Merge remote-tracking branch 'ggerganov/master'

* ggerganov/master: (32 commits)
  whisper.objc : fix build and CI
  talk-llama : sync llama.cpp
  sync : ggml
  GGUF: C++ refactor, backend support, misc fixes (skip) (llama/11030)
  ggml : add opencl backend (skip) (llama/10693)
  cuda : CUDA Graph Compute Function Refactor (precursor for performance improvements) (llama/11042)
  ggml : do not define GGML_USE_CUDA when building with GGML_BACKEND_DL (llama/11211)
  Vulkan: Fix float16 use on devices without float16 support + fix subgroup_size_control validation error (llama/11161)
  llama: add support for QRWKV6 model architecture (llama/11001)
  SYCL: Refactor ggml_sycl_compute_forward (llama/11121)
  fix: add missing msg in static_assert (llama/11143)
  llamafile : ppc64le MMA INT8 implementation (llama/10912)
  Disable GL_KHR_cooperative_matrix Vulkan extension if not available. (llama/11117)
  fix: Vulkan shader gen binary path when Cross-compiling (llama/11096)
  GGUF: C++ refactor, backend support, misc fixes (llama/11030)
  ggml-backend : only offload from host buffers (fix) (llama/11124)
  ggml-backend : only offload from host buffers (llama/11120)
  rpc : code cleanup (llama/11107)
  SYCL: Use get_multi_ptr instead of deprecated get_pointer in wkv6 (llama/11087)
  CUDA: add BF16 support (llama/11093)
  ...

Author: bygreencn

.github/workflows/build.yml

@@ -677,8 +677,8 @@ jobs:
         run: |
           xcodebuild -scheme whisper-Package -destination 'generic/platform=iOS'
 
-#- name: Build objc example
-#  run: xcodebuild -project examples/whisper.objc/whisper.objc.xcodeproj -scheme whisper.objc -configuration ${{ matrix.build }} -sdk iphoneos build
+      - name: Build objc example
+        run: xcodebuild -project examples/whisper.objc/whisper.objc.xcodeproj -scheme whisper.objc -configuration ${{ matrix.build }} -sdk iphoneos CODE_SIGN_IDENTITY="" CODE_SIGNING_REQUIRED=NO build
 
       - name: Build swiftui example
         run: xcodebuild -project examples/whisper.swiftui/whisper.swiftui.xcodeproj -scheme WhisperCppDemo -configuration ${{ matrix.build }} -sdk iphoneos CODE_SIGNING_REQUIRED=NO CODE_SIGN_IDENTITY= -destination 'generic/platform=iOS' build

README.md

@@ -360,6 +360,38 @@ Run the inference examples as usual, for example:
 - If you have trouble with Ascend NPU device, please create a issue with **[CANN]** prefix/tag.
 - If you run successfully with your Ascend NPU device, please help update the table `Verified devices`.
 
+## Docker
+
+### Prerequisites
+
+- Docker must be installed and running on your system.
+- Create a folder to store big models & intermediate files (ex. /whisper/models)
+
+### Images
+
+We have two Docker images available for this project:
+
+1. `ghcr.io/ggerganov/whisper.cpp:main`: This image includes the main executable file as well as `curl` and `ffmpeg`. (platforms: `linux/amd64`, `linux/arm64`)
+2. `ghcr.io/ggerganov/whisper.cpp:main-cuda`: Same as `main` but compiled with CUDA support. (platforms: `linux/amd64`)
+
+### Usage
+
+```shell
+# download model and persist it in a local folder
+docker run -it --rm \
+  -v path/to/models:/models \
+  whisper.cpp:main "./models/download-ggml-model.sh base /models"
+# transcribe an audio file
+docker run -it --rm \
+  -v path/to/models:/models \
+  -v path/to/audios:/audios \
+  whisper.cpp:main "./main -m /models/ggml-base.bin -f /audios/jfk.wav"
+# transcribe an audio file in samples folder
+docker run -it --rm \
+  -v path/to/models:/models \
+  whisper.cpp:main "./main -m /models/ggml-base.bin -f ./samples/jfk.wav"
+```
+
 ## Installing with Conan
 
 You can install pre-built binaries for whisper.cpp or build it from source using [Conan](https://conan.io/). Use the following command:

examples/server/server.cpp

@@ -223,6 +223,24 @@ void check_ffmpeg_availibility() {
     }
 }
 
+std::string generate_temp_filename(const std::string &prefix, const std::string &extension) {
+    auto now = std::chrono::system_clock::now();
+    auto now_time_t = std::chrono::system_clock::to_time_t(now);
+
+    static std::mt19937 rng{std::random_device{}()};
+    std::uniform_int_distribution<long long> dist(0, 1e9);
+
+    std::stringstream ss;
+    ss << prefix
+       << "-"
+       << std::put_time(std::localtime(&now_time_t), "%Y%m%d-%H%M%S")
+       << "-"
+       << dist(rng)
+       << extension;
+
+    return ss.str();
+}
+
 bool convert_to_wav(const std::string & temp_filename, std::string & error_resp) {
     std::ostringstream cmd_stream;
     std::string converted_filename_temp = temp_filename + "_temp.wav";
@@ -692,9 +710,7 @@ int main(int argc, char ** argv) {
         if (sparams.ffmpeg_converter) {
             // if file is not wav, convert to wav
             // write to temporary file
-            //const std::string temp_filename_base = std::tmpnam(nullptr);
-            const std::string temp_filename_base = "whisper-server-tmp"; // TODO: this is a hack, remove when the mutext is removed
-            const std::string temp_filename = temp_filename_base + ".wav";
+            const std::string temp_filename = generate_temp_filename("whisper-server", ".wav");
             std::ofstream temp_file{temp_filename, std::ios::binary};
             temp_file << audio_file.content;
             temp_file.close();

examples/talk-llama/CMakeLists.txt

@@ -1,10 +1,26 @@
 if (WHISPER_SDL2)
+    set(CMAKE_CXX_STANDARD 17)
+    set(CMAKE_CXX_STANDARD_REQUIRED ON)
+
     set(TARGET whisper-talk-llama)
     add_executable(${TARGET} talk-llama.cpp
         llama.cpp
-        llama-vocab.cpp
+        llama-adapter.cpp
+        llama-arch.cpp
+        llama-batch.cpp
+        llama-chat.cpp
+        llama-context.cpp
+        llama-cparams.cpp
         llama-grammar.cpp
+        llama-hparams.cpp
+        llama-impl.cpp
+        llama-kv-cache.cpp
+        llama-mmap.cpp
+        llama-model-loader.cpp
+        llama-model.cpp
+        llama-quant.cpp
         llama-sampling.cpp
+        llama-vocab.cpp
         unicode.cpp
         unicode-data.cpp)
     target_include_directories(${TARGET} PRIVATE ${SDL2_INCLUDE_DIRS})