localai-git: auto updated to v2.0.0.6.g997119c-1

x86_64/localai-git/PKGBUILD

@@ -0,0 +1,177 @@
+# Maintainer: wuxxin <wuxxin@gmail.com>
+
+# to only build for cpu, set ENABLE_CUDA and ENABLE_ROCM to 0
+_ENABLE_CUDA=1
+_ENABLE_ROCM=0
+_SKIP_CPU=0
+_GO_TAGS=""
+# _GO_TAGS="tts stablediffusion"
+_OPTIONAL_BACKENDS=""
+if test -n "$(echo "$_GO_TAGS" | grep -o "tts")"; then
+  _OPTIONAL_BACKENDS="backend-assets/grpc/piper $_OPTIONAL_BACKENDS"
+fi
+if test -n "$(echo "$_GO_TAGS" | grep -o "stablediffusion")"; then
+  _OPTIONAL_BACKENDS="backend-assets/grpc/stablediffusion $_OPTIONAL_BACKENDS"
+fi
+# list of backends to be build
+_GRPC_BACKENDS="backend-assets/grpc/bert-embeddings backend-assets/grpc/llama-cpp backend-assets/grpc/whisper $_OPTIONAL_BACKENDS"
+_pkgname="localai"
+
+pkgbase="${_pkgname}-git"
+pkgname=("${pkgbase}")
+pkgver=v2.0.0.6.g997119c
+pkgrel=1
+pkgdesc="The free, Open Source OpenAI alternative. Self-hosted, community-driven and local-first."
+url="https://github.com/mudler/LocalAI"
+license=('MIT')
+arch=('x86_64')
+
+provides=('localai')
+conflicts=('localai')
+
+depends=(
+  'grpc'
+  'opencv'
+  'blas-openblas'
+  'sdl2'
+  'ffmpeg'
+)
+makedepends=(
+  'go'
+  'git'
+  'cmake'
+)
+
+if test "$(echo "$_GO_TAGS" | grep -o "tts")" = "tts"; then
+  depends+=(
+    'onnxruntime'
+    'piper-phonemize'
+  )
+fi
+
+if [[ $_ENABLE_CUDA = 1 ]]; then
+  pkgname+=("${pkgbase}-cuda")
+  makedepends+=(
+    'cuda'
+    'cudnn'
+    'nccl'
+    'magma-cuda'
+  )
+fi
+
+if [[ $_ENABLE_ROCM = 1 ]]; then
+  pkgname+=("${pkgbase}-rocm")
+  makedepends+=(
+    'rocm-hip-sdk'
+    'miopen-hip'
+    'rccl'
+    'magma-hip'
+  )
+fi
+
+source=(
+  "${_pkgname}"::"git+https://github.com/mudler/LocalAI"
+)
+
+sha256sums=(
+  'SKIP'
+)
+
+pkgver() {
+  cd "${srcdir}/${_pkgname}"
+  (git describe --always --tags | tr "-" ".")
+}
+
+prepare() {
+  cd "${srcdir}/${_pkgname}"
+
+  # list of backend sources to be recursive git checked out before build()
+  _EXTERNAL_SOURCES="backend/cpp/llama/llama.cpp sources/go-piper sources/whisper.cpp sources/go-bert"
+  # fetch sources for active backends
+  mkdir -p "sources"
+  make $_EXTERNAL_SOURCES
+
+  # modify get-sources, remove go mod edits for inactive backend sources
+  sed -ri "s#get-sources: .*#get-sources: $_EXTERNAL_SOURCES#g" Makefile
+  sed -ri 's#.+\-replace github.com/nomic-ai/gpt4all/gpt4all.+##g' Makefile
+  sed -ri 's#.+\-replace github.com/donomii/go-rwkv.cpp.+##g' Makefile
+  sed -ri 's#.+\-replace github.com/go-skynet/go-ggml-transformers.cpp.+##g' Makefile
+  sed -ri 's#.+\-replace github.com/mudler/go-stable-diffusion.+##g' Makefile
+
+  # # patch stablediffusion
+  # sed -ri "s/^(#include <ncnn\/)(benchmark|net)(\.h>)/\1src\/\2\3/g" \
+  #   sources/go-stable-diffusion/stablediffusion.hpp
+
+  # copy for different build types
+  cd "${srcdir}"
+  for n in "${_pkgname}-cpu" "${_pkgname}-cuda" "${_pkgname}-rocm"; do
+    if test -d "$n"; then rm -rf "$n"; fi
+    cp -r "${_pkgname}" "$n"
+  done
+}
+
+_build() {
+  if test -n "$(echo "$_GO_TAGS" | grep -o "stablediffusion")"; then
+    make BUILD_TYPE="$1" GRPC_BACKENDS="backend-assets/grpc/stablediffusion" GO_TAGS="$_GO_TAGS" build
+  fi
+  make BUILD_TYPE="$1" GRPC_BACKENDS="$_GRPC_BACKENDS" GO_TAGS="$_GO_TAGS" build
+}
+
+build() {
+  if test "$_SKIP_CPU" != "1"; then
+    cd "${srcdir}/${_pkgname}-cpu"
+    _build openblas
+  fi
+
+  if [[ $_ENABLE_CUDA = 1 ]]; then
+    cd "${srcdir}/${_pkgname}-cuda"
+    export CUDA_HOME="${CUDA_HOME:-/opt/cuda}"
+    export PATH="$CUDA_HOME/bin:$PATH"
+    MAGMA_HOME="$CUDA_HOME/targets/x86_64-linux" CUDA_LIBPATH="$CUDA_HOME/lib64/" \
+      _build cublas
+  fi
+
+  if [[ $_ENABLE_ROCM = 1 ]]; then
+    cd "${srcdir}/${_pkgname}-rocm"
+    export ROCM_HOME="${ROCM_HOME:-/opt/rocm}"
+    export PATH="$ROC_HOME/bin:$PATH"
+    if test -n "$GPU_TARGETS"; then
+      _AMDGPU_TARGETS="$GPU_TARGETS"
+    else
+      _AMDGPU_TARGETS="${AMDGPU_TARGETS:-gfx900;gfx906;gfx908;gfx90a;gfx1030;gfx1100;gfx1101;gfx1102}"
+    fi
+    MAGMA_HOME="$ROCM_HOME" AMDGPU_TARGETS="$_AMDGPU_TARGETS" GPU_TARGETS="$_AMDGPU_TARGETS" \
+      _build hipblas
+  fi
+}
+
+_package_install() {
+  install -Dm755 "local-ai" "${pkgdir}/usr/bin/local-ai"
+  # sources/go-piper/piper/build/pi/lib/* /usr/lib/
+
+  # add 1-2 7b high performing models yaml configs based on mistral as gpt-3.5
+  # prefer chatml, add example working preload-models.yaml,
+
+  install -Dm644 README.md -t "${pkgdir}/usr/share/doc/${_pkgname}"
+}
+
+package_localai-git() {
+  if test "$_SKIP_CPU" != "1"; then
+    cd "${srcdir}/${_pkgname}-cpu"
+    _package_install
+  fi
+}
+
+package_localai-git-cuda() {
+  cd "${srcdir}/${_pkgname}-cuda"
+  pkgdesc+=' (with CUDA support)'
+  depends+=('cuda')
+  _package_install
+}
+
+package_localai-git-rocm() {
+  cd "${srcdir}/${_pkgname}-rocm"
+  pkgdesc+=' (with ROCM support)'
+  depends+=('rocm-hip-runtime')
+  _package_install
+}

x86_64/localai-git/whisper-1.5.1.patch

@@ -0,0 +1,66 @@
+From 90f96c5ea447ff7a9e796e772a4e53cec3d38235 Mon Sep 17 00:00:00 2001
+From: Felix Erkinger <wuxxin@gmail.com>
+Date: Sat, 18 Nov 2023 00:58:03 +0100
+Subject: [PATCH] update whisper_cpp to 1.5.1 with OPENBLAS, METAL, HIPBLAS,
+ CUBLAS, CLBLAST support
+
+---
+ Makefile | 8 +++++++-
+ 1 file changed, 7 insertions(+), 1 deletion(-)
+
+diff --git a/Makefile b/Makefile
+index 22689eb31..f7c7ca80e 100644
+--- a/Makefile
++++ b/Makefile
+@@ -22,7 +22,7 @@ RWKV_REPO?=https://github.com/donomii/go-rwkv.cpp
+ RWKV_VERSION?=c898cd0f62df8f2a7830e53d1d513bef4f6f792b
+
+ # whisper.cpp version
+-WHISPER_CPP_VERSION?=85ed71aaec8e0612a84c0b67804bde75aa75a273
++WHISPER_CPP_VERSION?=9d6ebd877ce7d99053423d186e6f5387a4a4753c
+
+ # bert.cpp version
+ BERT_VERSION?=6abe312cded14042f6b7c3cd8edf082713334a4d
+@@ -78,11 +78,13 @@ endif
+
+ ifeq ($(BUILD_TYPE),openblas)
+ 	CGO_LDFLAGS+=-lopenblas
++	export WHISPER_OPENBLAS=1
+ endif
+
+ ifeq ($(BUILD_TYPE),cublas)
+ 	CGO_LDFLAGS+=-lcublas -lcudart -L$(CUDA_LIBPATH)
+ 	export LLAMA_CUBLAS=1
++	export WHISPER_CUBLAS=1
+ endif
+
+ ifeq ($(BUILD_TYPE),hipblas)
+@@ -91,6 +93,7 @@ ifeq ($(BUILD_TYPE),hipblas)
+ 	export CC=$(ROCM_HOME)/llvm/bin/clang
+ 	# llama-ggml has no hipblas support, so override it here.
+ 	export STABLE_BUILD_TYPE=
++	export WHISPER_HIPBLAS=1
+ 	GPU_TARGETS ?= gfx900,gfx90a,gfx1030,gfx1031,gfx1100
+ 	AMDGPU_TARGETS ?= "$(GPU_TARGETS)"
+ 	CMAKE_ARGS+=-DLLAMA_HIPBLAS=ON -DAMDGPU_TARGETS="$(AMDGPU_TARGETS)" -DGPU_TARGETS="$(GPU_TARGETS)"
+@@ -100,10 +103,12 @@ endif
+ ifeq ($(BUILD_TYPE),metal)
+ 	CGO_LDFLAGS+=-framework Foundation -framework Metal -framework MetalKit -framework MetalPerformanceShaders
+ 	export LLAMA_METAL=1
++	export WHISPER_METAL=1
+ endif
+
+ ifeq ($(BUILD_TYPE),clblas)
+ 	CGO_LDFLAGS+=-lOpenCL -lclblast
++	export WHISPER_CLBLAST=1
+ endif
+
+ ifeq ($(OS),Darwin)
+@@ -233,6 +238,7 @@ replace:
+ 	$(GOCMD) mod edit -replace github.com/go-skynet/go-ggml-transformers.cpp=$(shell pwd)/sources/go-ggml-transformers
+ 	$(GOCMD) mod edit -replace github.com/donomii/go-rwkv.cpp=$(shell pwd)/sources/go-rwkv
+ 	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp=$(shell pwd)/sources/whisper.cpp
++	$(GOCMD) mod edit -replace github.com/ggerganov/whisper.cpp/bindings/go=$(shell pwd)/sources/whisper.cpp/bindings/go
+ 	$(GOCMD) mod edit -replace github.com/go-skynet/go-bert.cpp=$(shell pwd)/sources/go-bert
+ 	$(GOCMD) mod edit -replace github.com/mudler/go-stable-diffusion=$(shell pwd)/sources/go-stable-diffusion
+ 	$(GOCMD) mod edit -replace github.com/mudler/go-piper=$(shell pwd)/sources/go-piper