | #"`n`n `n `n `n `n CUDA $cudaVersion
`n `n" 76 | 77 | pause 78 | -------------------------------------------------------------------------------- /index/AVX/rocm5.5.1/llama-cpp-python/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python-0.2.7+rocm5.5.1avx-cp38-cp38-win_amd64.whl
5 | llama_cpp_python-0.2.7+rocm5.5.1avx-cp39-cp39-win_amd64.whl
6 | llama_cpp_python-0.2.7+rocm5.5.1avx-cp310-cp310-win_amd64.whl
7 | llama_cpp_python-0.2.7+rocm5.5.1avx-cp311-cp311-win_amd64.whl
8 | 9 | llama_cpp_python-0.2.8+rocm5.5.1avx-cp38-cp38-win_amd64.whl
10 | llama_cpp_python-0.2.8+rocm5.5.1avx-cp39-cp39-win_amd64.whl
11 | llama_cpp_python-0.2.8+rocm5.5.1avx-cp310-cp310-win_amd64.whl
12 | llama_cpp_python-0.2.8+rocm5.5.1avx-cp311-cp311-win_amd64.whl
13 | 14 | llama_cpp_python-0.2.9+rocm5.5.1avx-cp38-cp38-win_amd64.whl
15 | llama_cpp_python-0.2.9+rocm5.5.1avx-cp39-cp39-win_amd64.whl
16 | llama_cpp_python-0.2.9+rocm5.5.1avx-cp310-cp310-win_amd64.whl
17 | llama_cpp_python-0.2.9+rocm5.5.1avx-cp311-cp311-win_amd64.whl
18 | 19 | llama_cpp_python-0.2.10+rocm5.5.1avx-cp38-cp38-win_amd64.whl
20 | llama_cpp_python-0.2.10+rocm5.5.1avx-cp39-cp39-win_amd64.whl
21 | llama_cpp_python-0.2.10+rocm5.5.1avx-cp310-cp310-win_amd64.whl
22 | llama_cpp_python-0.2.10+rocm5.5.1avx-cp311-cp311-win_amd64.whl
23 | 24 | llama_cpp_python-0.2.11+rocm5.5.1avx-cp38-cp38-win_amd64.whl
25 | llama_cpp_python-0.2.11+rocm5.5.1avx-cp39-cp39-win_amd64.whl
26 | llama_cpp_python-0.2.11+rocm5.5.1avx-cp310-cp310-win_amd64.whl
27 | llama_cpp_python-0.2.11+rocm5.5.1avx-cp311-cp311-win_amd64.whl
28 | 29 | llama_cpp_python-0.2.14+rocm5.5.1avx-cp38-cp38-win_amd64.whl
30 | llama_cpp_python-0.2.14+rocm5.5.1avx-cp39-cp39-win_amd64.whl
31 | llama_cpp_python-0.2.14+rocm5.5.1avx-cp310-cp310-win_amd64.whl
32 | llama_cpp_python-0.2.14+rocm5.5.1avx-cp311-cp311-win_amd64.whl
33 | 34 | llama_cpp_python-0.2.15+rocm5.5.1avx-cp38-cp38-win_amd64.whl
35 | llama_cpp_python-0.2.15+rocm5.5.1avx-cp39-cp39-win_amd64.whl
36 | llama_cpp_python-0.2.15+rocm5.5.1avx-cp310-cp310-win_amd64.whl
37 | llama_cpp_python-0.2.15+rocm5.5.1avx-cp311-cp311-win_amd64.whl
38 | 39 | llama_cpp_python-0.2.16+rocm5.5.1avx-cp38-cp38-win_amd64.whl
40 | llama_cpp_python-0.2.16+rocm5.5.1avx-cp39-cp39-win_amd64.whl
41 | llama_cpp_python-0.2.16+rocm5.5.1avx-cp310-cp310-win_amd64.whl
42 | llama_cpp_python-0.2.16+rocm5.5.1avx-cp311-cp311-win_amd64.whl
43 | 44 | llama_cpp_python-0.2.17+rocm5.5.1avx-cp38-cp38-win_amd64.whl
45 | llama_cpp_python-0.2.17+rocm5.5.1avx-cp39-cp39-win_amd64.whl
46 | llama_cpp_python-0.2.17+rocm5.5.1avx-cp310-cp310-win_amd64.whl
47 | llama_cpp_python-0.2.17+rocm5.5.1avx-cp311-cp311-win_amd64.whl
48 | 49 | llama_cpp_python-0.2.18+rocm5.5.1avx-cp38-cp38-win_amd64.whl
50 | llama_cpp_python-0.2.18+rocm5.5.1avx-cp39-cp39-win_amd64.whl
51 | llama_cpp_python-0.2.18+rocm5.5.1avx-cp310-cp310-win_amd64.whl
52 | llama_cpp_python-0.2.18+rocm5.5.1avx-cp311-cp311-win_amd64.whl
53 | 54 | llama_cpp_python-0.2.19+rocm5.5.1avx-cp38-cp38-win_amd64.whl
55 | llama_cpp_python-0.2.19+rocm5.5.1avx-cp39-cp39-win_amd64.whl
56 | llama_cpp_python-0.2.19+rocm5.5.1avx-cp310-cp310-win_amd64.whl
57 | llama_cpp_python-0.2.19+rocm5.5.1avx-cp311-cp311-win_amd64.whl
58 | 59 | llama_cpp_python-0.2.20+rocm5.5.1avx-cp38-cp38-win_amd64.whl
60 | llama_cpp_python-0.2.20+rocm5.5.1avx-cp39-cp39-win_amd64.whl
61 | llama_cpp_python-0.2.20+rocm5.5.1avx-cp310-cp310-win_amd64.whl
62 | llama_cpp_python-0.2.20+rocm5.5.1avx-cp311-cp311-win_amd64.whl
63 | 64 | 65 | -------------------------------------------------------------------------------- /index/basic/rocm5.5.1/llama-cpp-python/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python-0.2.7+rocm5.5.1basic-cp38-cp38-win_amd64.whl
5 | llama_cpp_python-0.2.7+rocm5.5.1basic-cp39-cp39-win_amd64.whl
6 | llama_cpp_python-0.2.7+rocm5.5.1basic-cp310-cp310-win_amd64.whl
7 | llama_cpp_python-0.2.7+rocm5.5.1basic-cp311-cp311-win_amd64.whl
8 | 9 | llama_cpp_python-0.2.8+rocm5.5.1basic-cp38-cp38-win_amd64.whl
10 | llama_cpp_python-0.2.8+rocm5.5.1basic-cp39-cp39-win_amd64.whl
11 | llama_cpp_python-0.2.8+rocm5.5.1basic-cp310-cp310-win_amd64.whl
12 | llama_cpp_python-0.2.8+rocm5.5.1basic-cp311-cp311-win_amd64.whl
13 | 14 | llama_cpp_python-0.2.9+rocm5.5.1basic-cp38-cp38-win_amd64.whl
15 | llama_cpp_python-0.2.9+rocm5.5.1basic-cp39-cp39-win_amd64.whl
16 | llama_cpp_python-0.2.9+rocm5.5.1basic-cp310-cp310-win_amd64.whl
17 | llama_cpp_python-0.2.9+rocm5.5.1basic-cp311-cp311-win_amd64.whl
18 | 19 | llama_cpp_python-0.2.10+rocm5.5.1basic-cp38-cp38-win_amd64.whl
20 | llama_cpp_python-0.2.10+rocm5.5.1basic-cp39-cp39-win_amd64.whl
21 | llama_cpp_python-0.2.10+rocm5.5.1basic-cp310-cp310-win_amd64.whl
22 | llama_cpp_python-0.2.10+rocm5.5.1basic-cp311-cp311-win_amd64.whl
23 | 24 | llama_cpp_python-0.2.11+rocm5.5.1basic-cp38-cp38-win_amd64.whl
25 | llama_cpp_python-0.2.11+rocm5.5.1basic-cp39-cp39-win_amd64.whl
26 | llama_cpp_python-0.2.11+rocm5.5.1basic-cp310-cp310-win_amd64.whl
27 | llama_cpp_python-0.2.11+rocm5.5.1basic-cp311-cp311-win_amd64.whl
28 | 29 | llama_cpp_python-0.2.14+rocm5.5.1basic-cp38-cp38-win_amd64.whl
30 | llama_cpp_python-0.2.14+rocm5.5.1basic-cp39-cp39-win_amd64.whl
31 | llama_cpp_python-0.2.14+rocm5.5.1basic-cp310-cp310-win_amd64.whl
32 | llama_cpp_python-0.2.14+rocm5.5.1basic-cp311-cp311-win_amd64.whl
33 | 34 | llama_cpp_python-0.2.15+rocm5.5.1basic-cp38-cp38-win_amd64.whl
35 | llama_cpp_python-0.2.15+rocm5.5.1basic-cp39-cp39-win_amd64.whl
36 | llama_cpp_python-0.2.15+rocm5.5.1basic-cp310-cp310-win_amd64.whl
37 | llama_cpp_python-0.2.15+rocm5.5.1basic-cp311-cp311-win_amd64.whl
38 | 39 | llama_cpp_python-0.2.16+rocm5.5.1basic-cp38-cp38-win_amd64.whl
40 | llama_cpp_python-0.2.16+rocm5.5.1basic-cp39-cp39-win_amd64.whl
41 | llama_cpp_python-0.2.16+rocm5.5.1basic-cp310-cp310-win_amd64.whl
42 | llama_cpp_python-0.2.16+rocm5.5.1basic-cp311-cp311-win_amd64.whl
43 | 44 | llama_cpp_python-0.2.17+rocm5.5.1basic-cp38-cp38-win_amd64.whl
45 | llama_cpp_python-0.2.17+rocm5.5.1basic-cp39-cp39-win_amd64.whl
46 | llama_cpp_python-0.2.17+rocm5.5.1basic-cp310-cp310-win_amd64.whl
47 | llama_cpp_python-0.2.17+rocm5.5.1basic-cp311-cp311-win_amd64.whl
48 | 49 | llama_cpp_python-0.2.18+rocm5.5.1basic-cp38-cp38-win_amd64.whl
50 | llama_cpp_python-0.2.18+rocm5.5.1basic-cp39-cp39-win_amd64.whl
51 | llama_cpp_python-0.2.18+rocm5.5.1basic-cp310-cp310-win_amd64.whl
52 | llama_cpp_python-0.2.18+rocm5.5.1basic-cp311-cp311-win_amd64.whl
53 | 54 | llama_cpp_python-0.2.19+rocm5.5.1basic-cp38-cp38-win_amd64.whl
55 | llama_cpp_python-0.2.19+rocm5.5.1basic-cp39-cp39-win_amd64.whl
56 | llama_cpp_python-0.2.19+rocm5.5.1basic-cp310-cp310-win_amd64.whl
57 | llama_cpp_python-0.2.19+rocm5.5.1basic-cp311-cp311-win_amd64.whl
58 | 59 | llama_cpp_python-0.2.20+rocm5.5.1basic-cp38-cp38-win_amd64.whl
60 | llama_cpp_python-0.2.20+rocm5.5.1basic-cp39-cp39-win_amd64.whl
61 | llama_cpp_python-0.2.20+rocm5.5.1basic-cp310-cp310-win_amd64.whl
62 | llama_cpp_python-0.2.20+rocm5.5.1basic-cp311-cp311-win_amd64.whl
63 | 64 | 65 | -------------------------------------------------------------------------------- /index/textgen/AVX/rocm5.5.1/llama-cpp-python-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda-0.2.7+rocm5.5.1avx-cp38-cp38-win_amd64.whl
5 | llama_cpp_python_cuda-0.2.7+rocm5.5.1avx-cp39-cp39-win_amd64.whl
6 | llama_cpp_python_cuda-0.2.7+rocm5.5.1avx-cp310-cp310-win_amd64.whl
7 | llama_cpp_python_cuda-0.2.7+rocm5.5.1avx-cp311-cp311-win_amd64.whl
8 | 9 | llama_cpp_python_cuda-0.2.8+rocm5.5.1avx-cp38-cp38-win_amd64.whl
10 | llama_cpp_python_cuda-0.2.8+rocm5.5.1avx-cp39-cp39-win_amd64.whl
11 | llama_cpp_python_cuda-0.2.8+rocm5.5.1avx-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_cuda-0.2.8+rocm5.5.1avx-cp311-cp311-win_amd64.whl
13 | 14 | llama_cpp_python_cuda-0.2.9+rocm5.5.1avx-cp38-cp38-win_amd64.whl
15 | llama_cpp_python_cuda-0.2.9+rocm5.5.1avx-cp39-cp39-win_amd64.whl
16 | llama_cpp_python_cuda-0.2.9+rocm5.5.1avx-cp310-cp310-win_amd64.whl
17 | llama_cpp_python_cuda-0.2.9+rocm5.5.1avx-cp311-cp311-win_amd64.whl
18 | 19 | llama_cpp_python_cuda-0.2.10+rocm5.5.1avx-cp38-cp38-win_amd64.whl
20 | llama_cpp_python_cuda-0.2.10+rocm5.5.1avx-cp39-cp39-win_amd64.whl
21 | llama_cpp_python_cuda-0.2.10+rocm5.5.1avx-cp310-cp310-win_amd64.whl
22 | llama_cpp_python_cuda-0.2.10+rocm5.5.1avx-cp311-cp311-win_amd64.whl
23 | 24 | llama_cpp_python_cuda-0.2.11+rocm5.5.1avx-cp38-cp38-win_amd64.whl
25 | llama_cpp_python_cuda-0.2.11+rocm5.5.1avx-cp39-cp39-win_amd64.whl
26 | llama_cpp_python_cuda-0.2.11+rocm5.5.1avx-cp310-cp310-win_amd64.whl
27 | llama_cpp_python_cuda-0.2.11+rocm5.5.1avx-cp311-cp311-win_amd64.whl
28 | 29 | llama_cpp_python_cuda-0.2.14+rocm5.5.1avx-cp38-cp38-win_amd64.whl
30 | llama_cpp_python_cuda-0.2.14+rocm5.5.1avx-cp39-cp39-win_amd64.whl
31 | llama_cpp_python_cuda-0.2.14+rocm5.5.1avx-cp310-cp310-win_amd64.whl
32 | llama_cpp_python_cuda-0.2.14+rocm5.5.1avx-cp311-cp311-win_amd64.whl
33 | 34 | llama_cpp_python_cuda-0.2.15+rocm5.5.1avx-cp38-cp38-win_amd64.whl
35 | llama_cpp_python_cuda-0.2.15+rocm5.5.1avx-cp39-cp39-win_amd64.whl
36 | llama_cpp_python_cuda-0.2.15+rocm5.5.1avx-cp310-cp310-win_amd64.whl
37 | llama_cpp_python_cuda-0.2.15+rocm5.5.1avx-cp311-cp311-win_amd64.whl
38 | 39 | llama_cpp_python_cuda-0.2.16+rocm5.5.1avx-cp38-cp38-win_amd64.whl
40 | llama_cpp_python_cuda-0.2.16+rocm5.5.1avx-cp39-cp39-win_amd64.whl
41 | llama_cpp_python_cuda-0.2.16+rocm5.5.1avx-cp310-cp310-win_amd64.whl
42 | llama_cpp_python_cuda-0.2.16+rocm5.5.1avx-cp311-cp311-win_amd64.whl
43 | 44 | llama_cpp_python_cuda-0.2.17+rocm5.5.1avx-cp38-cp38-win_amd64.whl
45 | llama_cpp_python_cuda-0.2.17+rocm5.5.1avx-cp39-cp39-win_amd64.whl
46 | llama_cpp_python_cuda-0.2.17+rocm5.5.1avx-cp310-cp310-win_amd64.whl
47 | llama_cpp_python_cuda-0.2.17+rocm5.5.1avx-cp311-cp311-win_amd64.whl
48 | 49 | llama_cpp_python_cuda-0.2.18+rocm5.5.1avx-cp38-cp38-win_amd64.whl
50 | llama_cpp_python_cuda-0.2.18+rocm5.5.1avx-cp39-cp39-win_amd64.whl
51 | llama_cpp_python_cuda-0.2.18+rocm5.5.1avx-cp310-cp310-win_amd64.whl
52 | llama_cpp_python_cuda-0.2.18+rocm5.5.1avx-cp311-cp311-win_amd64.whl
53 | 54 | llama_cpp_python_cuda-0.2.19+rocm5.5.1avx-cp38-cp38-win_amd64.whl
55 | llama_cpp_python_cuda-0.2.19+rocm5.5.1avx-cp39-cp39-win_amd64.whl
56 | llama_cpp_python_cuda-0.2.19+rocm5.5.1avx-cp310-cp310-win_amd64.whl
57 | llama_cpp_python_cuda-0.2.19+rocm5.5.1avx-cp311-cp311-win_amd64.whl
58 | 59 | llama_cpp_python_cuda-0.2.20+rocm5.5.1avx-cp38-cp38-win_amd64.whl
60 | llama_cpp_python_cuda-0.2.20+rocm5.5.1avx-cp39-cp39-win_amd64.whl
61 | llama_cpp_python_cuda-0.2.20+rocm5.5.1avx-cp310-cp310-win_amd64.whl
62 | llama_cpp_python_cuda-0.2.20+rocm5.5.1avx-cp311-cp311-win_amd64.whl
63 | 64 | 65 | -------------------------------------------------------------------------------- /index/textgen/basic/rocm5.5.1/llama-cpp-python-cuda/index.html: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | llama_cpp_python_cuda-0.2.7+rocm5.5.1basic-cp38-cp38-win_amd64.whl
5 | llama_cpp_python_cuda-0.2.7+rocm5.5.1basic-cp39-cp39-win_amd64.whl
6 | llama_cpp_python_cuda-0.2.7+rocm5.5.1basic-cp310-cp310-win_amd64.whl
7 | llama_cpp_python_cuda-0.2.7+rocm5.5.1basic-cp311-cp311-win_amd64.whl
8 | 9 | llama_cpp_python_cuda-0.2.8+rocm5.5.1basic-cp38-cp38-win_amd64.whl
10 | llama_cpp_python_cuda-0.2.8+rocm5.5.1basic-cp39-cp39-win_amd64.whl
11 | llama_cpp_python_cuda-0.2.8+rocm5.5.1basic-cp310-cp310-win_amd64.whl
12 | llama_cpp_python_cuda-0.2.8+rocm5.5.1basic-cp311-cp311-win_amd64.whl
13 | 14 | llama_cpp_python_cuda-0.2.9+rocm5.5.1basic-cp38-cp38-win_amd64.whl
15 | llama_cpp_python_cuda-0.2.9+rocm5.5.1basic-cp39-cp39-win_amd64.whl
16 | llama_cpp_python_cuda-0.2.9+rocm5.5.1basic-cp310-cp310-win_amd64.whl
17 | llama_cpp_python_cuda-0.2.9+rocm5.5.1basic-cp311-cp311-win_amd64.whl
18 | 19 | llama_cpp_python_cuda-0.2.10+rocm5.5.1basic-cp38-cp38-win_amd64.whl
20 | llama_cpp_python_cuda-0.2.10+rocm5.5.1basic-cp39-cp39-win_amd64.whl
21 | llama_cpp_python_cuda-0.2.10+rocm5.5.1basic-cp310-cp310-win_amd64.whl
22 | llama_cpp_python_cuda-0.2.10+rocm5.5.1basic-cp311-cp311-win_amd64.whl
23 | 24 | llama_cpp_python_cuda-0.2.11+rocm5.5.1basic-cp38-cp38-win_amd64.whl
25 | llama_cpp_python_cuda-0.2.11+rocm5.5.1basic-cp39-cp39-win_amd64.whl
26 | llama_cpp_python_cuda-0.2.11+rocm5.5.1basic-cp310-cp310-win_amd64.whl
27 | llama_cpp_python_cuda-0.2.11+rocm5.5.1basic-cp311-cp311-win_amd64.whl
28 | 29 | llama_cpp_python_cuda-0.2.14+rocm5.5.1basic-cp38-cp38-win_amd64.whl
30 | llama_cpp_python_cuda-0.2.14+rocm5.5.1basic-cp39-cp39-win_amd64.whl
31 | llama_cpp_python_cuda-0.2.14+rocm5.5.1basic-cp310-cp310-win_amd64.whl
32 | llama_cpp_python_cuda-0.2.14+rocm5.5.1basic-cp311-cp311-win_amd64.whl
33 | 34 | llama_cpp_python_cuda-0.2.15+rocm5.5.1basic-cp38-cp38-win_amd64.whl
35 | llama_cpp_python_cuda-0.2.15+rocm5.5.1basic-cp39-cp39-win_amd64.whl
36 | llama_cpp_python_cuda-0.2.15+rocm5.5.1basic-cp310-cp310-win_amd64.whl
37 | llama_cpp_python_cuda-0.2.15+rocm5.5.1basic-cp311-cp311-win_amd64.whl
38 | 39 | llama_cpp_python_cuda-0.2.16+rocm5.5.1basic-cp38-cp38-win_amd64.whl
40 | llama_cpp_python_cuda-0.2.16+rocm5.5.1basic-cp39-cp39-win_amd64.whl
41 | llama_cpp_python_cuda-0.2.16+rocm5.5.1basic-cp310-cp310-win_amd64.whl
42 | llama_cpp_python_cuda-0.2.16+rocm5.5.1basic-cp311-cp311-win_amd64.whl
43 | 44 | llama_cpp_python_cuda-0.2.17+rocm5.5.1basic-cp38-cp38-win_amd64.whl
45 | llama_cpp_python_cuda-0.2.17+rocm5.5.1basic-cp39-cp39-win_amd64.whl
46 | llama_cpp_python_cuda-0.2.17+rocm5.5.1basic-cp310-cp310-win_amd64.whl
47 | llama_cpp_python_cuda-0.2.17+rocm5.5.1basic-cp311-cp311-win_amd64.whl
48 | 49 | llama_cpp_python_cuda-0.2.18+rocm5.5.1basic-cp38-cp38-win_amd64.whl
50 | llama_cpp_python_cuda-0.2.18+rocm5.5.1basic-cp39-cp39-win_amd64.whl
51 | llama_cpp_python_cuda-0.2.18+rocm5.5.1basic-cp310-cp310-win_amd64.whl
52 | llama_cpp_python_cuda-0.2.18+rocm5.5.1basic-cp311-cp311-win_amd64.whl
53 | 54 | llama_cpp_python_cuda-0.2.19+rocm5.5.1basic-cp38-cp38-win_amd64.whl
55 | llama_cpp_python_cuda-0.2.19+rocm5.5.1basic-cp39-cp39-win_amd64.whl
56 | llama_cpp_python_cuda-0.2.19+rocm5.5.1basic-cp310-cp310-win_amd64.whl
57 | llama_cpp_python_cuda-0.2.19+rocm5.5.1basic-cp311-cp311-win_amd64.whl
58 | 59 | llama_cpp_python_cuda-0.2.20+rocm5.5.1basic-cp38-cp38-win_amd64.whl
60 | llama_cpp_python_cuda-0.2.20+rocm5.5.1basic-cp39-cp39-win_amd64.whl
61 | llama_cpp_python_cuda-0.2.20+rocm5.5.1basic-cp310-cp310-win_amd64.whl
62 | llama_cpp_python_cuda-0.2.20+rocm5.5.1basic-cp311-cp311-win_amd64.whl
63 | 64 | 65 | -------------------------------------------------------------------------------- /.github/workflows/build-wheels-oobabooga.yml: -------------------------------------------------------------------------------- 1 | name: Build CUDA Wheels 2 | 3 | on: 4 | workflow_dispatch: 5 | inputs: 6 | version: 7 | description: 'Version tag of llama-cpp-python to build: v0.2.14' 8 | default: 'v0.2.14' 9 | required: true 10 | type: string 11 | config: 12 | description: 'Override configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 13 | default: 'Default' 14 | required: false 15 | type: string 16 | exclude: 17 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 18 | default: 'None' 19 | required: false 20 | type: string 21 | workflow_call: 22 | inputs: 23 | version: 24 | description: 'Version tag of llama-cpp-python to build: v0.2.14' 25 | default: 'v0.2.14' 26 | required: true 27 | type: string 28 | config: 29 | description: 'Configurations to build: key1:item1-1,item1-2;key2:item2-1,item2-2' 30 | default: 'Default' 31 | required: false 32 | type: string 33 | exclude: 34 | description: 'Exclude build configurations: key1-1:item1-1,key1-2:item1-2;key2-1:item2-1,key2-2:item2-2' 35 | default: 'None' 36 | required: false 37 | type: string 38 | 39 | permissions: 40 | contents: write 41 | 42 | jobs: 43 | define_matrix: 44 | name: Define Build Matrix 45 | runs-on: ubuntu-latest 46 | outputs: 47 | matrix: ${{ steps.set-matrix.outputs.matrix }} 48 | defaults: 49 | run: 50 | shell: pwsh 51 | env: 52 | CONFIGIN: ${{ inputs.config }} 53 | EXCLUDEIN: ${{ inputs.exclude }} 54 | 55 | steps: 56 | - name: Define Job Output 57 | id: set-matrix 58 | run: | 59 | $matrix = @{ 60 | 'os' = @('ubuntu-22.04', 'windows-2019') 61 | 'pyver' = @("3.11") 62 | 'cuda' = @("12.4.1") 63 | 'avxver' = @("AVX2", "AVX") 64 | } 65 | 66 | if ($env:CONFIGIN -ne 'Default') {$env:CONFIGIN.split(';').foreach({$matrix[$_.split(':')[0]] = $_.split(':')[1].split(',')})} 67 | 68 | if ($env:EXCLUDEIN -ne 'None') { 69 | $exclusions = @() 70 | $exclusions += $env:EXCLUDEIN.split(';').replace(':','=').replace(',',"`n") | ConvertFrom-StringData 71 | $matrix['exclude'] = $exclusions 72 | } 73 | 74 | $matrixOut = ConvertTo-Json $matrix -Compress 75 | Write-Output ('matrix=' + $matrixOut) >> $env:GITHUB_OUTPUT 76 | 77 | build_wheels: 78 | name: Build Wheel ${{ matrix.os }} ${{ matrix.pyver }} ${{ matrix.cuda }} ${{ matrix.avxver }} 79 | needs: define_matrix 80 | runs-on: ${{ matrix.os }} 81 | strategy: 82 | matrix: ${{ fromJSON(needs.define_matrix.outputs.matrix) }} 83 | defaults: 84 | run: 85 | shell: pwsh 86 | env: 87 | CUDAVER: ${{ matrix.cuda }} 88 | AVXVER: ${{ matrix.avxver }} 89 | PCKGVER: ${{ inputs.version }} 90 | 91 | steps: 92 | - name: Add MSBuild to PATH 93 | if: runner.os == 'Windows' 94 | uses: microsoft/setup-msbuild@v1.1 95 | with: 96 | vs-version: '[16.11,16.12)' 97 | 98 | - uses: actions/checkout@v4 99 | with: 100 | repository: 'abetlen/llama-cpp-python' 101 | ref: ${{ inputs.version }} 102 | submodules: 'recursive' 103 | 104 | - uses: actions/setup-python@v4 105 | with: 106 | python-version: ${{ matrix.pyver }} 107 | 108 | - name: Setup Mamba 109 | uses: conda-incubator/setup-miniconda@v3.1.0 110 | with: 111 | activate-environment: "TGW" 112 | python-version: ${{ matrix.pyver }} 113 | miniforge-version: latest 114 | add-pip-as-python-dependency: true 115 | auto-activate-base: false 116 | 117 | - name: VS Integration Cache 118 | id: vs-integration-cache 119 | if: runner.os == 'Windows' 120 | uses: actions/cache@v4 121 | with: 122 | path: ./MSBuildExtensions 123 | key: cuda-${{ matrix.cuda }}-vs-integration 124 | 125 | - name: Get Visual Studio Integration 126 | if: runner.os == 'Windows' && steps.vs-integration-cache.outputs.cache-hit != 'true' 127 | run: | 128 | if ($env:CUDAVER -eq '12.1.1') {$x = '12.1.0'} else {$x = $env:CUDAVER} 129 | $links = (Invoke-RestMethod 'https://raw.githubusercontent.com/Jimver/cuda-toolkit/95b5a29a2823ef87666dcd45ec236fc4f82498d0/src/links/windows-links.ts').Trim().split().where({$_ -ne ''}) 130 | for ($i=$q=0;$i -lt $links.count -and $q -lt 2;$i++) {if ($links[$i] -eq "'$x',") {$q++}} 131 | Invoke-RestMethod $links[$i].Trim("'") -OutFile 'cudainstaller.zip' 132 | & 'C:\Program Files\7-Zip\7z.exe' e cudainstaller.zip -oMSBuildExtensions -r *\MSBuildExtensions\* > $null 133 | Remove-Item 'cudainstaller.zip' 134 | 135 | - name: Install Visual Studio Integration 136 | if: runner.os == 'Windows' 137 | run: | 138 | $y = (gi '.\MSBuildExtensions').fullname + '\*' 139 | (gi 'C:\Program Files (x86)\Microsoft Visual Studio\2019\Enterprise\MSBuild\Microsoft\VC\*\BuildCustomizations').fullname.foreach({cp $y $_}) 140 | $cupath = 'CUDA_PATH_V' + $env:CUDAVER.Remove($env:CUDAVER.LastIndexOf('.')).Replace('.','_') 141 | echo "$cupath=$env:CONDA_PREFIX" >> $env:GITHUB_ENV 142 | 143 | - name: Install Dependencies 144 | run: | 145 | $cudaVersion = $env:CUDAVER 146 | $cudaChannels = '' 147 | $cudaNum = [int]$cudaVersion.substring($cudaVersion.LastIndexOf('.')+1) 148 | while ($cudaNum -ge 0) { $cudaChannels += '-c nvidia/label/cuda-' + $cudaVersion.Remove($cudaVersion.LastIndexOf('.')+1) + $cudaNum + ' '; $cudaNum-- } 149 | mamba install -y 'cuda' $cudaChannels.TrimEnd().Split() 150 | if (!(mamba list cuda)[-1].contains('cuda')) {sleep -s 10; mamba install -y 'cuda' $cudaChannels.TrimEnd().Split()} 151 | if (!(mamba list cuda)[-1].contains('cuda')) {throw 'CUDA Toolkit failed to install!'} 152 | 153 | python -m pip install build wheel 154 | 155 | - name: Change Package Name 156 | run: | 157 | $packageVersion = [version]$env:PCKGVER.TrimStart('v') 158 | $pyproject = Get-Content 'pyproject.toml' -raw 159 | $cmakelists = Get-Content 'CMakeLists.txt' -raw 160 | if ($packageVersion -lt [version]'0.2.0') { 161 | $setup = Get-Content 'setup.py' -raw 162 | $regexstr = '(?s)name="llama_cpp_python",(.+)(package_dir={"llama_cpp": "llama_cpp", "llama_cpp.server": "llama_cpp/server"},.+?packages=\["llama_cpp", "llama_cpp.server"],)' 163 | if ($packageVersion -gt [version]'0.1.77') {$regexstr = '(?s)name="llama_cpp_python",(.+)(package_dir={"llama_cpp": "llama_cpp", "llama_cpp.server": "llama_cpp/server"},.+?package_data={"llama_cpp": \["py.typed"]},.+?packages=\["llama_cpp", "llama_cpp.server"],)'} 164 | $regexmatch = [Regex]::Matches($setup,$regexstr) 165 | if (!($regexmatch[0].Success)) {throw 'setup.py parsing failed'} 166 | $newstr = 'name="llama_cpp_python_cuda",' + $regexmatch[0].Groups[1].Value + $regexmatch[0].Groups[2].Value.Replace('llama_cpp','llama_cpp_cuda') 167 | $newsetup = $regexmatch[0].Result(('$`'+$newstr+'$''')) 168 | New-Item 'setup.py' -itemType File -value $newsetup -force 169 | $regexstr = '(?s)(?<=name = ")llama_cpp_python(".+?packages = \[{include = ")llama_cpp(".+)' 170 | $regexmatch = [Regex]::Matches($pyproject,$regexstr) 171 | if (!($regexmatch[0].Success)) {throw 'pyproject.toml parsing failed'} 172 | $newpyproject = $regexmatch[0].Result(('$`'+'llama_cpp_python_cuda'+'$1llama_cpp_cuda$2')) 173 | } else { 174 | $regexstr = '(?s)(?<=\[project]\s+?name = ")llama_cpp_python(".+?all = \[\s+?")llama_cpp_python(\[.+?wheel.packages = \[")llama_cpp("].+?input = ")llama_cpp(?=/__init__.py")' 175 | $regexmatch = [Regex]::Matches($pyproject,$regexstr) 176 | if (!($regexmatch[0].Success)) {throw 'pyproject.toml parsing failed'} 177 | $newpyproject = $regexmatch[0].Result(('$`' + 'llama_cpp_python_cuda' + '$1llama_cpp_cuda$2' + 'llama_cpp_cuda$3llama_cpp_cuda' + '$''')) 178 | } 179 | Copy-Item 'llama_cpp' 'llama_cpp_cuda' -recurse 180 | New-Item 'pyproject.toml' -itemType File -value $newpyproject -force 181 | New-Item 'CMakeLists.txt' -itemType File -value $cmakelists.Replace('llama_cpp','llama_cpp_cuda') -force 182 | 183 | if ($packageVersion -gt [version]'0.2.13') 184 | { 185 | $pyScripts = (Get-ChildItem $(Join-Path '.' 'llama_cpp_cuda' '*.py')) 186 | $pyScripts.fullname.foreach({ 187 | $content = Get-Content $_ -raw 188 | $content = $content -replace '(?m)^import llama_cpp\.', 'from . import ' 189 | $content = $content -replace '(?m)^from llama_cpp\.', 'from .' 190 | $content = $content -replace '(?m)^(\s*)import llama_cpp(\s*)$', '${1}import llama_cpp_cuda as llama_cpp${2}' 191 | $content | Set-Content $_ -NoNewline 192 | }) 193 | } 194 | 195 | - name: Build Wheel 196 | run: | 197 | $packageVersion = [version]$env:PCKGVER.TrimStart('v') 198 | $cudaVersion = $env:CUDAVER.Remove($env:CUDAVER.LastIndexOf('.')).Replace('.','') 199 | $env:CUDA_PATH = $env:CONDA_PREFIX 200 | $env:CUDA_HOME = $env:CONDA_PREFIX 201 | if ($IsLinux) {$env:LD_LIBRARY_PATH = $env:CONDA_PREFIX + '/lib:' + $env:LD_LIBRARY_PATH} 202 | $env:VERBOSE = '1' 203 | $env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=all' 204 | if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.71') {$env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=35-real;37-real;52;61-real;70-real;72-real;75-real;80-real;86-real;89-real;90'} 205 | if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.71' -and [version]$env:CUDAVER -ge [version]'12.0') {$env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=52;61-real;70-real;72-real;75-real;80-real;86-real;89-real;90'} 206 | if ($packageVersion -gt [version]'0.1.68' -and $packageVersion -lt [version]'0.1.71' -and [version]$env:CUDAVER -lt [version]'11.8') {$env:CMAKE_ARGS = '-DGGML_CUDA=on -DCMAKE_CUDA_ARCHITECTURES=35-real;37-real;52;61-real;70-real;72-real;75-real;80-real;86'} 207 | if ($packageVersion -gt [version]'0.2.13') {$env:CMAKE_ARGS = "-DGGML_NATIVE=off -DGGML_CUDA_FORCE_MMQ=ON $env:CMAKE_ARGS"} 208 | if ($packageVersion -lt [version]'0.1.66') {$env:CUDAFLAGS = '-arch=all'} 209 | if ($env:AVXVER -eq 'AVX') {$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off'; $cputag = 'avx'} 210 | if ($env:AVXVER -eq 'basic') {$env:CMAKE_ARGS = $env:CMAKE_ARGS + ' -DGGML_AVX=off -DGGML_AVX2=off -DGGML_FMA=off -DGGML_F16C=off'; $cputag = 'basic'} 211 | $buildtag = "+cu$cudaVersion$cputag" 212 | if ($packageVersion -lt [version]'0.2.0') { 213 | $env:FORCE_CMAKE = '1' 214 | python -m build --wheel -C--build-option=egg_info "-C--build-option=--tag-build=$buildtag" 215 | } else { 216 | $initpath = Join-Path '.' 'llama_cpp_cuda' '__init__.py' -resolve 217 | $initcontent = Get-Content $initpath -raw 218 | $regexstr = '(?s)(?<=__version__ \= ")\d+(?:\.\d+)*(?=")' 219 | $regexmatch = [Regex]::Matches($initcontent,$regexstr) 220 | if (!($regexmatch[0].Success)) {throw '__init__.py parsing failed'} 221 | $newinit = $regexmatch[0].Result(('$`' + '$&' + $buildtag + '$''')) 222 | New-Item $initpath -itemType File -value $newinit -force 223 | python -m build --wheel 224 | } 225 | 226 | - name: Upload files to a GitHub release 227 | id: upload-release 228 | uses: svenstaro/upload-release-action@2.7.0 229 | continue-on-error: true 230 | with: 231 | repo_token: ${{ secrets.GITHUB_TOKEN }} 232 | file: ./dist/*.whl 233 | tag: textgen-webui 234 | file_glob: true 235 | make_latest: false 236 | overwrite: true 237 | 238 | - uses: actions/upload-artifact@v4 239 | if: steps.upload-release.outcome == 'failure' 240 | with: 241 | name: textgen-webui-wheels 242 | path: ./dist/*.whl 243 | --------------------------------------------------------------------------------