Errors fixed.

Author: popovaan

.github/workflows/causal_lm_cpp.yml

@@ -704,21 +704,19 @@ jobs:
           source ./ov/setupvars.sh
           cmake -DCMAKE_BUILD_TYPE=Release -S ./ -B ./build/
           cmake --build ./build/ --config Release --target visual_language_chat -j
-      - name: Download and convert and model
+      - name: Download and convert a model and an image
         run: |
           source ./ov/setupvars.sh
-          python -m pip install --upgrade-strategy eager -r ./samples/requirements.txt --pre --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
           python -m pip install ./thirdparty/openvino_tokenizers/[transformers] --pre --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
+          python -m pip install --upgrade-strategy eager -r ./samples/requirements.txt --pre --extra-index-url https://storage.openvinotoolkit.org/simple/wheels/nightly
           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
           
-      - name: Run chat sample
+      - name: Run chat chat sample
         run: |
           source ./ov/setupvars.sh
-          export PYTHONPATH=./build/:$PYTHONPATH
-          printf 'What is on the image?\nWhat is special on the image?\n' > ./input.txt
-          timeout 120s python ./samples/python/vlm_chat_sample/vlm_chat_sample.py ./miniCPM-V-2_6/ d5fbbd1a-d484-415c-88cb-9986625b7b11 < input.txt > ./pred.txt || ( [[ $? -eq 124 ]] && \
-          echo "Timeout reached, but it's excpected." )
+          && timeout 120s ./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?'
 
   cpp-continuous-batching-ubuntu:
     runs-on: ubuntu-20.04-8-cores

src/cpp/include/openvino/genai/vlm_pipeline.hpp

@@ -163,6 +163,7 @@ class OPENVINO_GENAI_EXPORTS VLMPipeline {
     class VLMPipelineImpl;
     std::unique_ptr<VLMPipelineImpl> m_pimpl;
 };
+
 /*
  * utils that allow to use generate() in the following way:
  * pipe.generate(prompt, ov::genai::image(std::move(image_tensor))).

src/cpp/src/vlm_pipeline.cpp

@@ -533,8 +533,25 @@ DecodedResults VLMPipeline::generate(
             variable.reset();
         }
         m_language.get_tensor("attention_mask").set_shape({1, 0});
-    }
-    return {{m_tokenizer.decode(generated)}};
+    } 
+    DecodedResults results;
+    results.texts = {m_tokenizer.decode(generated)};
+
+    // TODO: implement performance metrics
+    results.perf_metrics = ov::genai::PerfMetrics();
+    results.perf_metrics.m_evaluated = false;
+    results.perf_metrics.generate_duration = {0, 0};
+    results.perf_metrics.inference_duration= {0, 0};
+    results.perf_metrics.tokenization_duration = {0, 0};
+    results.perf_metrics.detokenization_duration= {0, 0};
+    results.perf_metrics.ttft = {0, 0};
+    results.perf_metrics.tpot= {0, 0};
+    results.perf_metrics.ipot= {0, 0};
+    results.perf_metrics.throughput= {0, 0};
+    results.perf_metrics.num_generated_tokens = generated.size();
+    results.perf_metrics.num_input_tokens= 0;
+
+    return results;
 }
 
 DecodedResults VLMPipeline::generate(

src/python/py_vlm_pipeline.cpp

@@ -137,9 +137,9 @@ void init_vlm_pipeline(py::module_& m) {
             tokenizer (openvino_genai.Tokenizer): tokenizer object.
             device (str): Device to run the model on (e.g., CPU, GPU). Default is 'CPU'.
         )")
-        .def("start_chat", &ov::genai::VLMPipeline::start_chat)
+        .def("start_chat", &ov::genai::VLMPipeline::start_chat, py::arg("system_message") = "")
         .def("finish_chat", &ov::genai::VLMPipeline::finish_chat) 
-        .def("get_generation_config", static_cast<const ov::genai::GenerationConfig& (ov::genai::VLMPipeline::*)() const>(&ov::genai::VLMPipeline::get_generation_config))
+        .def("get_generation_config", &ov::genai::VLMPipeline::get_generation_config)
         .def(
             "generate", 
             [](ov::genai::VLMPipeline& pipe,