4 cores

Author: Wovchena

.github/workflows/causal_lm_cpp.yml

@@ -684,23 +684,23 @@ jobs:
   visual_language_sample:
     strategy:
       fail-fast: false
-      matrix: {runs-on: [ubuntu-20.04-16-core, macos-12]}
+      matrix: {runs-on: [ubuntu-20.04-4-cores, macos-12]}
     runs-on: ${{ matrix.runs-on }}
     steps:
       - uses: actions/checkout@v4
         with: {submodules: recursive}
       - uses: actions/setup-python@v4
         with: {python-version: 3.12}
       - run: mkdir ./ov/
-      - if: ${{ 'ubuntu-20.04-16-core' == matrix.runs-on }}
+      - if: ${{ 'ubuntu-20.04-4-cores' == matrix.runs-on }}
         run: >
           curl ${{ env.l_ov_link }} | tar --directory ./ov/ --strip-components 1 -xz
           && sudo ./ov/install_dependencies/install_openvino_dependencies.sh
       - if: ${{ 'macos-12' == matrix.runs-on }}
         run: >
           curl ${{ env.m_ov_link }} | tar --directory ./ov/ --strip-components 1 -xz
           && brew install coreutils scons
-      - run: OpenVINO_DIR=./ov/runtime/cmake/ cmake -DCMAKE_BUILD_TYPE=Release -B ./build/ ./
+      - run: cmake -DOpenVINO_DIR=./ov/runtime/cmake/ -DCMAKE_BUILD_TYPE=Release -B ./build/ ./
       - run: cmake --build ./build/ --config Release --target visual_language_chat -j
       - name: Download and convert a model and an image
         run: |
@@ -710,8 +710,9 @@ jobs:
           python ./samples/cpp/visual_language_chat/export_MiniCPM-V-2_6.py ./miniCPM-V-2_6/
           wget https://github.com/openvinotoolkit/openvino_notebooks/assets/29454499/d5fbbd1a-d484-415c-88cb-9986625b7b11
       - run: >
-          timeout 2m ./build/samples/cpp/visual_language_chat/visual_language_chat ./miniCPM-V-2_6/ d5fbbd1a-d484-415c-88cb-9986625b7b11
+          ./build/samples/cpp/visual_language_chat/visual_language_chat ./miniCPM-V-2_6/ d5fbbd1a-d484-415c-88cb-9986625b7b11
           <<< $'What is on the image?\nWhat is special on the image?'
+        timeout-minutes: 2
 
   cpp-continuous-batching-ubuntu:
     runs-on: ubuntu-20.04-8-cores
@@ -855,6 +856,7 @@ jobs:
             cpp-beam_search_causal_lm-Qwen-7B-Chat, cpp-beam_search_causal_lm-Qwen1_5-7B-Chat, cpp-beam_search_causal_lm-Phi-2,
             cpp-beam_search_causal_lm-notus-7b-v1, cpp-speculative_decoding_lm-ubuntu, cpp-prompt_lookup_decoding_lm-ubuntu,
             cpp-Phi-1_5, cpp-greedy_causal_lm-redpajama-3b-chat, cpp-chat_sample-ubuntu, cpp-continuous-batching-ubuntu,
+            visual_language_sample,
             cpp-continuous-batching-windows, cpp-continuous-batching-macos]
     if: ${{ always() }}
     runs-on: ubuntu-latest

samples/CMakeLists.txt

@@ -25,7 +25,7 @@ install(DIRECTORY
             cpp/greedy_causal_lm
             cpp/multinomial_causal_lm
             # Don't install prompt_lookup_decoding_lm and speculative_decoding_lm because they don't use openvino_genai library and arent verifyed yet.
-            # Don't install continuous_batching_accuracy and continuous_batching_benchmark because they depend on json.
+            # Don't install continuous_batching_accuracy and continuous_batching_benchmark because CB isn't ready.
             cpp/visual_language_chat
             cpp/whisper_speech_recognition
             cpp/stable_diffusion

samples/cpp/visual_language_chat/CMakeLists.txt

@@ -1,9 +1,11 @@
 # Copyright (C) 2023-2024 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
-find_package(OpenVINOGenAI REQUIRED PATHS
-    "${CMAKE_BINARY_DIR}"  # Reuse the package from the build.
-    ${OpenVINO_DIR}  # GenAI may be installed alogside OpenVINO.
+find_package(OpenVINOGenAI REQUIRED
+    PATHS
+        "${CMAKE_BINARY_DIR}"  # Reuse the package from the build.
+        ${OpenVINO_DIR}  # GenAI may be installed alogside OpenVINO.
+    NO_CMAKE_FIND_ROOT_PATH
 )
 
 file(DOWNLOAD
@@ -14,11 +16,11 @@ file(DOWNLOAD
 add_executable(visual_language_chat visual_language_chat.cpp load_image.cpp)
 target_include_directories(visual_language_chat PRIVATE "${CMAKE_CURRENT_SOUCE_DIR}" "${CMAKE_BINARY_DIR}")
 target_link_libraries(visual_language_chat PRIVATE openvino::genai)
+
 set_target_properties(visual_language_chat PROPERTIES
-    COMPILE_PDB_NAME chat_sample
+    COMPILE_PDB_NAME visual_language_chat
     # Ensure out of box LC_RPATH on macOS with SIP
     INSTALL_RPATH_USE_LINK_PATH ON)
-target_compile_features(visual_language_chat PRIVATE cxx_std_11)
 
 install(TARGETS visual_language_chat
         RUNTIME DESTINATION samples_bin/

samples/cpp/visual_language_chat/export_MiniCPM-V-2_6.py

@@ -958,8 +958,8 @@ def main():
     gc.collect()
 
     convert_vision_encoder(model, model_dir)
-    ov_cpm = init_model(model_dir, "CPU")
-    print(ov_cpm.chat(Image.open(requests.get("https://github.com/openvinotoolkit/openvino_notebooks/assets/29454499/d5fbbd1a-d484-415c-88cb-9986625b7b11", stream=True).raw), [{"role": "user", "content": "What is unusual on this image?"}], ov_cpm.processor.tokenizer))
+    # ov_cpm = init_model(model_dir, "CPU")
+    # print(ov_cpm.chat(Image.open(requests.get("https://github.com/openvinotoolkit/openvino_notebooks/assets/29454499/d5fbbd1a-d484-415c-88cb-9986625b7b11", stream=True).raw), [{"role": "user", "content": "What is unusual on this image?"}], ov_cpm.processor.tokenizer))
 
 if "__main__" == __name__:
     main()

samples/cpp/visual_language_chat/visual_language_chat.cpp

@@ -10,11 +10,10 @@ bool print_subword(std::string&& subword) {
 }
 
 int main(int argc, char* argv[]) {
-    if (4 != argc) {
+    if (3 != argc) {
         throw std::runtime_error(std::string{"Usage "} + argv[0] + " <MODEL_DIR> <IMAGE_FILE>");
     }
     ov::Tensor image = utils::load_image(argv[2]);
-    ov::Tensor image2 = utils::load_image(argv[3]);
     std::string device = "CPU";  // GPU can be used as well
     ov::AnyMap enable_compile_cache;
     if ("GPU" == device) {
@@ -32,7 +31,7 @@ int main(int argc, char* argv[]) {
     }
     pipe.generate(
         prompt,
-        ov::genai::images(std::vector{image2, image}),
+        ov::genai::image(std::move(image)),
         ov::genai::streamer(print_subword)
     );
     std::cout << "\n----------\n"