revert llm changes

llama/example/main.go

@@ -6,6 +6,7 @@ import (
 	"io"
 	"log"
 	"os"
+	"runtime"
 	"strings"
 
 	"github.com/ollama/ollama/llama"
@@ -28,9 +29,11 @@ func main() {
 
 	// load the model
 	llama.BackendInit()
-	params := llama.NewModelParams()
+	params := llama.NewModelParams(999, 0, func(p float32) {
+		fmt.Printf("loading... %f\n", p)
+	})
 	model := llama.LoadModelFromFile(*mpath, params)
-	ctxParams := llama.NewContextParams()
+	ctxParams := llama.NewContextParams(2048, runtime.NumCPU(), false)
 
 	// language model context
 	lc := llama.NewContextWithModel(model, ctxParams)
@@ -65,7 +68,7 @@ func main() {
 			panic("prompt must contain exactly one <image>")
 		}
 
-		beforeTokens, err := lc.Model().Tokenize(parts[0], 2048, true, true)
+		beforeTokens, err := lc.Model().Tokenize(parts[0], true, true)
 		if err != nil {
 			panic(err)
 		}
@@ -82,7 +85,7 @@ func main() {
 
 		llama.LlavaEvalImageEmbed(lc, embedding, 512, &nPast)
 
-		afterTokens, err := lc.Model().Tokenize(parts[1], 2048, true, true)
+		afterTokens, err := lc.Model().Tokenize(parts[1], true, true)
 		if err != nil {
 			panic(err)
 		}
@@ -92,7 +95,7 @@ func main() {
 			nPast++
 		}
 	} else {
-		tokens, err := lc.Model().Tokenize(*prompt, 2048, true, true)
+		tokens, err := lc.Model().Tokenize(*prompt, true, true)
 		if err != nil {
 			panic(err)
 		}

llm/filetype.go

@@ -2,10 +2,10 @@ package llm
 
 import "fmt"
 
-type FileType uint32
+type fileType uint32
 
 const (
-	fileTypeF32 FileType = iota
+	fileTypeF32 fileType = iota
 	fileTypeF16
 	fileTypeQ4_0
 	fileTypeQ4_1
@@ -41,7 +41,7 @@ const (
 	fileTypeUnknown
 )
 
-func ParseFileType(s string) (FileType, error) {
+func ParseFileType(s string) (fileType, error) {
 	switch s {
 	case "F32":
 		return fileTypeF32, nil
@@ -108,7 +108,7 @@ func ParseFileType(s string) (FileType, error) {
 	}
 }
 
-func (t FileType) String() string {
+func (t fileType) String() string {
 	switch t {
 	case fileTypeF32:
 		return "F32"
@@ -175,6 +175,6 @@ func (t FileType) String() string {
 	}
 }
 
-func (t FileType) Value() uint32 {
+func (t fileType) Value() uint32 {
 	return uint32(t)
 }

llm/generate/gen_darwin.sh

@@ -100,4 +100,4 @@ esac
 
 cleanup
 wait_for_compress
-echo "go generate completed.  LLM runners: $(cd ${BUILD_DIR}/..; echo *)"
+echo "go generate completed.  LLM runners: $(cd ${BUILD_DIR}/..; echo *)"

llm/generate/gen_linux.sh

@@ -58,6 +58,19 @@ init_vars
 git_module_setup
 apply_patches
 
+init_vars
+if [ -z "${OLLAMA_SKIP_STATIC_GENERATE}" -o "${OLLAMA_CPU_TARGET}" = "static" ]; then
+    # Builds by default, allows skipping, forces build if OLLAMA_CPU_TARGET="static"
+    # Enables optimized Dockerfile builds using a blanket skip and targeted overrides
+    # Static build for linking into the Go binary
+    init_vars
+    CMAKE_TARGETS="--target llama --target ggml"
+    CMAKE_DEFS="-DBUILD_SHARED_LIBS=off -DLLAMA_NATIVE=off -DLLAMA_AVX=off -DLLAMA_AVX2=off -DLLAMA_AVX512=off -DLLAMA_FMA=off -DLLAMA_F16C=off ${CMAKE_DEFS}"
+    BUILD_DIR="../build/linux/${ARCH}_static"
+    echo "Building static library"
+    build
+fi
+
 init_vars
 if [ -z "${OLLAMA_SKIP_CPU_GENERATE}" ]; then
     # Users building from source can tune the exact flags we pass to cmake for configuring

llm/generate/gen_windows.ps1

@@ -177,6 +177,39 @@ function cleanup {
 # -DGGML_AVX2 -- 2013 Intel Haswell & 2015 AMD Excavator / 2017 AMD Zen
 # -DGGML_FMA (FMA3) -- 2013 Intel Haswell & 2012 AMD Piledriver
 
+
+function build_static() {
+    if ((-not "${env:OLLAMA_SKIP_STATIC_GENERATE}") -and ((-not "${env:OLLAMA_CPU_TARGET}") -or ("${env:OLLAMA_CPU_TARGET}" -eq "static"))) {
+        # GCC build for direct linking into the Go binary
+        init_vars
+        # cmake will silently fallback to msvc compilers if mingw isn't in the path, so detect and fail fast
+        # as we need this to be compiled by gcc for golang to be able to link with itx
+        write-host "Checking for MinGW..."
+        # error action ensures we exit on failure
+        get-command gcc
+        get-command mingw32-make
+        $oldTargets = $script:cmakeTargets
+        $script:cmakeTargets = @("llama", "ggml")
+        $script:cmakeDefs = @(
+            "-G", "MinGW Makefiles"
+            "-DCMAKE_C_COMPILER=gcc.exe",
+            "-DCMAKE_CXX_COMPILER=g++.exe",
+            "-DBUILD_SHARED_LIBS=off",
+            "-DLLAMA_NATIVE=off",
+            "-DLLAMA_AVX=off",
+            "-DLLAMA_AVX2=off",
+            "-DLLAMA_AVX512=off",
+            "-DLLAMA_F16C=off",
+            "-DLLAMA_FMA=off")
+        $script:buildDir="../build/windows/${script:ARCH}_static"
+        write-host "Building static library"
+        build
+        $script:cmakeTargets = $oldTargets
+    } else {
+        write-host "Skipping CPU generation step as requested"
+    }
+}
+
 function build_cpu($gen_arch) {
     if ((-not "${env:OLLAMA_SKIP_CPU_GENERATE}" ) -and ((-not "${env:OLLAMA_CPU_TARGET}") -or ("${env:OLLAMA_CPU_TARGET}" -eq "cpu"))) {
         # remaining llama.cpp builds use MSVC 
@@ -364,6 +397,7 @@ init_vars
 if ($($args.count) -eq 0) {
     git_module_setup
     apply_patches
+    build_static
     if ($script:ARCH -eq "arm64") {
         build_cpu("ARM64")
     } else { # amd64

llm/ggml.go

@@ -55,9 +55,9 @@ func (kv KV) ParameterCount() uint64 {
 	return kv.u64("general.parameter_count")
 }
 
-func (kv KV) FileType() FileType {
+func (kv KV) FileType() fileType {
 	if u64 := kv.u64("general.file_type"); u64 > 0 {
-		return FileType(uint32(u64))
+		return fileType(uint32(u64))
 	}
 
 	return fileTypeUnknown

llm/llm.go

@@ -0,0 +1,39 @@
+package llm
+
+// #cgo CFLAGS: -Illama.cpp
+// #cgo darwin,arm64 LDFLAGS: ${SRCDIR}/build/darwin/arm64_static/libllama.a -lstdc++
+// #cgo darwin,amd64 LDFLAGS: ${SRCDIR}/build/darwin/x86_64_static/libllama.a -lstdc++
+// #cgo windows,amd64 LDFLAGS: ${SRCDIR}/build/windows/amd64_static/libllama.a -static -lstdc++
+// #cgo windows,arm64 LDFLAGS: ${SRCDIR}/build/windows/arm64_static/libllama.a -static -lstdc++
+// #cgo linux,amd64 LDFLAGS: ${SRCDIR}/build/linux/x86_64_static/libllama.a -lstdc++
+// #cgo linux,arm64 LDFLAGS: ${SRCDIR}/build/linux/arm64_static/libllama.a -lstdc++
+// #include <stdlib.h>
+// #include "llama.h"
+import "C"
+import (
+	"fmt"
+	"unsafe"
+)
+
+// SystemInfo is an unused example of calling llama.cpp functions using CGo
+func SystemInfo() string {
+	return C.GoString(C.llama_print_system_info())
+}
+
+func Quantize(infile, outfile string, ftype fileType) error {
+	cinfile := C.CString(infile)
+	defer C.free(unsafe.Pointer(cinfile))
+
+	coutfile := C.CString(outfile)
+	defer C.free(unsafe.Pointer(coutfile))
+
+	params := C.llama_model_quantize_default_params()
+	params.nthread = -1
+	params.ftype = ftype.Value()
+
+	if rc := C.llama_model_quantize(cinfile, coutfile, &params); rc != 0 {
+		return fmt.Errorf("llama_model_quantize: %d", rc)
+	}
+
+	return nil
+}

server/images.go

@@ -26,7 +26,6 @@ import (
 	"github.com/ollama/ollama/auth"
 	"github.com/ollama/ollama/envconfig"
 	"github.com/ollama/ollama/format"
-	"github.com/ollama/ollama/llama"
 	"github.com/ollama/ollama/llm"
 	"github.com/ollama/ollama/parser"
 	"github.com/ollama/ollama/template"
@@ -454,7 +453,7 @@ func CreateModel(ctx context.Context, name model.Name, modelFileDir, quantizatio
 						defer temp.Close()
 						defer os.Remove(temp.Name())
 
-						if err := llama.Quantize(blob, temp.Name(), want); err != nil {
+						if err := llm.Quantize(blob, temp.Name(), want); err != nil {
 							return err
 						}