Updates properties (openvinotoolkit#2403)

CVS-120640

Author: aleksandr-mokrov

.ci/skipped_notebooks.yml

@@ -586,4 +586,8 @@
     - python:
       - '3.8'
     - os:
-        - macos-12
+        - macos-12
+- notebook: notebooks/llm-agent-react/llm-agent-react-langchain.ipynb
+  skips:
+    - python:
+      - '3.8'

.ci/spellcheck/.pyspelling.wordlist.txt

@@ -617,6 +617,7 @@ perceptron
 Patil
 PEFT
 perceiver
+PerformanceMode
 performant
 PersonaGPT
 PGI

notebooks/auto-device/auto-device.ipynb

@@ -70,7 +70,7 @@
     "import platform\n",
     "\n",
     "# Install required packages\n",
-    "%pip install -q \"openvino>=2023.1.0\" Pillow torch torchvision tqdm --extra-index-url https://download.pytorch.org/whl/cpu\n",
+    "%pip install -q \"openvino>=2023.1.0\" \"numpy<2\" Pillow torch torchvision tqdm --extra-index-url https://download.pytorch.org/whl/cpu\n",
     "\n",
     "if platform.system() != \"Windows\":\n",
     "    %pip install -q \"matplotlib>=3.4\"\n",
@@ -187,8 +187,11 @@
     }
    ],
    "source": [
+    "import openvino.properties.log as log\n",
+    "\n",
+    "\n",
     "# Set LOG_LEVEL to LOG_INFO.\n",
-    "core.set_property(\"AUTO\", {\"LOG_LEVEL\": \"LOG_INFO\"})\n",
+    "core.set_property(\"AUTO\", {log.level(): log.Level.INFO})\n",
     "\n",
     "# Load the model onto the target device.\n",
     "compiled_model = core.compile_model(ov_model)\n",
@@ -249,7 +252,7 @@
    ],
    "source": [
     "# Set LOG_LEVEL to LOG_NONE.\n",
-    "core.set_property(\"AUTO\", {\"LOG_LEVEL\": \"LOG_NONE\"})\n",
+    "core.set_property(\"AUTO\", {log.level(): log.Level.NO})\n",
     "\n",
     "compiled_model = core.compile_model(model=ov_model, device_name=\"AUTO\")\n",
     "\n",
@@ -611,12 +614,15 @@
     }
    ],
    "source": [
+    "import openvino.properties.hint as hints\n",
+    "\n",
+    "\n",
     "THROUGHPUT_hint_context = InferContext(metrics_update_interval, metrics_update_num)\n",
     "\n",
     "print(\"Compiling Model for AUTO device with THROUGHPUT hint\")\n",
     "sys.stdout.flush()\n",
     "\n",
-    "compiled_model = core.compile_model(model=ov_model, config={\"PERFORMANCE_HINT\": \"THROUGHPUT\"})\n",
+    "compiled_model = core.compile_model(model=ov_model, config={hints.performance_mode(): hints.PerformanceMode.THROUGHPUT})\n",
     "\n",
     "infer_queue = ov.AsyncInferQueue(compiled_model, 0)  # Setting to 0 will query optimal number by default.\n",
     "infer_queue.set_callback(completion_callback)\n",
@@ -680,7 +686,7 @@
     "print(\"Compiling Model for AUTO Device with LATENCY hint\")\n",
     "sys.stdout.flush()\n",
     "\n",
-    "compiled_model = core.compile_model(model=ov_model, config={\"PERFORMANCE_HINT\": \"LATENCY\"})\n",
+    "compiled_model = core.compile_model(model=ov_model, config={hints.performance_mode(): hints.PerformanceMode.LATENCY})\n",
     "\n",
     "# Setting to 0 will query optimal number by default.\n",
     "infer_queue = ov.AsyncInferQueue(compiled_model, 0)\n",

notebooks/clip-language-saliency-map/clip-language-saliency-map.ipynb

@@ -90,7 +90,7 @@
    "source": [
     "# Install requirements\n",
     "%pip install -q \"openvino>=2023.1.0\"\n",
-    "%pip install -q --extra-index-url https://download.pytorch.org/whl/cpu transformers \"torch>=2.1\" \"gradio>=4.19\""
+    "%pip install -q --extra-index-url https://download.pytorch.org/whl/cpu transformers \"numpy<2\" \"torch>=2.1\" \"gradio>=4.19\""
    ]
   },
   {
@@ -759,14 +759,15 @@
    "outputs": [],
    "source": [
     "from typing import Dict, Any\n",
+    "import openvino.properties.hint as hints\n",
     "\n",
     "\n",
     "image_model = core.read_model(image_model_path)\n",
     "\n",
     "image_model = core.compile_model(\n",
     "    model=image_model,\n",
     "    device_name=device.value,\n",
-    "    config={\"PERFORMANCE_HINT\": \"THROUGHPUT\"},\n",
+    "    config={hints.performance_mode(): hints.PerformanceMode.THROUGHPUT},\n",
     ")"
    ]
   },

notebooks/cross-lingual-books-alignment/cross-lingual-books-alignment.ipynb

@@ -1105,11 +1105,13 @@
    "source": [
     "from typing import Any\n",
     "\n",
+    "import openvino.properties.hint as hints\n",
+    "\n",
     "\n",
     "compiled_throughput_hint = core.compile_model(\n",
     "    ov_model,\n",
     "    device_name=device.value,\n",
-    "    config={\"PERFORMANCE_HINT\": \"THROUGHPUT\"},\n",
+    "    config={hints.performance_mode(): hints.PerformanceMode.THROUGHPUT},\n",
     ")"
    ]
   },
@@ -1323,7 +1325,10 @@
     }
    ],
    "source": [
-    "cpu_name = core.get_property(\"CPU\", \"FULL_DEVICE_NAME\")\n",
+    "import openvino.properties as props\n",
+    "\n",
+    "\n",
+    "cpu_name = core.get_property(\"CPU\", props.device.full_name)\n",
     "\n",
     "plot = sns.barplot(benchmark_dataframe, errorbar=\"sd\")\n",
     "plot.set(ylabel=\"Sentences Per Second\", title=f\"Sentence Embeddings Benchmark\\n{cpu_name}\")\n",

notebooks/ct-segmentation-quantize/async_pipeline.py

@@ -24,6 +24,7 @@
 
 import cv2
 
+import openvino.properties as props
 from custom_segmentation import Model
 
 
@@ -169,15 +170,15 @@ def __init__(self, ie, model, plugin_config, device="CPU", max_num_requests=0):
         cache_path.mkdir(exist_ok=True)
         # Enable model caching for GPU devices
         if "GPU" in device and "GPU" in ie.available_devices:
-            ie.set_property(device_name="GPU", properties={"CACHE_DIR": str(cache_path)})
+            ie.set_property(device_name="GPU", properties={props.cache_dir(): str(cache_path)})
 
         self.model = model
         self.logger = logging.getLogger()
 
         self.logger.info("Loading network to {} plugin...".format(device))
         self.exec_net = ie.compile_model(self.model.net, device, plugin_config)
         if max_num_requests == 0:
-            max_num_requests = self.exec_net.get_property("OPTIMAL_NUMBER_OF_INFER_REQUESTS") + 1
+            max_num_requests = self.exec_net.get_property(props.optimal_number_of_infer_requests) + 1
         self.requests = [self.exec_net.create_infer_request() for _ in range(max_num_requests)]
         self.empty_requests = deque(self.requests)
         self.completed_request_results = {}

notebooks/dolly-2-instruction-following/dolly-2-instruction-following.ipynb

@@ -564,9 +564,15 @@
    ],
    "source": [
     "from pathlib import Path\n",
+    "\n",
     "from transformers import AutoTokenizer\n",
     "from optimum.intel.openvino import OVModelForCausalLM\n",
     "\n",
+    "import openvino.properties as props\n",
+    "import openvino.properties.hint as hints\n",
+    "import openvino.properties.streams as streams\n",
+    "\n",
+    "\n",
     "if model_to_run.value == \"INT4\":\n",
     "    model_dir = int4_model_dir\n",
     "elif model_to_run.value == \"INT8\":\n",
@@ -579,7 +585,7 @@
     "\n",
     "current_device = device.value\n",
     "\n",
-    "ov_config = {\"PERFORMANCE_HINT\": \"LATENCY\", \"NUM_STREAMS\": \"1\", \"CACHE_DIR\": \"\"}\n",
+    "ov_config = {hints.performance_mode(): hints.PerformanceMode.LATENCY, streams.num(): \"1\", props.cache_dir(): \"\"}\n",
     "\n",
     "ov_model = OVModelForCausalLM.from_pretrained(model_dir, device=current_device, ov_config=ov_config)"
    ]

notebooks/florence2/gradio_helper.py

@@ -63,7 +63,6 @@ def plot_bbox(image, data):
 
 
 def draw_polygons(image, prediction, fill_mask=False):
-
     draw = ImageDraw.Draw(image)
     scale = 1
     for polygons, label in zip(prediction["polygons"], prediction["labels"]):

notebooks/florence2/ov_florence2_helper.py

@@ -353,7 +353,6 @@ def __init__(self, model_dir, device, ov_config=None) -> None:
         self.language_model = OVFlorence2LangModel(model_dir, self.config.text_config, device, ov_config)
 
     def generate(self, input_ids, inputs_embeds=None, pixel_values=None, **kwargs):
-
         if inputs_embeds is None:
             # 1. Extra the input embeddings
             if input_ids is not None:

notebooks/gpu-device/gpu-device.ipynb

@@ -256,9 +256,12 @@
     }
    ],
    "source": [
+    "import openvino.properties as props\n",
+    "\n",
+    "\n",
     "device = \"GPU\"\n",
     "\n",
-    "core.get_property(device, \"FULL_DEVICE_NAME\")"
+    "core.get_property(device, props.device.full_name)"
    ]
   },
   {
@@ -267,7 +270,7 @@
    "id": "aac3129a-129f-49aa-aba0-71ae1e892ada",
    "metadata": {},
    "source": [
-    "Each device also has a specific property called `SUPPORTED_PROPERTIES`, that enables viewing all the available properties in the device. We can check the value for each property by simply looping through the dictionary returned by `core.get_property(\"GPU\", \"SUPPORTED_PROPERTIES\")` and then querying for that property."
+    "Each device also has a specific property called `SUPPORTED_PROPERTIES`, that enables viewing all the available properties in the device. We can check the value for each property by simply looping through the dictionary returned by `core.get_property(\"GPU\", props.supported_properties)` and then querying for that property."
    ]
   },
   {
@@ -321,7 +324,7 @@
    ],
    "source": [
     "print(f\"{device} SUPPORTED_PROPERTIES:\\n\")\n",
-    "supported_properties = core.get_property(device, \"SUPPORTED_PROPERTIES\")\n",
+    "supported_properties = core.get_property(device, props.supported_properties)\n",
     "indent = len(max(supported_properties, key=len))\n",
     "\n",
     "for property_key in supported_properties:\n",
@@ -677,7 +680,7 @@
     "core = ov.Core()\n",
     "\n",
     "# Set cache folder\n",
-    "core.set_property({\"CACHE_DIR\": cache_folder})\n",
+    "core.set_property({props.cache_dir(): cache_folder})\n",
     "\n",
     "# Compile the model as before\n",
     "model = core.read_model(model=model_path)\n",
@@ -717,7 +720,7 @@
    "source": [
     "start = time.time()\n",
     "core = ov.Core()\n",
-    "core.set_property({\"CACHE_DIR\": \"cache\"})\n",
+    "core.set_property({props.cache_dir(): \"cache\"})\n",
     "model = core.read_model(model=model_path)\n",
     "compiled_model = core.compile_model(model, device)\n",
     "print(f\"Cache enabled  - compile time: {time.time() - start}s\")\n",
@@ -765,7 +768,7 @@
    "id": "7077b662-22f3-4c52-9c80-e5ac1309c482",
    "metadata": {},
    "source": [
-    "To use the \"LATENCY\" performance hint, add `{\"PERFORMANCE_HINT\": \"LATENCY\"}` when compiling the model as shown below. For GPUs, this automatically minimizes the batch size and number of parallel streams such that all of the compute resources can focus on completing a single inference as fast as possible."
+    "To use the \"LATENCY\" performance hint, add `{hints.performance_mode(): hints.PerformanceMode.LATENCY}` when compiling the model as shown below. For GPUs, this automatically minimizes the batch size and number of parallel streams such that all of the compute resources can focus on completing a single inference as fast as possible."
    ]
   },
   {
@@ -780,7 +783,10 @@
    },
    "outputs": [],
    "source": [
-    "compiled_model = core.compile_model(model, device, {\"PERFORMANCE_HINT\": \"LATENCY\"})"
+    "import openvino.properties.hint as hints\n",
+    "\n",
+    "\n",
+    "compiled_model = core.compile_model(model, device, {hints.performance_mode(): hints.PerformanceMode.LATENCY})"
    ]
   },
   {
@@ -789,7 +795,7 @@
    "id": "06589f38-ce35-457f-8395-a4a3f6327ea0",
    "metadata": {},
    "source": [
-    "To use the \"THROUGHPUT\" performance hint, add `{\"PERFORMANCE_HINT\": \"THROUGHPUT\"}` when compiling the model. For GPUs, this creates multiple processing streams to efficiently utilize all the execution cores and optimizes the batch size to fill the available memory."
+    "To use the \"THROUGHPUT\" performance hint, add `{hints.performance_mode(): hints.PerformanceMode.THROUGHPUT}` when compiling the model. For GPUs, this creates multiple processing streams to efficiently utilize all the execution cores and optimizes the batch size to fill the available memory."
    ]
   },
   {
@@ -804,7 +810,7 @@
    },
    "outputs": [],
    "source": [
-    "compiled_model = core.compile_model(model, device, {\"PERFORMANCE_HINT\": \"THROUGHPUT\"})"
+    "compiled_model = core.compile_model(model, device, {hints.performance_mode(): hints.PerformanceMode.THROUGHPUT})"
    ]
   },
   {
@@ -836,7 +842,9 @@
     "Note that we always need to explicitly specify the device list for MULTI to work, otherwise MULTI does not know which devices are available for inference. However, this is not the only way to use multiple devices in OpenVINO. There is another performance hint called \"CUMULATIVE_THROUGHPUT\" that works similar to MULTI, except it uses the devices automatically selected by AUTO. This way, we do not need to manually specify devices to use. Below is an example showing how to use \"CUMULATIVE_THROUGHPUT\", equivalent to the MULTI one:\n",
     "\n",
     "`\n",
-    "compiled_model = core.compile_model(model=model, device_name=\"AUTO\", config={\"PERFORMANCE_HINT\": \"CUMULATIVE_THROUGHPUT\"})\n",
+    "\n",
+    "\n",
+    "compiled_model = core.compile_model(model=model, device_name=\"AUTO\", config={hints.performance_mode(): hints.PerformanceMode.CUMULATIVE_THROUGHPUT})\n",
     "`\n",
     "\n",
     "> **Important**: **The “THROUGHPUT”, “MULTI”, and “CUMULATIVE_THROUGHPUT” modes are only applicable to asynchronous inferencing pipelines. The example at the end of this article shows how to set up an asynchronous pipeline that takes advantage of parallelism to increase throughput.** To learn more, see [Asynchronous Inferencing](https://docs.openvino.ai/2024/documentation/openvino-extensibility/openvino-plugin-library/asynch-inference-request.html) in OpenVINO as well as the [Asynchronous Inference notebook](../async-api/async-api.ipynb)."
@@ -1584,7 +1592,7 @@
     "# Read model and compile it on GPU in THROUGHPUT mode\n",
     "model = core.read_model(model=model_path)\n",
     "device_name = \"GPU\"\n",
-    "compiled_model = core.compile_model(model=model, device_name=device_name, config={\"PERFORMANCE_HINT\": \"THROUGHPUT\"})\n",
+    "compiled_model = core.compile_model(model=model, device_name=device_name, config={hints.performance_mode(): hints.PerformanceMode.THROUGHPUT})\n",
     "\n",
     "# Get the input and output nodes\n",
     "input_layer = compiled_model.input(0)\n",
@@ -1996,7 +2004,7 @@
     "    )\n",
     "    cv2.putText(\n",
     "        frame,\n",
-    "        f\"hint {compiled_model.get_property('PERFORMANCE_HINT')}\",\n",
+    "        f\"hint {compiled_model.get_property(hints.performance_mode)}\",\n",
     "        (5, 60),\n",
     "        cv2.FONT_ITALIC,\n",
     "        0.6,\n",

notebooks/grounded-segment-anything/grounded-segment-anything.ipynb

@@ -966,7 +966,6 @@
    "outputs": [],
    "source": [
     "def draw_mask(mask, draw, random_color=False):\n",
-    "\n",
     "    if random_color:\n",
     "        color = (\n",
     "            np.random.randint(0, 255),\n",