diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index bc295d52d..22be233e6 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -498,6 +498,17 @@ jobs:
           path: |
             cudart-llama-bin-win-cu${{ matrix.cuda }}-x64.zip
 
+  ios-xcode-build:
+    runs-on: macos-latest
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Build Xcode project
+        run: xcodebuild -project examples/llama.swiftui/llama.swiftui.xcodeproj -scheme llama.swiftui -sdk iphoneos CODE_SIGNING_REQUIRED=NO CODE_SIGN_IDENTITY= -destination 'generic/platform=iOS' build
+
+
 #  freeBSD-latest:
 #    runs-on: macos-12
 #    steps:
diff --git a/CMakeLists.txt b/CMakeLists.txt
index f32df5fe5..3e0009415 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -43,6 +43,7 @@ else()
 endif()
 
 # general
+option(BUILD_SHARED_LIBS                "build shared libraries"                                OFF)
 option(LLAMA_STATIC                     "llama: static link libraries"                          OFF)
 option(LLAMA_NATIVE                     "llama: enable -march=native flag"                      ON)
 option(LLAMA_LTO                        "llama: enable link time optimization"                  OFF)
@@ -100,6 +101,9 @@ option(LLAMA_BUILD_TESTS                "llama: build tests"    ${LLAMA_STANDALO
 option(LLAMA_BUILD_EXAMPLES             "llama: build examples" ${LLAMA_STANDALONE})
 option(LLAMA_BUILD_SERVER               "llama: build server example"                           ON)
 
+# Required for relocatable CMake package
+include(${CMAKE_CURRENT_SOURCE_DIR}/scripts/build-info.cmake)
+
 #
 # Compile flags
 #
diff --git a/README.md b/README.md
index 2892132c4..0118eb03e 100644
--- a/README.md
+++ b/README.md
@@ -1,7 +1,53 @@
+
+
+# This fork is ONLY a modified server UI.
+
+I will maintain it and keep it up-to-date with the origin repo, so almost everything here will be like a mirror, except the server ui (which is in `./examples/server/public`)
+
+![Server UI-1](https://github.com/mounta11n/plusplus-caMalL/blob/master/media/ppcamall-ui-min_p.jpeg?raw=true)
+---
+
+- UI with CSS to make it look nicer and tidier overall.
+- CSS outsourced as a separate file
+- Added a dropdown menu with prompt style templates
+- Added a dropdown menu with system prompts
+- Prompt Styles and System Prompts are separate files, so editing is very easy.
+- Created a script that uses "dialog" to compose the command for the server.
+- Script offers the possibility to save and load configs
+
+
+
+In planning or already started:
+
+- WIP Multilingual: You will be able to select the language from a dropdown menu. So far language files only for English and German. (concerns UI elements and system prompts).
+- Dark Mode
+- Templates for the values of the UI options (samplers etc.), e.g. deterministic template, creative template, balanced template etc...
+
+
+---
+
+
+![Server UI-2](https://github.com/mounta11n/plusplus-caMalL/blob/master/media/ppCamall-ui02.jpeg?raw=true)
+
+
+---
+
+
+![Server TUI-1](https://github.com/mounta11n/plusplus-caMalL/blob/master/media/ppCamall-tui01.jpeg?raw=true)
+
+
+---
+
+
 # llama.cpp
 
+
 ![llama](https://user-images.githubusercontent.com/1991296/230134379-7181e485-c521-4d23-a0d6-f7b3b61ba524.png)
 
+
+
+
+
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 
 [Roadmap](https://github.com/users/ggerganov/projects/7) / [Project status](https://github.com/ggerganov/llama.cpp/discussions/3471) / [Manifesto](https://github.com/ggerganov/llama.cpp/discussions/205) / [ggml](https://github.com/ggerganov/ggml)
@@ -116,6 +162,7 @@ as the main playground for developing new features for the [ggml](https://github
 - [nat/openplayground](https://github.com/nat/openplayground)
 - [oobabooga/text-generation-webui](https://github.com/oobabooga/text-generation-webui)
 - [withcatai/catai](https://github.com/withcatai/catai)
+- [semperai/amica](https://github.com/semperai/amica)
 
 ---
 
diff --git a/common/CMakeLists.txt b/common/CMakeLists.txt
index 4f930bdc5..71891edc3 100644
--- a/common/CMakeLists.txt
+++ b/common/CMakeLists.txt
@@ -26,7 +26,7 @@ add_custom_command(
     COMMENT "Generating build details from Git"
     COMMAND ${CMAKE_COMMAND} -DMSVC=${MSVC} -DCMAKE_C_COMPILER_VERSION=${CMAKE_C_COMPILER_VERSION}
             -DCMAKE_C_COMPILER_ID=${CMAKE_C_COMPILER_ID} -DCMAKE_VS_PLATFORM_NAME=${CMAKE_VS_PLATFORM_NAME}
-            -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -P "${CMAKE_CURRENT_SOURCE_DIR}/../scripts/build-info.cmake"
+            -DCMAKE_C_COMPILER=${CMAKE_C_COMPILER} -P "${CMAKE_CURRENT_SOURCE_DIR}/../scripts/gen-build-info-cpp.cmake"
     WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}/.."
     DEPENDS "${CMAKE_CURRENT_SOURCE_DIR}/build-info.cpp.in" ${GIT_INDEX}
     VERBATIM
diff --git a/examples/llama.swiftui/.gitignore b/examples/llama.swiftui/.gitignore
new file mode 100644
index 000000000..9bce6af39
--- /dev/null
+++ b/examples/llama.swiftui/.gitignore
@@ -0,0 +1 @@
+xcuserdata
diff --git a/examples/llama.swiftui/README.md b/examples/llama.swiftui/README.md
new file mode 100644
index 000000000..fa68e6ed8
--- /dev/null
+++ b/examples/llama.swiftui/README.md
@@ -0,0 +1,7 @@
+# llama.swiftui
+
+Local inference of llama.cpp on an iPhone.
+So far I only tested with starcoder 1B model, but it can most likely handle 7B models as well.
+
+https://github.com/bachittle/llama.cpp/assets/39804642/e290827a-4edb-4093-9642-2a5e399ec545
+
diff --git a/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift b/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift
new file mode 100644
index 000000000..aaef09611
--- /dev/null
+++ b/examples/llama.swiftui/llama.cpp.swift/LibLlama.swift
@@ -0,0 +1,176 @@
+import Foundation
+
+// import llama
+
+enum LlamaError: Error {
+    case couldNotInitializeContext
+}
+
+actor LlamaContext {
+    private var model: OpaquePointer
+    private var context: OpaquePointer
+    private var batch: llama_batch
+    private var tokens_list: [llama_token]
+
+    var n_len: Int32 = 512
+    var n_cur: Int32 = 0
+    var n_decode: Int32 = 0
+
+    init(model: OpaquePointer, context: OpaquePointer) {
+        self.model = model
+        self.context = context
+        self.tokens_list = []
+        self.batch = llama_batch_init(512, 0, 1)
+    }
+
+    deinit {
+        llama_free(context)
+        llama_free_model(model)
+        llama_backend_free()
+    }
+
+    static func createContext(path: String) throws -> LlamaContext {
+        llama_backend_init(false)
+        let model_params = llama_model_default_params()
+
+        let model = llama_load_model_from_file(path, model_params)
+        guard let model else {
+            print("Could not load model at \(path)")
+            throw LlamaError.couldNotInitializeContext
+        }
+        var ctx_params = llama_context_default_params()
+        ctx_params.seed = 1234
+        ctx_params.n_ctx = 2048
+        ctx_params.n_threads = 8
+        ctx_params.n_threads_batch = 8
+
+        let context = llama_new_context_with_model(model, ctx_params)
+        guard let context else {
+            print("Could not load context!")
+            throw LlamaError.couldNotInitializeContext
+        }
+
+        return LlamaContext(model: model, context: context)
+    }
+
+    func get_n_tokens() -> Int32 {
+        return batch.n_tokens;
+    }
+
+    func completion_init(text: String) {
+        print("attempting to complete \"\(text)\"")
+
+        tokens_list = tokenize(text: text, add_bos: true)
+
+        let n_ctx = llama_n_ctx(context)
+        let n_kv_req = tokens_list.count + (Int(n_len) - tokens_list.count)
+
+        print("\n n_len = \(n_len), n_ctx = \(n_ctx), n_kv_req = \(n_kv_req)")
+
+        if n_kv_req > n_ctx {
+            print("error: n_kv_req > n_ctx, the required KV cache size is not big enough")
+        }
+
+        for id in tokens_list {
+            print(token_to_piece(token: id))
+        }
+
+        // batch = llama_batch_init(512, 0) // done in init()
+        batch.n_tokens = Int32(tokens_list.count)
+
+        for i1 in 0..<batch.n_tokens {
+            let i = Int(i1)
+            batch.token[i] = tokens_list[i]
+            batch.pos[i] = i1
+            batch.n_seq_id[Int(i)] = 1
+            batch.seq_id[Int(i)]![0] = 0
+            batch.logits[i] = 0
+        }
+        batch.logits[Int(batch.n_tokens) - 1] = 1 // true
+
+        if llama_decode(context, batch) != 0 {
+            print("llama_decode() failed")
+        }
+
+        n_cur = batch.n_tokens
+    }
+
+    func completion_loop() -> String {
+        var new_token_id: llama_token = 0
+
+        let n_vocab = llama_n_vocab(model)
+        let logits = llama_get_logits_ith(context, batch.n_tokens - 1)
+
+        var candidates = Array<llama_token_data>()
+        candidates.reserveCapacity(Int(n_vocab))
+
+        for token_id in 0..<n_vocab {
+            candidates.append(llama_token_data(id: token_id, logit: logits![Int(token_id)], p: 0.0))
+        }
+        candidates.withUnsafeMutableBufferPointer() { buffer in
+            var candidates_p = llama_token_data_array(data: buffer.baseAddress, size: buffer.count, sorted: false)
+
+            new_token_id = llama_sample_token_greedy(context, &candidates_p)
+        }
+
+        if new_token_id == llama_token_eos(context) || n_cur == n_len {
+            print("\n")
+            return ""
+        }
+
+        let new_token_str = token_to_piece(token: new_token_id)
+        print(new_token_str)
+        // tokens_list.append(new_token_id)
+
+        batch.n_tokens = 0
+
+        batch.token[Int(batch.n_tokens)] = new_token_id
+        batch.pos[Int(batch.n_tokens)] = n_cur
+        batch.n_seq_id[Int(batch.n_tokens)] = 1
+        batch.seq_id[Int(batch.n_tokens)]![0] = 0
+        batch.logits[Int(batch.n_tokens)] = 1 // true
+        batch.n_tokens += 1
+
+        n_decode += 1
+
+        n_cur += 1
+
+        if llama_decode(context, batch) != 0 {
+            print("failed to evaluate llama!")
+        }
+
+        return new_token_str
+    }
+
+    func clear() {
+        tokens_list.removeAll()
+    }
+
+    private func tokenize(text: String, add_bos: Bool) -> [llama_token] {
+        let n_tokens = text.count + (add_bos ? 1 : 0)
+        let tokens = UnsafeMutablePointer<llama_token>.allocate(capacity: n_tokens)
+        let tokenCount = llama_tokenize(model, text, Int32(text.count), tokens, Int32(n_tokens), add_bos, false)
+
+        var swiftTokens: [llama_token] = []
+        for i in 0..<tokenCount {
+            swiftTokens.append(tokens[Int(i)])
+        }
+
+        tokens.deallocate()
+
+        return swiftTokens
+    }
+
+    private func token_to_piece(token: llama_token) -> String {
+        let result = UnsafeMutablePointer<Int8>.allocate(capacity: 8)
+        result.initialize(repeating: Int8(0), count: 8)
+
+        let _ = llama_token_to_piece(model, token, result, 8)
+
+        let resultStr = String(cString: result)
+
+        result.deallocate()
+
+        return resultStr
+    }
+}
diff --git a/examples/llama.swiftui/llama.cpp.swift/bridging-header.h b/examples/llama.swiftui/llama.cpp.swift/bridging-header.h
new file mode 100644
index 000000000..6cd72c979
--- /dev/null
+++ b/examples/llama.swiftui/llama.cpp.swift/bridging-header.h
@@ -0,0 +1,5 @@
+//
+//  Use this file to import your target's public headers that you would like to expose to Swift.
+//
+
+#import "llama.h"
diff --git a/examples/llama.swiftui/llama.swiftui.xcodeproj/project.pbxproj b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.pbxproj
new file mode 100644
index 000000000..bc1fd15ce
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.pbxproj
@@ -0,0 +1,481 @@
+// !$*UTF8*$!
+{
+    archiveVersion = 1;
+    classes = {
+    };
+    objectVersion = 56;
+    objects = {
+
+/* Begin PBXBuildFile section */
+        542376082B0D9BFB008E6A1C /* ggml-quants.c in Sources */ = {isa = PBXBuildFile; fileRef = 542376072B0D9BFB008E6A1C /* ggml-quants.c */; };
+        5423760B2B0D9C4B008E6A1C /* ggml-backend.c in Sources */ = {isa = PBXBuildFile; fileRef = 5423760A2B0D9C4B008E6A1C /* ggml-backend.c */; };
+        542378792ACE3F3500834A7B /* ggml-metal.metal in Resources */ = {isa = PBXBuildFile; fileRef = 549479C82AC9E10B00E0F78B /* ggml-metal.metal */; };
+        542EA09D2AC8723900A8AEE9 /* ggml.c in Sources */ = {isa = PBXBuildFile; fileRef = 542EA09B2AC8723900A8AEE9 /* ggml.c */; settings = {COMPILER_FLAGS = "-DGGML_USE_ACCELERATE -DGGML_USE_METAL -DGGML_USE_K_QUANTS -O3"; }; };
+        542EA0A02AC8725700A8AEE9 /* ggml-alloc.c in Sources */ = {isa = PBXBuildFile; fileRef = 542EA09F2AC8725700A8AEE9 /* ggml-alloc.c */; };
+        542EA0A32AC8729100A8AEE9 /* llama.cpp in Sources */ = {isa = PBXBuildFile; fileRef = 542EA0A12AC8729100A8AEE9 /* llama.cpp */; settings = {COMPILER_FLAGS = "-DGGML_USE_K_QUANTS -DGGML_USE_METAL -O3"; }; };
+        549479CB2AC9E16000E0F78B /* Metal.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 549479CA2AC9E16000E0F78B /* Metal.framework */; };
+        549479CD2AC9E42A00E0F78B /* ggml-metal.m in Sources */ = {isa = PBXBuildFile; fileRef = 549479C52AC9E0F200E0F78B /* ggml-metal.m */; settings = {COMPILER_FLAGS = "-fno-objc-arc -DGGML_SWIFT -DGGML_USE_METAL -O3"; }; };
+        8A1C83772AC328BD0096AF73 /* llama_swiftuiApp.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A1C83762AC328BD0096AF73 /* llama_swiftuiApp.swift */; };
+        8A1C83792AC328BD0096AF73 /* ContentView.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A1C83782AC328BD0096AF73 /* ContentView.swift */; };
+        8A1C837B2AC328BE0096AF73 /* Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 8A1C837A2AC328BE0096AF73 /* Assets.xcassets */; };
+        8A1C837E2AC328BE0096AF73 /* Preview Assets.xcassets in Resources */ = {isa = PBXBuildFile; fileRef = 8A1C837D2AC328BE0096AF73 /* Preview Assets.xcassets */; };
+        8A39BE0A2AC7601100BFEB40 /* Accelerate.framework in Frameworks */ = {isa = PBXBuildFile; fileRef = 8A39BE092AC7601000BFEB40 /* Accelerate.framework */; };
+        8A3F84242AC4C891005E2EE8 /* models in Resources */ = {isa = PBXBuildFile; fileRef = 8A3F84232AC4C891005E2EE8 /* models */; };
+        8A907F332AC7138A006146EA /* LibLlama.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A907F322AC7134E006146EA /* LibLlama.swift */; };
+        8A9F7C4D2AC332EE008AE1EA /* LlamaState.swift in Sources */ = {isa = PBXBuildFile; fileRef = 8A9F7C4C2AC332EE008AE1EA /* LlamaState.swift */; };
+/* End PBXBuildFile section */
+
+/* Begin PBXFileReference section */
+        542376062B0D9BEA008E6A1C /* ggml-quants.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = "ggml-quants.h"; path = "../../ggml-quants.h"; sourceTree = "<group>"; };
+        542376072B0D9BFB008E6A1C /* ggml-quants.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = "ggml-quants.c"; path = "../../ggml-quants.c"; sourceTree = "<group>"; };
+        542376092B0D9C40008E6A1C /* ggml-backend.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; name = "ggml-backend.h"; path = "../../ggml-backend.h"; sourceTree = "<group>"; };
+        5423760A2B0D9C4B008E6A1C /* ggml-backend.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = "ggml-backend.c"; path = "../../ggml-backend.c"; sourceTree = "<group>"; };
+        542EA09B2AC8723900A8AEE9 /* ggml.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = ggml.c; path = ../../ggml.c; sourceTree = "<group>"; };
+        542EA09C2AC8723900A8AEE9 /* ggml.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = ggml.h; path = ../../ggml.h; sourceTree = "<group>"; };
+        542EA09E2AC8725700A8AEE9 /* ggml-alloc.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = "ggml-alloc.h"; path = "../../ggml-alloc.h"; sourceTree = "<group>"; };
+        542EA09F2AC8725700A8AEE9 /* ggml-alloc.c */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.c; name = "ggml-alloc.c"; path = "../../ggml-alloc.c"; sourceTree = "<group>"; };
+        542EA0A12AC8729100A8AEE9 /* llama.cpp */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.cpp.cpp; name = llama.cpp; path = ../../llama.cpp; sourceTree = "<group>"; };
+        542EA0A22AC8729100A8AEE9 /* llama.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = llama.h; path = ../../llama.h; sourceTree = "<group>"; };
+        549479C52AC9E0F200E0F78B /* ggml-metal.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; name = "ggml-metal.m"; path = "../../ggml-metal.m"; sourceTree = "<group>"; };
+        549479C62AC9E0F200E0F78B /* ggml-metal.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; name = "ggml-metal.h"; path = "../../ggml-metal.h"; sourceTree = "<group>"; };
+        549479C82AC9E10B00E0F78B /* ggml-metal.metal */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.metal; name = "ggml-metal.metal"; path = "../../ggml-metal.metal"; sourceTree = "<group>"; };
+        549479CA2AC9E16000E0F78B /* Metal.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Metal.framework; path = System/Library/Frameworks/Metal.framework; sourceTree = SDKROOT; };
+        8A08D20A2AC73B1500FE6CD4 /* bridging-header.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = "bridging-header.h"; sourceTree = "<group>"; };
+        8A1C83732AC328BD0096AF73 /* llama.swiftui.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = llama.swiftui.app; sourceTree = BUILT_PRODUCTS_DIR; };
+        8A1C83762AC328BD0096AF73 /* llama_swiftuiApp.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = llama_swiftuiApp.swift; sourceTree = "<group>"; };
+        8A1C83782AC328BD0096AF73 /* ContentView.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = ContentView.swift; sourceTree = "<group>"; };
+        8A1C837A2AC328BE0096AF73 /* Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = Assets.xcassets; sourceTree = "<group>"; };
+        8A1C837D2AC328BE0096AF73 /* Preview Assets.xcassets */ = {isa = PBXFileReference; lastKnownFileType = folder.assetcatalog; path = "Preview Assets.xcassets"; sourceTree = "<group>"; };
+        8A39BE092AC7601000BFEB40 /* Accelerate.framework */ = {isa = PBXFileReference; lastKnownFileType = wrapper.framework; name = Accelerate.framework; path = System/Library/Frameworks/Accelerate.framework; sourceTree = SDKROOT; };
+        8A3F841F2AC4C824005E2EE8 /* llama-2-7b-chat.Q2_K.gguf */ = {isa = PBXFileReference; lastKnownFileType = file; path = "llama-2-7b-chat.Q2_K.gguf"; sourceTree = "<group>"; };
+        8A3F84232AC4C891005E2EE8 /* models */ = {isa = PBXFileReference; lastKnownFileType = folder; name = models; path = llama.swiftui/Resources/models; sourceTree = "<group>"; };
+        8A907F322AC7134E006146EA /* LibLlama.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LibLlama.swift; sourceTree = "<group>"; };
+        8A9F7C4C2AC332EE008AE1EA /* LlamaState.swift */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.swift; path = LlamaState.swift; sourceTree = "<group>"; };
+/* End PBXFileReference section */
+
+/* Begin PBXFrameworksBuildPhase section */
+        8A1C83702AC328BD0096AF73 /* Frameworks */ = {
+            isa = PBXFrameworksBuildPhase;
+            buildActionMask = 2147483647;
+            files = (
+                549479CB2AC9E16000E0F78B /* Metal.framework in Frameworks */,
+                8A39BE0A2AC7601100BFEB40 /* Accelerate.framework in Frameworks */,
+            );
+            runOnlyForDeploymentPostprocessing = 0;
+        };
+/* End PBXFrameworksBuildPhase section */
+
+/* Begin PBXGroup section */
+        8A08D1F62AC7383900FE6CD4 /* llama.cpp */ = {
+            isa = PBXGroup;
+            children = (
+                5423760A2B0D9C4B008E6A1C /* ggml-backend.c */,
+                542376092B0D9C40008E6A1C /* ggml-backend.h */,
+                542376062B0D9BEA008E6A1C /* ggml-quants.h */,
+                542376072B0D9BFB008E6A1C /* ggml-quants.c */,
+                549479C82AC9E10B00E0F78B /* ggml-metal.metal */,
+                549479C62AC9E0F200E0F78B /* ggml-metal.h */,
+                549479C52AC9E0F200E0F78B /* ggml-metal.m */,
+                542EA09B2AC8723900A8AEE9 /* ggml.c */,
+                542EA09C2AC8723900A8AEE9 /* ggml.h */,
+                542EA09F2AC8725700A8AEE9 /* ggml-alloc.c */,
+                542EA09E2AC8725700A8AEE9 /* ggml-alloc.h */,
+                542EA0A12AC8729100A8AEE9 /* llama.cpp */,
+                542EA0A22AC8729100A8AEE9 /* llama.h */,
+            );
+            name = llama.cpp;
+            sourceTree = "<group>";
+        };
+        8A1C836A2AC328BD0096AF73 = {
+            isa = PBXGroup;
+            children = (
+                8A08D1F62AC7383900FE6CD4 /* llama.cpp */,
+                8A907F312AC7134E006146EA /* llama.cpp.swift */,
+                8A3F84232AC4C891005E2EE8 /* models */,
+                8A1C83752AC328BD0096AF73 /* llama.swiftui */,
+                8A1C83742AC328BD0096AF73 /* Products */,
+                8A39BE082AC7601000BFEB40 /* Frameworks */,
+            );
+            sourceTree = "<group>";
+        };
+        8A1C83742AC328BD0096AF73 /* Products */ = {
+            isa = PBXGroup;
+            children = (
+                8A1C83732AC328BD0096AF73 /* llama.swiftui.app */,
+            );
+            name = Products;
+            sourceTree = "<group>";
+        };
+        8A1C83752AC328BD0096AF73 /* llama.swiftui */ = {
+            isa = PBXGroup;
+            children = (
+                8A3F84102AC4BD85005E2EE8 /* Resources */,
+                8A9F7C4B2AC332DC008AE1EA /* Models */,
+                8A9F7C4A2AC332BF008AE1EA /* UI */,
+                8A1C83762AC328BD0096AF73 /* llama_swiftuiApp.swift */,
+                8A1C837A2AC328BE0096AF73 /* Assets.xcassets */,
+                8A1C837C2AC328BE0096AF73 /* Preview Content */,
+            );
+            path = llama.swiftui;
+            sourceTree = "<group>";
+        };
+        8A1C837C2AC328BE0096AF73 /* Preview Content */ = {
+            isa = PBXGroup;
+            children = (
+                8A1C837D2AC328BE0096AF73 /* Preview Assets.xcassets */,
+            );
+            path = "Preview Content";
+            sourceTree = "<group>";
+        };
+        8A39BE082AC7601000BFEB40 /* Frameworks */ = {
+            isa = PBXGroup;
+            children = (
+                549479CA2AC9E16000E0F78B /* Metal.framework */,
+                8A39BE092AC7601000BFEB40 /* Accelerate.framework */,
+            );
+            name = Frameworks;
+            sourceTree = "<group>";
+        };
+        8A3F84102AC4BD85005E2EE8 /* Resources */ = {
+            isa = PBXGroup;
+            children = (
+                8A3F84112AC4BD8C005E2EE8 /* models */,
+            );
+            path = Resources;
+            sourceTree = "<group>";
+        };
+        8A3F84112AC4BD8C005E2EE8 /* models */ = {
+            isa = PBXGroup;
+            children = (
+                8A3F841F2AC4C824005E2EE8 /* llama-2-7b-chat.Q2_K.gguf */,
+            );
+            path = models;
+            sourceTree = "<group>";
+        };
+        8A907F312AC7134E006146EA /* llama.cpp.swift */ = {
+            isa = PBXGroup;
+            children = (
+                8A08D20A2AC73B1500FE6CD4 /* bridging-header.h */,
+                8A907F322AC7134E006146EA /* LibLlama.swift */,
+            );
+            path = llama.cpp.swift;
+            sourceTree = "<group>";
+        };
+        8A9F7C4A2AC332BF008AE1EA /* UI */ = {
+            isa = PBXGroup;
+            children = (
+                8A1C83782AC328BD0096AF73 /* ContentView.swift */,
+            );
+            path = UI;
+            sourceTree = "<group>";
+        };
+        8A9F7C4B2AC332DC008AE1EA /* Models */ = {
+            isa = PBXGroup;
+            children = (
+                8A9F7C4C2AC332EE008AE1EA /* LlamaState.swift */,
+            );
+            path = Models;
+            sourceTree = "<group>";
+        };
+/* End PBXGroup section */
+
+/* Begin PBXNativeTarget section */
+        8A1C83722AC328BD0096AF73 /* llama.swiftui */ = {
+            isa = PBXNativeTarget;
+            buildConfigurationList = 8A1C83812AC328BE0096AF73 /* Build configuration list for PBXNativeTarget "llama.swiftui" */;
+            buildPhases = (
+                8A1C836F2AC328BD0096AF73 /* Sources */,
+                8A1C83702AC328BD0096AF73 /* Frameworks */,
+                8A1C83712AC328BD0096AF73 /* Resources */,
+            );
+            buildRules = (
+            );
+            dependencies = (
+            );
+            name = llama.swiftui;
+            packageProductDependencies = (
+            );
+            productName = llama.swiftui;
+            productReference = 8A1C83732AC328BD0096AF73 /* llama.swiftui.app */;
+            productType = "com.apple.product-type.application";
+        };
+/* End PBXNativeTarget section */
+
+/* Begin PBXProject section */
+        8A1C836B2AC328BD0096AF73 /* Project object */ = {
+            isa = PBXProject;
+            attributes = {
+                BuildIndependentTargetsInParallel = 1;
+                LastSwiftUpdateCheck = 1500;
+                LastUpgradeCheck = 1500;
+                TargetAttributes = {
+                    8A1C83722AC328BD0096AF73 = {
+                        CreatedOnToolsVersion = 15.0;
+                        LastSwiftMigration = 1500;
+                    };
+                };
+            };
+            buildConfigurationList = 8A1C836E2AC328BD0096AF73 /* Build configuration list for PBXProject "llama.swiftui" */;
+            compatibilityVersion = "Xcode 14.0";
+            developmentRegion = en;
+            hasScannedForEncodings = 0;
+            knownRegions = (
+                en,
+                Base,
+            );
+            mainGroup = 8A1C836A2AC328BD0096AF73;
+            packageReferences = (
+            );
+            productRefGroup = 8A1C83742AC328BD0096AF73 /* Products */;
+            projectDirPath = "";
+            projectRoot = "";
+            targets = (
+                8A1C83722AC328BD0096AF73 /* llama.swiftui */,
+            );
+        };
+/* End PBXProject section */
+
+/* Begin PBXResourcesBuildPhase section */
+        8A1C83712AC328BD0096AF73 /* Resources */ = {
+            isa = PBXResourcesBuildPhase;
+            buildActionMask = 2147483647;
+            files = (
+                542378792ACE3F3500834A7B /* ggml-metal.metal in Resources */,
+                8A3F84242AC4C891005E2EE8 /* models in Resources */,
+                8A1C837E2AC328BE0096AF73 /* Preview Assets.xcassets in Resources */,
+                8A1C837B2AC328BE0096AF73 /* Assets.xcassets in Resources */,
+            );
+            runOnlyForDeploymentPostprocessing = 0;
+        };
+/* End PBXResourcesBuildPhase section */
+
+/* Begin PBXSourcesBuildPhase section */
+        8A1C836F2AC328BD0096AF73 /* Sources */ = {
+            isa = PBXSourcesBuildPhase;
+            buildActionMask = 2147483647;
+            files = (
+                542376082B0D9BFB008E6A1C /* ggml-quants.c in Sources */,
+                549479CD2AC9E42A00E0F78B /* ggml-metal.m in Sources */,
+                542EA09D2AC8723900A8AEE9 /* ggml.c in Sources */,
+                8A907F332AC7138A006146EA /* LibLlama.swift in Sources */,
+                542EA0A32AC8729100A8AEE9 /* llama.cpp in Sources */,
+                8A9F7C4D2AC332EE008AE1EA /* LlamaState.swift in Sources */,
+                8A1C83792AC328BD0096AF73 /* ContentView.swift in Sources */,
+                8A1C83772AC328BD0096AF73 /* llama_swiftuiApp.swift in Sources */,
+                542EA0A02AC8725700A8AEE9 /* ggml-alloc.c in Sources */,
+                5423760B2B0D9C4B008E6A1C /* ggml-backend.c in Sources */,
+            );
+            runOnlyForDeploymentPostprocessing = 0;
+        };
+/* End PBXSourcesBuildPhase section */
+
+/* Begin XCBuildConfiguration section */
+        8A1C837F2AC328BE0096AF73 /* Debug */ = {
+            isa = XCBuildConfiguration;
+            buildSettings = {
+                ALWAYS_SEARCH_USER_PATHS = NO;
+                ASSETCATALOG_COMPILER_GENERATE_SWIFT_ASSET_SYMBOL_EXTENSIONS = YES;
+                CLANG_ANALYZER_NONNULL = YES;
+                CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
+                CLANG_CXX_LANGUAGE_STANDARD = "gnu++20";
+                CLANG_ENABLE_MODULES = YES;
+                CLANG_ENABLE_OBJC_ARC = YES;
+                CLANG_ENABLE_OBJC_WEAK = YES;
+                CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
+                CLANG_WARN_BOOL_CONVERSION = YES;
+                CLANG_WARN_COMMA = YES;
+                CLANG_WARN_CONSTANT_CONVERSION = YES;
+                CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
+                CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+                CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
+                CLANG_WARN_EMPTY_BODY = YES;
+                CLANG_WARN_ENUM_CONVERSION = YES;
+                CLANG_WARN_INFINITE_RECURSION = YES;
+                CLANG_WARN_INT_CONVERSION = YES;
+                CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+                CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
+                CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
+                CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+                CLANG_WARN_QUOTED_INCLUDE_IN_FRAMEWORK_HEADER = YES;
+                CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+                CLANG_WARN_STRICT_PROTOTYPES = YES;
+                CLANG_WARN_SUSPICIOUS_MOVE = YES;
+                CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
+                CLANG_WARN_UNREACHABLE_CODE = YES;
+                CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+                COPY_PHASE_STRIP = NO;
+                DEBUG_INFORMATION_FORMAT = dwarf;
+                ENABLE_STRICT_OBJC_MSGSEND = YES;
+                ENABLE_TESTABILITY = YES;
+                ENABLE_USER_SCRIPT_SANDBOXING = YES;
+                GCC_C_LANGUAGE_STANDARD = gnu17;
+                GCC_DYNAMIC_NO_PIC = NO;
+                GCC_NO_COMMON_BLOCKS = YES;
+                GCC_OPTIMIZATION_LEVEL = 0;
+                GCC_PREPROCESSOR_DEFINITIONS = (
+                    "DEBUG=1",
+                    "$(inherited)",
+                );
+                GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+                GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+                GCC_WARN_UNDECLARED_SELECTOR = YES;
+                GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+                GCC_WARN_UNUSED_FUNCTION = YES;
+                GCC_WARN_UNUSED_VARIABLE = YES;
+                IPHONEOS_DEPLOYMENT_TARGET = 17.0;
+                LOCALIZATION_PREFERS_STRING_CATALOGS = YES;
+                MTL_ENABLE_DEBUG_INFO = INCLUDE_SOURCE;
+                MTL_FAST_MATH = YES;
+                ONLY_ACTIVE_ARCH = YES;
+                SDKROOT = iphoneos;
+                SWIFT_ACTIVE_COMPILATION_CONDITIONS = "DEBUG $(inherited)";
+                SWIFT_OPTIMIZATION_LEVEL = "-Onone";
+            };
+            name = Debug;
+        };
+        8A1C83802AC328BE0096AF73 /* Release */ = {
+            isa = XCBuildConfiguration;
+            buildSettings = {
+                ALWAYS_SEARCH_USER_PATHS = NO;
+                ASSETCATALOG_COMPILER_GENERATE_SWIFT_ASSET_SYMBOL_EXTENSIONS = YES;
+                CLANG_ANALYZER_NONNULL = YES;
+                CLANG_ANALYZER_NUMBER_OBJECT_CONVERSION = YES_AGGRESSIVE;
+                CLANG_CXX_LANGUAGE_STANDARD = "gnu++20";
+                CLANG_ENABLE_MODULES = YES;
+                CLANG_ENABLE_OBJC_ARC = YES;
+                CLANG_ENABLE_OBJC_WEAK = YES;
+                CLANG_WARN_BLOCK_CAPTURE_AUTORELEASING = YES;
+                CLANG_WARN_BOOL_CONVERSION = YES;
+                CLANG_WARN_COMMA = YES;
+                CLANG_WARN_CONSTANT_CONVERSION = YES;
+                CLANG_WARN_DEPRECATED_OBJC_IMPLEMENTATIONS = YES;
+                CLANG_WARN_DIRECT_OBJC_ISA_USAGE = YES_ERROR;
+                CLANG_WARN_DOCUMENTATION_COMMENTS = YES;
+                CLANG_WARN_EMPTY_BODY = YES;
+                CLANG_WARN_ENUM_CONVERSION = YES;
+                CLANG_WARN_INFINITE_RECURSION = YES;
+                CLANG_WARN_INT_CONVERSION = YES;
+                CLANG_WARN_NON_LITERAL_NULL_CONVERSION = YES;
+                CLANG_WARN_OBJC_IMPLICIT_RETAIN_SELF = YES;
+                CLANG_WARN_OBJC_LITERAL_CONVERSION = YES;
+                CLANG_WARN_OBJC_ROOT_CLASS = YES_ERROR;
+                CLANG_WARN_QUOTED_INCLUDE_IN_FRAMEWORK_HEADER = YES;
+                CLANG_WARN_RANGE_LOOP_ANALYSIS = YES;
+                CLANG_WARN_STRICT_PROTOTYPES = YES;
+                CLANG_WARN_SUSPICIOUS_MOVE = YES;
+                CLANG_WARN_UNGUARDED_AVAILABILITY = YES_AGGRESSIVE;
+                CLANG_WARN_UNREACHABLE_CODE = YES;
+                CLANG_WARN__DUPLICATE_METHOD_MATCH = YES;
+                COPY_PHASE_STRIP = NO;
+                DEBUG_INFORMATION_FORMAT = "dwarf-with-dsym";
+                ENABLE_NS_ASSERTIONS = NO;
+                ENABLE_STRICT_OBJC_MSGSEND = YES;
+                ENABLE_USER_SCRIPT_SANDBOXING = YES;
+                GCC_C_LANGUAGE_STANDARD = gnu17;
+                GCC_NO_COMMON_BLOCKS = YES;
+                GCC_WARN_64_TO_32_BIT_CONVERSION = YES;
+                GCC_WARN_ABOUT_RETURN_TYPE = YES_ERROR;
+                GCC_WARN_UNDECLARED_SELECTOR = YES;
+                GCC_WARN_UNINITIALIZED_AUTOS = YES_AGGRESSIVE;
+                GCC_WARN_UNUSED_FUNCTION = YES;
+                GCC_WARN_UNUSED_VARIABLE = YES;
+                IPHONEOS_DEPLOYMENT_TARGET = 17.0;
+                LOCALIZATION_PREFERS_STRING_CATALOGS = YES;
+                MTL_ENABLE_DEBUG_INFO = NO;
+                MTL_FAST_MATH = YES;
+                SDKROOT = iphoneos;
+                SWIFT_COMPILATION_MODE = wholemodule;
+                VALIDATE_PRODUCT = YES;
+            };
+            name = Release;
+        };
+        8A1C83822AC328BE0096AF73 /* Debug */ = {
+            isa = XCBuildConfiguration;
+            buildSettings = {
+                ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
+                ASSETCATALOG_COMPILER_GLOBAL_ACCENT_COLOR_NAME = AccentColor;
+                CLANG_ENABLE_MODULES = YES;
+                CODE_SIGN_STYLE = Automatic;
+                CURRENT_PROJECT_VERSION = 1;
+                DEVELOPMENT_ASSET_PATHS = "\"llama.swiftui/Preview Content\"";
+                DEVELOPMENT_TEAM = STLSG3FG8Q;
+                ENABLE_PREVIEWS = YES;
+                GENERATE_INFOPLIST_FILE = YES;
+                INFOPLIST_KEY_UIApplicationSceneManifest_Generation = YES;
+                INFOPLIST_KEY_UIApplicationSupportsIndirectInputEvents = YES;
+                INFOPLIST_KEY_UILaunchScreen_Generation = YES;
+                INFOPLIST_KEY_UISupportedInterfaceOrientations_iPad = "UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
+                INFOPLIST_KEY_UISupportedInterfaceOrientations_iPhone = "UIInterfaceOrientationPortrait UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
+                IPHONEOS_DEPLOYMENT_TARGET = 16.0;
+                LD_RUNPATH_SEARCH_PATHS = (
+                    "$(inherited)",
+                    "@executable_path/Frameworks",
+                );
+                MARKETING_VERSION = 1.0;
+                PRODUCT_BUNDLE_IDENTIFIER = "com.bachittle.llama-swift";
+                PRODUCT_NAME = "$(TARGET_NAME)";
+                SWIFT_EMIT_LOC_STRINGS = YES;
+                SWIFT_OBJC_BRIDGING_HEADER = "llama.cpp.swift/bridging-header.h";
+                SWIFT_OPTIMIZATION_LEVEL = "-Onone";
+                SWIFT_VERSION = 5.0;
+                TARGETED_DEVICE_FAMILY = "1,2";
+            };
+            name = Debug;
+        };
+        8A1C83832AC328BE0096AF73 /* Release */ = {
+            isa = XCBuildConfiguration;
+            buildSettings = {
+                ASSETCATALOG_COMPILER_APPICON_NAME = AppIcon;
+                ASSETCATALOG_COMPILER_GLOBAL_ACCENT_COLOR_NAME = AccentColor;
+                CLANG_ENABLE_MODULES = YES;
+                CODE_SIGN_STYLE = Automatic;
+                CURRENT_PROJECT_VERSION = 1;
+                DEVELOPMENT_ASSET_PATHS = "\"llama.swiftui/Preview Content\"";
+                DEVELOPMENT_TEAM = STLSG3FG8Q;
+                ENABLE_PREVIEWS = YES;
+                GENERATE_INFOPLIST_FILE = YES;
+                INFOPLIST_KEY_UIApplicationSceneManifest_Generation = YES;
+                INFOPLIST_KEY_UIApplicationSupportsIndirectInputEvents = YES;
+                INFOPLIST_KEY_UILaunchScreen_Generation = YES;
+                INFOPLIST_KEY_UISupportedInterfaceOrientations_iPad = "UIInterfaceOrientationPortrait UIInterfaceOrientationPortraitUpsideDown UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
+                INFOPLIST_KEY_UISupportedInterfaceOrientations_iPhone = "UIInterfaceOrientationPortrait UIInterfaceOrientationLandscapeLeft UIInterfaceOrientationLandscapeRight";
+                IPHONEOS_DEPLOYMENT_TARGET = 16.0;
+                LD_RUNPATH_SEARCH_PATHS = (
+                    "$(inherited)",
+                    "@executable_path/Frameworks",
+                );
+                MARKETING_VERSION = 1.0;
+                PRODUCT_BUNDLE_IDENTIFIER = "com.bachittle.llama-swift";
+                PRODUCT_NAME = "$(TARGET_NAME)";
+                SWIFT_EMIT_LOC_STRINGS = YES;
+                SWIFT_OBJC_BRIDGING_HEADER = "llama.cpp.swift/bridging-header.h";
+                SWIFT_VERSION = 5.0;
+                TARGETED_DEVICE_FAMILY = "1,2";
+            };
+            name = Release;
+        };
+/* End XCBuildConfiguration section */
+
+/* Begin XCConfigurationList section */
+        8A1C836E2AC328BD0096AF73 /* Build configuration list for PBXProject "llama.swiftui" */ = {
+            isa = XCConfigurationList;
+            buildConfigurations = (
+                8A1C837F2AC328BE0096AF73 /* Debug */,
+                8A1C83802AC328BE0096AF73 /* Release */,
+            );
+            defaultConfigurationIsVisible = 0;
+            defaultConfigurationName = Release;
+        };
+        8A1C83812AC328BE0096AF73 /* Build configuration list for PBXNativeTarget "llama.swiftui" */ = {
+            isa = XCConfigurationList;
+            buildConfigurations = (
+                8A1C83822AC328BE0096AF73 /* Debug */,
+                8A1C83832AC328BE0096AF73 /* Release */,
+            );
+            defaultConfigurationIsVisible = 0;
+            defaultConfigurationName = Release;
+        };
+/* End XCConfigurationList section */
+    };
+    rootObject = 8A1C836B2AC328BD0096AF73 /* Project object */;
+}
diff --git a/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/contents.xcworkspacedata b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/contents.xcworkspacedata
new file mode 100644
index 000000000..919434a62
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/contents.xcworkspacedata
@@ -0,0 +1,7 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<Workspace
+   version = "1.0">
+   <FileRef
+      location = "self:">
+   </FileRef>
+</Workspace>
diff --git a/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist
new file mode 100644
index 000000000..3d4c1e552
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui.xcodeproj/project.xcworkspace/xcshareddata/IDEWorkspaceChecks.plist
@@ -0,0 +1,8 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
+<plist version="1.0">
+<dict>
+    <key>IDEDidComputeMac32BitWarning</key>
+    <true/>
+</dict>
+</plist>
diff --git a/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AccentColor.colorset/Contents.json b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AccentColor.colorset/Contents.json
new file mode 100644
index 000000000..eb8789700
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AccentColor.colorset/Contents.json
@@ -0,0 +1,11 @@
+{
+  "colors" : [
+    {
+      "idiom" : "universal"
+    }
+  ],
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}
diff --git a/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AppIcon.appiconset/Contents.json b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AppIcon.appiconset/Contents.json
new file mode 100644
index 000000000..13613e3ee
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/AppIcon.appiconset/Contents.json
@@ -0,0 +1,13 @@
+{
+  "images" : [
+    {
+      "idiom" : "universal",
+      "platform" : "ios",
+      "size" : "1024x1024"
+    }
+  ],
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}
diff --git a/examples/llama.swiftui/llama.swiftui/Assets.xcassets/Contents.json b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/Contents.json
new file mode 100644
index 000000000..73c00596a
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/Assets.xcassets/Contents.json
@@ -0,0 +1,6 @@
+{
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}
diff --git a/examples/llama.swiftui/llama.swiftui/Models/LlamaState.swift b/examples/llama.swiftui/llama.swiftui/Models/LlamaState.swift
new file mode 100644
index 000000000..babc60cdc
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/Models/LlamaState.swift
@@ -0,0 +1,45 @@
+import Foundation
+
+@MainActor
+class LlamaState: ObservableObject {
+    @Published var messageLog = ""
+
+    private var llamaContext: LlamaContext?
+    private var modelUrl: URL? {
+        Bundle.main.url(forResource: "q8_0", withExtension: "gguf", subdirectory: "models")
+        // Bundle.main.url(forResource: "llama-2-7b-chat", withExtension: "Q2_K.gguf", subdirectory: "models")
+    }
+    init() {
+        do {
+            try loadModel()
+        } catch {
+            messageLog += "Error!\n"
+        }
+    }
+
+    private func loadModel() throws {
+        messageLog += "Loading model...\n"
+        if let modelUrl {
+            llamaContext = try LlamaContext.createContext(path: modelUrl.path())
+            messageLog += "Loaded model \(modelUrl.lastPathComponent)\n"
+        } else {
+            messageLog += "Could not locate model\n"
+        }
+    }
+
+    func complete(text: String) async {
+        guard let llamaContext else {
+            return
+        }
+        messageLog += "Attempting to complete text...\n"
+        await llamaContext.completion_init(text: text)
+        messageLog += "\(text)"
+
+        while await llamaContext.n_cur <= llamaContext.n_len {
+            let result = await llamaContext.completion_loop()
+            messageLog += "\(result)"
+        }
+        await llamaContext.clear()
+        messageLog += "\n\ndone\n"
+    }
+}
diff --git a/examples/llama.swiftui/llama.swiftui/Preview Content/Preview Assets.xcassets/Contents.json b/examples/llama.swiftui/llama.swiftui/Preview Content/Preview Assets.xcassets/Contents.json
new file mode 100644
index 000000000..73c00596a
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/Preview Content/Preview Assets.xcassets/Contents.json	
@@ -0,0 +1,6 @@
+{
+  "info" : {
+    "author" : "xcode",
+    "version" : 1
+  }
+}
diff --git a/examples/llama.swiftui/llama.swiftui/Resources/models/.gitignore b/examples/llama.swiftui/llama.swiftui/Resources/models/.gitignore
new file mode 100644
index 000000000..e69de29bb
diff --git a/examples/llama.swiftui/llama.swiftui/UI/ContentView.swift b/examples/llama.swiftui/llama.swiftui/UI/ContentView.swift
new file mode 100644
index 000000000..0bd16a806
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/UI/ContentView.swift
@@ -0,0 +1,42 @@
+import SwiftUI
+
+struct ContentView: View {
+    @StateObject var llamaState = LlamaState()
+
+    @State private var multiLineText = ""
+
+    var body: some View {
+        VStack {
+            ScrollView(.vertical) {
+                Text(llamaState.messageLog)
+            }
+
+            TextEditor(text: $multiLineText)
+                .frame(height: 200)
+                .padding()
+                .border(Color.gray, width: 0.5)
+            Button(action: {
+                sendText()
+            }) {
+                Text("Send")
+                    .padding()
+                    .background(Color.blue)
+                    .foregroundColor(.white)
+                    .cornerRadius(8)
+            }
+        }
+        .padding()
+    }
+
+    func sendText() {
+        Task {
+            await llamaState.complete(text: multiLineText)
+            multiLineText = ""
+        }
+    }
+}
+/*
+#Preview {
+    ContentView()
+}
+*/
diff --git a/examples/llama.swiftui/llama.swiftui/llama_swiftuiApp.swift b/examples/llama.swiftui/llama.swiftui/llama_swiftuiApp.swift
new file mode 100644
index 000000000..cccda8a97
--- /dev/null
+++ b/examples/llama.swiftui/llama.swiftui/llama_swiftuiApp.swift
@@ -0,0 +1,10 @@
+import SwiftUI
+
+@main
+struct llama_swiftuiApp: App {
+    var body: some Scene {
+        WindowGroup {
+            ContentView()
+        }
+    }
+}
diff --git a/ggml.c b/ggml.c
index 0c7264a36..c522a101f 100644
--- a/ggml.c
+++ b/ggml.c
@@ -9373,7 +9373,7 @@ static bool ggml_compute_forward_mul_mat_use_blas(
     // TODO: find the optimal values for these
     if (ggml_is_contiguous(src0) &&
         ggml_is_contiguous(src1) &&
-        src0->type == GGML_TYPE_F32 &&
+      //src0->type == GGML_TYPE_F32 &&
         src1->type == GGML_TYPE_F32 &&
         (ne0 >= 32 && ne1 >= 32 && ne10 >= 32)) {
 
diff --git a/icon.png b/icon.png
new file mode 100755
index 000000000..c071a5b3c
Binary files /dev/null and b/icon.png differ
diff --git a/llama.cpp b/llama.cpp
index f2b5967d7..cb544228b 100644
--- a/llama.cpp
+++ b/llama.cpp
@@ -5550,18 +5550,8 @@ static int llama_decode_internal(
         n_threads = std::min(4, n_threads);
     }
 
-    // If all tensors can be run on the GPU then using more than 1 thread is detrimental.
-    const bool full_offload_supported =
-        model.arch == LLM_ARCH_LLAMA      ||
-        model.arch == LLM_ARCH_BAICHUAN   ||
-        model.arch == LLM_ARCH_FALCON     ||
-        model.arch == LLM_ARCH_REFACT     ||
-        model.arch == LLM_ARCH_MPT        ||
-        model.arch == LLM_ARCH_STARCODER  ||
-        model.arch == LLM_ARCH_STABLELM;
-
     const bool fully_offloaded = model.n_gpu_layers >= (int) hparams.n_layer + 3;
-    if (ggml_cpu_has_cublas() && full_offload_supported && fully_offloaded) {
+    if (ggml_cpu_has_cublas() && fully_offloaded) {
         n_threads = 1;
     }
 
diff --git a/media/llama-server-icon.png b/media/llama-server-icon.png
new file mode 100755
index 000000000..c071a5b3c
Binary files /dev/null and b/media/llama-server-icon.png differ
diff --git a/media/ppcamall-ui-min_p.jpeg b/media/ppcamall-ui-min_p.jpeg
new file mode 100644
index 000000000..e21cd041d
Binary files /dev/null and b/media/ppcamall-ui-min_p.jpeg differ
diff --git a/scripts/build-info.cmake b/scripts/build-info.cmake
index 73853dfa4..ea3dc55c8 100644
--- a/scripts/build-info.cmake
+++ b/scripts/build-info.cmake
@@ -1,5 +1,3 @@
-set(TEMPLATE_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp.in")
-set(OUTPUT_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp")
 set(BUILD_NUMBER 0)
 set(BUILD_COMMIT "unknown")
 set(BUILD_COMPILER "unknown")
@@ -58,23 +56,3 @@ else()
     )
     set(BUILD_TARGET ${OUT})
 endif()
-
-# Only write the build info if it changed
-if(EXISTS ${OUTPUT_FILE})
-    file(READ ${OUTPUT_FILE} CONTENTS)
-    string(REGEX MATCH "LLAMA_COMMIT = \"([^\"]*)\";" _ ${CONTENTS})
-    set(OLD_COMMIT ${CMAKE_MATCH_1})
-    string(REGEX MATCH "LLAMA_COMPILER = \"([^\"]*)\";" _ ${CONTENTS})
-    set(OLD_COMPILER ${CMAKE_MATCH_1})
-    string(REGEX MATCH "LLAMA_BUILD_TARGET = \"([^\"]*)\";" _ ${CONTENTS})
-    set(OLD_TARGET ${CMAKE_MATCH_1})
-    if (
-        NOT OLD_COMMIT   STREQUAL BUILD_COMMIT   OR
-        NOT OLD_COMPILER STREQUAL BUILD_COMPILER OR
-        NOT OLD_TARGET   STREQUAL BUILD_TARGET
-    )
-        configure_file(${TEMPLATE_FILE} ${OUTPUT_FILE})
-    endif()
-else()
-    configure_file(${TEMPLATE_FILE} ${OUTPUT_FILE})
-endif()
diff --git a/scripts/gen-build-info-cpp.cmake b/scripts/gen-build-info-cpp.cmake
new file mode 100644
index 000000000..d89338920
--- /dev/null
+++ b/scripts/gen-build-info-cpp.cmake
@@ -0,0 +1,24 @@
+include(${CMAKE_CURRENT_SOURCE_DIR}/scripts/build-info.cmake)
+
+set(TEMPLATE_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp.in")
+set(OUTPUT_FILE "${CMAKE_CURRENT_SOURCE_DIR}/common/build-info.cpp")
+
+# Only write the build info if it changed
+if(EXISTS ${OUTPUT_FILE})
+    file(READ ${OUTPUT_FILE} CONTENTS)
+    string(REGEX MATCH "LLAMA_COMMIT = \"([^\"]*)\";" _ ${CONTENTS})
+    set(OLD_COMMIT ${CMAKE_MATCH_1})
+    string(REGEX MATCH "LLAMA_COMPILER = \"([^\"]*)\";" _ ${CONTENTS})
+    set(OLD_COMPILER ${CMAKE_MATCH_1})
+    string(REGEX MATCH "LLAMA_BUILD_TARGET = \"([^\"]*)\";" _ ${CONTENTS})
+    set(OLD_TARGET ${CMAKE_MATCH_1})
+    if (
+        NOT OLD_COMMIT   STREQUAL BUILD_COMMIT   OR
+        NOT OLD_COMPILER STREQUAL BUILD_COMPILER OR
+        NOT OLD_TARGET   STREQUAL BUILD_TARGET
+    )
+        configure_file(${TEMPLATE_FILE} ${OUTPUT_FILE})
+    endif()
+else()
+    configure_file(${TEMPLATE_FILE} ${OUTPUT_FILE})
+endif()
diff --git a/start-server-dialog.sh b/start-server-dialog.sh
new file mode 100755
index 000000000..aeb6535e8
--- /dev/null
+++ b/start-server-dialog.sh
@@ -0,0 +1,343 @@
+#!/bin/bash
+
+# Set default values
+model_path="./models/"
+mmproj_path=""
+threads=4
+ctx_size=512
+batch_size=512
+n_gpu_layers=0
+cont_batching="off"
+mlock="off"
+no_mmap="off"
+host="127.0.0.1"
+port="8080"
+advanced_options=""
+
+
+
+model_selection() {
+    # User selects a file or folder
+  exec 3>&1
+  model_path=$(dialog --backtitle "Model Selection" \
+                      --title "Select Model File or Folder" \
+                      --fselect "$HOME/" 14 60 \
+                      2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # If a folder has been selected, search for *.gguf files
+  if [ -d "$model_path" ]; then
+    model_files=($(find "$model_path" -name "*.gguf" 2>/dev/null))
+  elif [ -f "$model_path" ]; then
+    model_files=("$model_path")
+  else
+    dialog --backtitle "Model Selection" \
+           --title "Invalid Selection" \
+           --msgbox "The selected path is not valid." 7 50
+    return
+  fi
+
+# Selection menu for models found
+exec 3>&1
+model_choice=$(dialog --backtitle "Model Selection" \
+                      --title "Select a Model File" \
+                      --menu "Choose one of the found models:" 15 60 4 \
+                      $(for i in "${!model_files[@]}"; do echo "$((i+1))" "$(basename "${model_files[$i]}")"; done) \
+                      2>&1 1>&3)
+exit_status=$?
+exec 3>&-
+
+# Check whether user has selected 'Cancel'
+if [ $exit_status = 1 ]; then
+  return
+fi
+
+# Set path to the selected model
+model_path=${model_files[$((model_choice-1))]}
+}
+
+
+
+multimodal_model_selection() {
+    # User selects a file or folder
+  exec 3>&1
+  mmproj_path=$(dialog --backtitle "Multimodal Model" \
+                      --title "Select Model File or Folder" \
+                      --fselect "$HOME/" 14 60 \
+                      2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # If a folder has been selected, search for *.bin files
+  if [ -d "$mmproj_path" ]; then
+    multi_modal_files=($(find "$mmproj_path" -name "*.bin" 2>/dev/null))
+  elif [ -f "$mmproj_path" ]; then
+    multi_modal_files=("$mmproj_path")
+  else
+    dialog --backtitle "Multimodal Model" \
+           --title "Invalid Selection" \
+           --msgbox "The selected path is not valid." 7 50
+    return
+  fi
+
+# Selection menu for models found
+exec 3>&1
+multi_modal_choice=$(dialog --backtitle "Multimodal Model" \
+                      --title "Select a Model File" \
+                      --menu "Choose one of the found models:" 15 60 4 \
+                      $(for i in "${!multi_modal_files[@]}"; do echo "$((i+1))" "$(basename "${multi_modal_files[$i]}")"; done) \
+                      2>&1 1>&3)
+exit_status=$?
+exec 3>&-
+
+# Check whether user has selected 'Cancel'
+if [ $exit_status = 1 ]; then
+  return
+fi
+
+# Set path to the selected model
+mmproj_path=${multi_modal_files[$((multi_modal_choice-1))]}
+}
+
+
+
+options() {
+  # Show form for entering the options
+  exec 3>&1
+  form_values=$(dialog --backtitle "Options Configuration" \
+                       --title "Set Options" \
+                       --form "Enter the values for the following options:" \
+                       15 50 0 \
+                       "Number of Threads (-t):" 1 1 "$threads" 1 25 25 5 \
+                       "Context Size (-c):" 2 1 "$ctx_size" 2 25 25 5 \
+                       "Batch Size (-b):" 3 1 "$batch_size" 3 25 25 5 \
+                       "GPU Layers (-ngl):" 4 1 "$n_gpu_layers" 4 25 25 5 \
+                       2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Save the entered values in the corresponding variables
+  IFS=$'\n' read -r threads ctx_size batch_size n_gpu_layers <<< "$form_values"
+}
+
+
+
+further_options() {
+  # Initial values for the checkboxes based on current settings
+  cb_value=$([ "$cont_batching" = "on" ] && echo "on" || echo "off")
+  mlock_value=$([ "$mlock" = "on" ] && echo "on" || echo "off")
+  no_mmap_value=$([ "$no_mmap" = "on" ] && echo "on" || echo "off")
+
+  # Show dialog for setting options
+  exec 3>&1
+  choices=$(dialog --backtitle "Further Options" \
+                   --title "Boolean Options" \
+                   --checklist "Select options:" 15 60 3 \
+                   "1" "Continuous Batching (-cb)" $cb_value \
+                   "2" "Memory Lock (--mlock)" $mlock_value \
+                   "3" "No Memory Map (--no-mmap)" $no_mmap_value \
+                   2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Set options based on user selection
+  cont_batching="off"
+  mlock="off"
+  no_mmap="off"
+  for choice in $choices; do
+    case $choice in
+      1) cont_batching="on" ;;
+      2) mlock="on" ;;
+      3) no_mmap="on" ;;
+    esac
+  done
+}
+
+
+
+advanced_options() {
+  # Input fields for Advanced Options
+  exec 3>&1
+  advanced_values=$(dialog --backtitle "Advanced Options" \
+                           --title "Advanced Server Configuration" \
+                           --form "Enter the advanced configuration options:" \
+                           15 60 0 \
+                           "Host IP:" 1 1 "$host" 1 15 15 0 \
+                           "Port:" 2 1 "$port" 2 15 5 0 \
+                           "Additional Options:" 3 1 "$advanced_options" 3 15 30 0 \
+                           2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Read the entries and save them in the corresponding variables
+  read -r host port advanced_options <<< "$advanced_values"
+}
+
+
+
+start_server() {
+  # Compiling the command with the selected options
+  cmd="./server"
+  [ -n "$model_path" ] && cmd+=" -m $model_path"
+  [ -n "$mmproj_path" ] && cmd+=" --mmproj $mmproj_path"
+  [ "$threads" -ne 4 ] && cmd+=" -t $threads"
+  [ "$ctx_size" -ne 512 ] && cmd+=" -c $ctx_size"
+  [ "$batch_size" -ne 512 ] && cmd+=" -b $batch_size"
+  [ "$n_gpu_layers" -ne 0 ] && cmd+=" -ngl $n_gpu_layers"
+  [ "$cont_batching" = "on" ] && cmd+=" -cb"
+  [ "$mlock" = "on" ] && cmd+=" --mlock"
+  [ "$no_mmap" = "off" ] && cmd+=" --no-mmap"
+  [ -n "$host" ] && cmd+=" --host $host"
+  [ -n "$port" ] && cmd+=" --port $port"
+  [ -n "$advanced_options" ] && cmd+=" $advanced_options"
+
+    eval "$cmd"
+  read -p 'Do not forget to quit the server later with Ctrl+C as soon as you are finished. Press Enter to continue...'
+}
+
+
+
+# Function to save the current configuration
+save_config() {
+  exec 3>&1
+  config_file=$(dialog --backtitle "Save Configuration" \
+                       --title "Save Configuration File" \
+                       --fselect "$HOME/" 14 60 \
+                       2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Saving the configuration to the file
+  cat > "$config_file" << EOF
+model_path=$model_path
+mmproj_path=$mmproj_path
+threads=$threads
+ctx_size=$ctx_size
+batch_size=$batch_size
+n_gpu_layers=$n_gpu_layers
+cont_batching=$cont_batching
+mlock=$mlock
+no_mmap=$no_mmap
+host=$host
+port=$port
+advanced_options=$advanced_options
+EOF
+
+  dialog --backtitle "Save Configuration" \
+         --title "Configuration Saved" \
+         --msgbox "Configuration has been saved to $config_file" 7 50
+}
+
+
+
+# Function for loading the configuration from a file
+load_config() {
+  exec 3>&1
+  config_file=$(dialog --backtitle "Load Configuration" \
+                       --title "Load Configuration File" \
+                       --fselect "$HOME/" 14 60 \
+                       2>&1 1>&3)
+  exit_status=$?
+  exec 3>&-
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Check whether the configuration file exists
+  if [ ! -f "$config_file" ]; then
+    dialog --backtitle "Load Configuration" \
+           --title "File Not Found" \
+           --msgbox "The file $config_file was not found." 7 50
+    return
+  fi
+
+  # Load configuration from the file
+  source "$config_file"
+
+  dialog --backtitle "Load Configuration" \
+         --title "Configuration Loaded" \
+         --msgbox "Configuration has been loaded from $config_file" 7 50
+}
+
+
+
+# Function to show the main menu
+show_main_menu() {
+  while true; do
+    exec 3>&1
+    selection=$(dialog \
+      --backtitle "Server Configuration" \
+      --title "Main Menu" \
+      --clear \
+      --cancel-label "Exit" \
+      --menu "Please select:" 15 50 6 \
+      "1" "Model Selection" \
+      "2" "Multimodal Model Selection" \
+      "3" "Options" \
+      "4" "Further Options" \
+      "5" "Advanced Options" \
+      "6" "Save Config" \
+      "7" "Load Config" \
+      "8" "Start Server" \
+      2>&1 1>&3)
+    exit_status=$?
+    exec 3>&-
+
+    # Check whether user has selected 'Exit'
+    if [ $exit_status = 1 ]; then
+      clear
+      exit
+    fi
+
+    # Call up the corresponding function based on the selection
+    case $selection in
+      1) model_selection ;;
+      2) multimodal_model_selection ;;
+      3) options ;;
+      4) further_options ;;
+      5) advanced_options ;;
+      6) save_config ;;
+      7) load_config ;;
+      8) start_server ;;
+      *) clear ;;
+    esac
+  done
+}
+
+
+
+# Show main menu
+show_main_menu
diff --git a/start-server-zenity.sh b/start-server-zenity.sh
new file mode 100755
index 000000000..5415dcfda
--- /dev/null
+++ b/start-server-zenity.sh
@@ -0,0 +1,294 @@
+#!/bin/bash
+
+# Set default values
+model_path="./models/"
+mmproj_path=""
+threads=4
+ctx_size=512
+batch_size=512
+n_gpu_layers=0
+cont_batching="off"
+mlock="off"
+no_mmap="off"
+host="127.0.0.1"
+port="8080"
+advanced_options=""
+
+
+
+# Function to install Zenity
+install_zenity() {
+    echo "Try to install Zenity with $1..."
+    if ! $1 install zenity -y; then
+        echo "Error: Zenity could not be installed."
+        exit 1
+    fi
+    echo "Zenity was successfully installed."
+}
+
+# Check whether Zenity is already installed
+if ! command -v zenity &> /dev/null; then
+    # Zenity is not installed, try to find the package manager
+    PACKAGE_MANAGERS=(brew apt apt-get yum pacman)
+    for manager in "${PACKAGE_MANAGERS[@]}"; do
+        if command -v $manager &> /dev/null; then
+            # Package manager found, ask the user for permission
+            read -p "Zenity is not installed. Would you like to install Zenity with $manager? (y/N) " response
+            if [[ "$response" =~ ^[Yy]$ ]]; then
+                # User has agreed, install Zenity
+                install_zenity $manager
+                break
+            else
+                echo "Installation canceled."
+                exit 1
+            fi
+        fi
+    done
+    if ! command -v zenity &> /dev/null; then
+        echo "No supported package manager found or Zenity could not be installed. Please install Zenity manually."
+        exit 1
+    fi
+fi
+
+
+
+model_selection() {
+  # User selects a file or folder
+  model_path=$(zenity --file-selection --title="Select Model File or Folder" --filename="$HOME/" --file-filter="*.gguf" --file-filter="*" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # If a folder has been selected, search for *.gguf files
+  if [ -d "$model_path" ]; then
+    model_files=($(find "$model_path" -name "*.gguf" 2>/dev/null))
+  elif [ -f "$model_path" ]; then
+    model_files=("$model_path")
+  else
+    zenity --error --title="Invalid Selection" --text="The selected path is not valid."
+    return
+  fi
+
+  # Selection menu for models found
+  model_choice=$(zenity --list --title="Select a Model File" --column="Index" --column="Model File" $(for i in "${!model_files[@]}"; do echo "$((i+1))" "$(basename "${model_files[$i]}")"; done) --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Set path to the selected model
+  model_path=${model_files[$((model_choice-1))]}
+}
+
+
+
+multimodal_model_selection() {
+  # User selects a file or folder
+  mmproj_path=$(zenity --file-selection --title="Select Multimodal Model File or Folder" --filename="$HOME/" --file-filter="*.bin" --file-filter="*" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # If a folder has been selected, search for *.bin files
+  if [ -d "$mmproj_path" ]; then
+    multi_modal_files=($(find "$mmproj_path" -name "*.bin" 2>/dev/null))
+  elif [ -f "$mmproj_path" ]; then
+    multi_modal_files=("$mmproj_path")
+  else
+    zenity --error --title="Invalid Selection" --text="The selected path is not valid."
+    return
+  fi
+
+  # Selection menu for models found
+  multi_modal_choice=$(zenity --list --title="Select a Multimodal Model File" --column="Index" --column="Model File" $(for i in "${!multi_modal_files[@]}"; do echo "$((i+1))" "$(basename "${multi_modal_files[$i]}")"; done) --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Set path to the selected model
+  mmproj_path=${multi_modal_files[$((multi_modal_choice-1))]}
+}
+
+
+
+options() {
+  # Show form for entering the options
+  form_values=$(zenity --forms --title="Set Options" --text="Enter the values for the following options:" --add-entry="Number of Threads (-t):" --add-entry="Context Size (-c):" --add-entry="Batch Size (-b):" --add-entry="GPU Layers (-ngl):" --separator="|" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Save the entered values in the corresponding variables
+  IFS="|" read -r threads ctx_size batch_size n_gpu_layers <<< "$form_values"
+}
+
+
+
+further_options() {
+  # Initial values for the checkboxes based on current settings
+  cb_value=$([ "$cont_batching" = "on" ] && echo "TRUE" || echo "FALSE")
+  mlock_value=$([ "$mlock" = "on" ] && echo "TRUE" || echo "FALSE")
+  no_mmap_value=$([ "$no_mmap" = "on" ] && echo "TRUE" || echo "FALSE")
+
+  # Show dialog for setting options
+  choices=$(zenity --list --title="Boolean Options" --text="Select options:" --checklist --column="Select" --column="Option" TRUE "Continuous Batching (-cb)" FALSE "Memory Lock (--mlock)" FALSE "No Memory Map (--no-mmap)" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Set options based on user selection
+  cont_batching="off"
+  mlock="off"
+  no_mmap="off"
+  for choice in $choices; do
+    case $choice in
+      "Continuous Batching (-cb)") cont_batching="on" ;;
+      "Memory Lock (--mlock)") mlock="on" ;;
+      "No Memory Map (--no-mmap)") no_mmap="on" ;;
+    esac
+  done
+}
+
+
+
+advanced_options() {
+  # Input fields for Advanced Options
+  advanced_values=$(zenity --forms --title="Advanced Server Configuration" --text="Enter the advanced configuration options:" --add-entry="Host IP:" --add-entry="Port:" --add-entry="Additional Options:" --separator="|" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Read the entries and save them in the corresponding variables
+  IFS="|" read -r host port advanced_options <<< "$advanced_values"
+}
+
+
+
+start_server() {
+  # Compiling the command with the selected options
+  cmd="./server"
+  [ -n "$model_path" ] && cmd+=" -m $model_path"
+  [ -n "$mmproj_path" ] && cmd+=" --mmproj $mmproj_path"
+  [ "$threads" -ne 4 ] && cmd+=" -t $threads"
+  [ "$ctx_size" -ne 512 ] && cmd+=" -c $ctx_size"
+  [ "$batch_size" -ne 512 ] && cmd+=" -b $batch_size"
+  [ "$n_gpu_layers" -ne 0 ] && cmd+=" -ngl $n_gpu_layers"
+  [ "$cont_batching" = "on" ] && cmd+=" -cb"
+  [ "$mlock" = "on" ] && cmd+=" --mlock"
+  [ "$no_mmap" = "off" ] && cmd+=" --no-mmap"
+  [ -n "$host" ] && cmd+=" --host $host"
+  [ -n "$port" ] && cmd+=" --port $port"
+  [ -n "$advanced_options" ] && cmd+=" $advanced_options"
+
+  eval "$cmd"
+  read -p 'Press Enter to continue...'
+}
+
+
+
+# Function to save the current configuration
+save_config() {
+  config_file=$(zenity --file-selection --title="Save Configuration File" --filename="$HOME/" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Saving the configuration to the file
+  cat > "$config_file" << EOF
+model_path=$model_path
+mmproj_path=$mmproj_path
+threads=$threads
+ctx_size=$ctx_size
+batch_size=$batch_size
+n_gpu_layers=$n_gpu_layers
+cont_batching=$cont_batching
+mlock=$mlock
+no_mmap=$no_mmap
+host=$host
+port=$port
+advanced_options=$advanced_options
+EOF
+
+  zenity --info --title="Configuration Saved" --text="Configuration has been saved to $config_file" --width=300 --height=400
+}
+
+
+
+# Function for loading the configuration from a file
+load_config() {
+  config_file=$(zenity --file-selection --title="Load Configuration File" --filename="$HOME/" --width=300 --height=400)
+  exit_status=$?
+
+  # Check whether user has selected 'Cancel'
+  if [ $exit_status = 1 ]; then
+    return
+  fi
+
+  # Check whether the configuration file exists
+  if [ ! -f "$config_file" ]; then
+    zenity --error --title="File Not Found" --text="The file $config_file was not found." --width=300 --height=400
+    return
+  fi
+
+  # Load configuration from the file
+  source "$config_file"
+
+  zenity --info --title="Configuration Loaded" --text="Configuration has been loaded from $config_file" --width=300 --height=400
+}
+
+
+
+# Function to show the main menu
+show_main_menu() {
+  while true; do
+    selection=$(zenity --list --title="Main Menu" --text="Please select:" --cancel-label="Exit" --column="Index" --column="Option" 1 "Model Selection" 2 "Multimodal Model Selection" 3 "Options" 4 "Further Options" 5 "Advanced Options" 6 "Save Config" 7 "Load Config" 8 "Start Server" --width=300 --height=400)
+    exit_status=$?
+
+    # Check whether user has selected 'Exit'
+    if [ $exit_status = 1 ]; then
+      clear
+      exit
+    fi
+
+    # Call up the corresponding function based on the selection
+    case $selection in
+      1) model_selection ;;
+      2) multimodal_model_selection ;;
+      3) options ;;
+      4) further_options ;;
+      5) advanced_options ;;
+      6) save_config ;;
+      7) load_config ;;
+      8) start_server ;;
+      *) clear ;;
+    esac
+  done
+}
+
+
+
+# Show main menu
+show_main_menu