{"id":204039,"date":"2025-05-29T12:26:32","date_gmt":"2025-05-29T04:26:32","guid":{"rendered":"https:\/\/server.hk\/cnblog\/204039\/"},"modified":"2025-05-29T12:26:32","modified_gmt":"2025-05-29T04:26:32","slug":"%e4%bd%bf%e7%94%a8-openvino-%e5%92%8c-postgres-%e6%9e%84%e5%bb%ba%e5%bf%ab%e9%80%9f%e9%ab%98%e6%95%88%e7%9a%84%e8%af%ad%e4%b9%89%e6%90%9c%e7%b4%a2%e7%b3%bb%e7%bb%9f","status":"publish","type":"post","link":"https:\/\/server.hk\/cnblog\/204039\/","title":{"rendered":"\u4f7f\u7528 OpenVINO \u548c Postgres \u6784\u5efa\u5feb\u901f\u9ad8\u6548\u7684\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf"},"content":{"rendered":"<p><b><\/b>     <\/p>\n<h1>\u4f7f\u7528 OpenVINO \u548c Postgres \u6784\u5efa\u5feb\u901f\u9ad8\u6548\u7684\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf<\/h1>\n<p>\u6765\u5230\u7684\u5927\u5bb6\uff0c\u76f8\u4fe1\u90fd\u662f\u7f16\u7a0b\u5b66\u4e60\u7231\u597d\u8005\uff0c\u5e0c\u671b\u5728\u8fd9\u91cc\u5b66\u4e60\u6587\u7ae0\u76f8\u5173\u7f16\u7a0b\u77e5\u8bc6\u3002\u4e0b\u9762\u672c\u7bc7\u6587\u7ae0\u5c31\u6765\u5e26\u5927\u5bb6\u804a\u804a\u300a\u4f7f\u7528 OpenVINO \u548c Postgres \u6784\u5efa\u5feb\u901f\u9ad8\u6548\u7684\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf\u300b\uff0c\u4ecb\u7ecd\u4e00\u4e0b\uff0c\u5e0c\u671b\u5bf9\u5927\u5bb6\u7684\u77e5\u8bc6\u79ef\u7d2f\u6709\u6240\u5e2e\u52a9\uff0c\u52a9\u529b\u5b9e\u6218\u5f00\u53d1\uff01<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.17golang.com\/uploads\/20241024\/1729767113671a26c9701d3.jpg\" class=\"aligncenter\"><\/p>\n<p>\u7167\u7247\u7531 real-napster \u5728 pixabay\u4e0a<\/p>\n<p>\u5728\u6211\u6700\u8fd1\u7684\u4e00\u4e2a\u9879\u76ee\u4e2d\uff0c\u6211\u5fc5\u987b\u6784\u5efa\u4e00\u4e2a\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf\uff0c\u8be5\u7cfb\u7edf\u53ef\u4ee5\u9ad8\u6027\u80fd\u6269\u5c55\u5e76\u4e3a\u62a5\u544a\u641c\u7d22\u63d0\u4f9b\u5b9e\u65f6\u54cd\u5e94\u3002\u6211\u4eec\u5728 aws rds \u4e0a\u4f7f\u7528 postgresql \u548c pgvector\uff0c\u5e76\u642d\u914d aws lambda \u6765\u5b9e\u73b0\u8fd9\u4e00\u76ee\u6807\u3002\u9762\u4e34\u7684\u6311\u6218\u662f\u5141\u8bb8\u7528\u6237\u4f7f\u7528\u81ea\u7136\u8bed\u8a00\u67e5\u8be2\u800c\u4e0d\u662f\u4f9d\u8d56\u6b7b\u677f\u7684\u5173\u952e\u5b57\u8fdb\u884c\u641c\u7d22\uff0c\u540c\u65f6\u786e\u4fdd\u54cd\u5e94\u65f6\u95f4\u5728 1-2 \u79d2\u751a\u81f3\u66f4\u77ed\uff0c\u5e76\u4e14\u53ea\u80fd\u5229\u7528 cpu \u8d44\u6e90\u3002<\/p>\n<p>\u5728\u8fd9\u7bc7\u6587\u7ae0\u4e2d\uff0c\u6211\u5c06\u9010\u6b65\u4ecb\u7ecd\u6784\u5efa\u6b64\u641c\u7d22\u7cfb\u7edf\u7684\u6b65\u9aa4\uff0c\u4ece\u68c0\u7d22\u5230\u91cd\u65b0\u6392\u540d\uff0c\u4ee5\u53ca\u4f7f\u7528 openvino \u548c\u667a\u80fd\u6279\u5904\u7406\u8fdb\u884c\u6807\u8bb0\u5316\u7684\u4f18\u5316\u3002<\/p>\n<p>\u73b0\u4ee3\u6700\u5148\u8fdb\u7684\u641c\u7d22\u7cfb\u7edf\u901a\u5e38\u5305\u542b\u4e24\u4e2a\u4e3b\u8981\u6b65\u9aa4\uff1a<strong>\u68c0\u7d22<\/strong>\u548c<strong>\u91cd\u65b0\u6392\u540d<\/strong>\u3002<\/p>\n<p>1) <strong>\u68c0\u7d22\uff1a<\/strong> \u7b2c\u4e00\u6b65\u6d89\u53ca\u6839\u636e\u7528\u6237\u67e5\u8be2\u68c0\u7d22\u76f8\u5173\u6587\u6863\u7684\u5b50\u96c6\u3002\u8fd9\u53ef\u4ee5\u4f7f\u7528\u9884\u5148\u8bad\u7ec3\u7684\u5d4c\u5165\u6a21\u578b\u6765\u5b8c\u6210\uff0c\u4f8b\u5982 openai \u7684\u5c0f\u578b\u548c\u5927\u578b\u5d4c\u5165\u3001cohere \u7684\u5d4c\u5165\u6a21\u578b\u6216 mixbread \u7684 mxbai \u5d4c\u5165\u3002\u68c0\u7d22\u7684\u91cd\u70b9\u662f\u901a\u8fc7\u6d4b\u91cf\u6587\u6863\u4e0e\u67e5\u8be2\u7684\u76f8\u4f3c\u6027\u6765\u7f29\u5c0f\u6587\u6863\u6c60\u7684\u8303\u56f4\u3002<\/p>\n<p>\u8fd9\u662f\u4e00\u4e2a\u4f7f\u7528 huggingface \u7684\u53e5\u5b50\u8f6c\u6362\u5668\u5e93\u8fdb\u884c\u68c0\u7d22\u7684\u7b80\u5316\u793a\u4f8b\uff0c\u8fd9\u662f\u6211\u6700\u559c\u6b22\u7684\u5e93\u4e4b\u4e00\uff1a<\/p>\n<pre>from sentence_transformers import sentencetransformer\nimport numpy as np\n\n# load a pre-trained sentence transformer model\nmodel = sentencetransformer(\"sentence-transformers\/all-minilm-l6-v2\")\n\n# sample query and documents (vectorize the query and the documents)\nquery = \"how do i fix a broken landing gear?\"\ndocuments = [\"report 1 on landing gear failure\", \"report 2 on engine problems\"]\n\n# get embeddings for query and documents\nquery_embedding = model.encode(query)\ndocument_embeddings = model.encode(documents)\n\n# calculate cosine similarity between query and documents\nsimilarities = np.dot(document_embeddings, query_embedding)\n\n# retrieve top-k most relevant documents\ntop_k = np.argsort(similarities)[-5:]\nprint(\"top 5 documents:\", [documents[i] for i in top_k])\n<\/pre>\n<p>2\uff09<strong>\u91cd\u65b0\u6392\u540d\uff1a<\/strong>\u68c0\u7d22\u5230\u6700\u76f8\u5173\u7684\u6587\u6863\u540e\uff0c\u6211\u4eec\u4f7f\u7528<strong>\u4ea4\u53c9\u7f16\u7801\u5668<\/strong>\u6a21\u578b\u8fdb\u4e00\u6b65\u63d0\u9ad8\u8fd9\u4e9b\u6587\u6863\u7684\u6392\u540d\u3002\u6b64\u6b65\u9aa4\u4f1a\u66f4\u51c6\u786e\u5730\u91cd\u65b0\u8bc4\u4f30\u4e0e\u67e5\u8be2\u76f8\u5173\u7684\u6bcf\u4e2a\u6587\u6863\uff0c\u91cd\u70b9\u5173\u6ce8\u66f4\u6df1\u5165\u7684\u4e0a\u4e0b\u6587\u7406\u89e3\u3002<br \/> \u91cd\u65b0\u6392\u540d\u662f\u6709\u76ca\u7684\uff0c\u56e0\u4e3a\u5b83\u901a\u8fc7\u66f4\u7cbe\u786e\u5730\u8bc4\u200b\u200b\u5206\u6bcf\u4e2a\u6587\u6863\u7684\u76f8\u5173\u6027\u6765\u589e\u52a0\u989d\u5916\u7684\u7ec6\u5316\u5c42\u3002<\/p>\n<p>\u8fd9\u662f\u4f7f\u7528 cross-encoder\/ms-marco-tinybert-l-2-v2\uff08\u4e00\u79cd\u8f7b\u91cf\u7ea7\u4ea4\u53c9\u7f16\u7801\u5668\uff09\u8fdb\u884c\u91cd\u65b0\u6392\u540d\u7684\u4ee3\u7801\u793a\u4f8b\uff1a<\/p>\n<pre>from sentence_transformers import crossencoder\n\n# load the cross-encoder model\ncross_encoder = crossencoder(\"cross-encoder\/ms-marco-tinybert-l-2-v2\")\n\n# use the cross-encoder to rerank top-k retrieved documents\nquery_document_pairs = [(query, doc) for doc in documents]\nscores = cross_encoder.predict(query_document_pairs)\n\n# rank documents based on the new scores\ntop_k_reranked = np.argsort(scores)[-5:]\nprint(\"top 5 reranked documents:\", [documents[i] for i in top_k_reranked])\n<\/pre>\n<p>\u5728\u5f00\u53d1\u8fc7\u7a0b\u4e2d\uff0c\u6211\u53d1\u73b0\u5728\u4f7f\u7528\u53e5\u5b50\u8f6c\u6362\u5668\u7684\u9ed8\u8ba4\u8bbe\u7f6e\u5904\u7406 1,000 \u4e2a\u62a5\u544a\u65f6\uff0c\u6807\u8bb0\u5316\u548c\u9884\u6d4b\u9636\u6bb5\u82b1\u8d39\u4e86\u76f8\u5f53\u957f\u7684\u65f6\u95f4\u3002\u8fd9\u9020\u6210\u4e86\u6027\u80fd\u74f6\u9888\uff0c\u7279\u522b\u662f\u56e0\u4e3a\u6211\u4eec\u7684\u76ee\u6807\u662f\u5b9e\u65f6\u54cd\u5e94\u3002<\/p>\n<p>\u4e0b\u9762\u6211\u4f7f\u7528 snakeviz \u5206\u6790\u4e86\u6211\u7684\u4ee3\u7801\u4ee5\u53ef\u89c6\u5316\u6027\u80fd\uff1a<\/p>\n<p><img decoding=\"async\" src=\"https:\/\/www.17golang.com\/uploads\/20241024\/1729767113671a26c971805.jpg\" class=\"aligncenter\"><\/p>\n<p>\u5982\u60a8\u6240\u89c1\uff0c\u6807\u8bb0\u5316\u548c\u9884\u6d4b\u6b65\u9aa4\u5f02\u5e38\u7f13\u6162\uff0c\u5bfc\u81f4\u641c\u7d22\u7ed3\u679c\u7684\u63d0\u4f9b\u51fa\u73b0\u4e25\u91cd\u5ef6\u8fdf\u3002\u603b\u7684\u6765\u8bf4\uff0c\u5e73\u5747\u9700\u8981 4-5 \u79d2\u3002\u8fd9\u662f\u56e0\u4e3a\u6807\u8bb0\u5316\u548c\u9884\u6d4b\u6b65\u9aa4\u4e4b\u95f4\u5b58\u5728\u963b\u585e\u64cd\u4f5c\u3002\u5982\u679c\u6211\u4eec\u8fd8\u6dfb\u52a0\u5176\u4ed6\u64cd\u4f5c\uff0c\u4f8b\u5982\u6570\u636e\u5e93\u8c03\u7528\u3001\u8fc7\u6ee4\u7b49\uff0c\u6211\u4eec\u5f88\u5bb9\u6613\u5c31\u603b\u5171\u9700\u8981 8-9 \u79d2\u3002<\/p>\n<p>\u6211\u9762\u4e34\u7684\u95ee\u9898\u662f\uff1a<strong>\u6211\u4eec\u53ef\u4ee5\u8ba9\u5b83\u66f4\u5feb\u5417\uff1f<\/strong>\u7b54\u6848\u662f\u80af\u5b9a\u7684\uff0c\u901a\u8fc7\u5229\u7528 <strong>openvino<\/strong>\uff0c\u4e00\u4e2a\u9488\u5bf9 cpu \u63a8\u7406\u4f18\u5316\u7684\u540e\u7aef\u3002 openvino \u6709\u52a9\u4e8e\u52a0\u901f\u82f1\u7279\u5c14\u786c\u4ef6\u4e0a\u7684\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u63a8\u7406\uff0c\u6211\u4eec\u5728 aws lambda \u4e0a\u4f7f\u7528\u8be5\u786c\u4ef6\u3002<\/p>\n<p><strong>openvino \u4f18\u5316\u7684\u4ee3\u7801\u793a\u4f8b<\/strong><br \/> \u4ee5\u4e0b\u662f\u6211\u5982\u4f55\u5c06 openvino \u96c6\u6210\u5230\u641c\u7d22\u7cfb\u7edf\u4e2d\u4ee5\u52a0\u5feb\u63a8\u7406\u901f\u5ea6\uff1a<\/p>\n<pre>import argparse\nimport numpy as np\nimport pandas as pd\nfrom typing import Any\nfrom openvino.runtime import Core\nfrom transformers import AutoTokenizer\n\n\ndef load_openvino_model(model_path: str) -&gt; Core:\n    core = Core()\n    model = core.read_model(model_path + \".xml\")\n    compiled_model = core.compile_model(model, \"CPU\")\n    return compiled_model\n\n\ndef rerank(\n    compiled_model: Core,\n    query: str,\n    results: list[str],\n    tokenizer: AutoTokenizer,\n    batch_size: int,\n) -&gt; np.ndarray[np.float32, Any]:\n    max_length = 512\n    all_logits = []\n\n    # Split results into batches\n    for i in range(0, len(results), batch_size):\n        batch_results = results[i : i + batch_size]\n        inputs = tokenizer(\n            [(query, item) for item in batch_results],\n            padding=True,\n            truncation=\"longest_first\",\n            max_length=max_length,\n            return_tensors=\"np\",\n        )\n\n        # Extract input tensors (convert to NumPy arrays)\n        input_ids = inputs[\"input_ids\"].astype(np.int32)\n        attention_mask = inputs[\"attention_mask\"].astype(np.int32)\n        token_type_ids = inputs.get(\"token_type_ids\", np.zeros_like(input_ids)).astype(\n            np.int32\n        )\n\n        infer_request = compiled_model.create_infer_request()\n        output = infer_request.infer(\n            {\n                \"input_ids\": input_ids,\n                \"attention_mask\": attention_mask,\n                \"token_type_ids\": token_type_ids,\n            }\n        )\n\n        logits = output[\"logits\"]\n        all_logits.append(logits)\n\n    all_logits = np.concatenate(all_logits, axis=0)\n    return all_logits\n\n\ndef fetch_search_data(search_text: str) -&gt; pd.DataFrame:\n    # Usually you would fetch the data from a database\n    df = pd.read_csv(\"cnbc_headlines.csv\")\n    df = df[~df[\"Headlines\"].isnull()]\n\n    texts = df[\"Headlines\"].tolist()\n\n    # Load the model and rerank\n    openvino_model = load_openvino_model(\"cross-encoder-openvino-model\/model\")\n    tokenizer = AutoTokenizer.from_pretrained(\"cross-encoder\/ms-marco-TinyBERT-L-2-v2\")\n    rerank_scores = rerank(openvino_model, search_text, texts, tokenizer, batch_size=16)\n\n    # Add the rerank scores to the DataFrame and sort by the new scores\n    df[\"rerank_score\"] = rerank_scores\n    df = df.sort_values(by=\"rerank_score\", ascending=False)\n\n    return df\n\n\nif __name__ == \"__main__\":\n    parser = argparse.ArgumentParser(\n        description=\"Fetch search results with reranking using OpenVINO\"\n    )\n\n    parser.add_argument(\n        \"--search_text\",\n        type=str,\n        required=True,\n        help=\"The search text to use for reranking\",\n    )\n\n    args = parser.parse_args()\n\n    df = fetch_search_data(args.search_text)\n    print(df)\n<\/pre>\n<p>\u901a\u8fc7\u8fd9\u79cd\u65b9\u6cd5\uff0c\u6211\u4eec\u53ef\u4ee5\u83b7\u5f97 2-3 \u500d\u7684\u52a0\u901f\uff0c\u5c06\u539f\u6765\u7684 4-5 \u79d2\u51cf\u5c11\u5230 1-2 \u79d2\u3002\u5b8c\u6574\u7684\u5de5\u4f5c\u4ee3\u7801\u4f4d\u4e8e github \u4e0a\u3002<\/p>\n<p>\u63d0\u9ad8\u6027\u80fd\u7684\u53e6\u4e00\u4e2a\u5173\u952e\u56e0\u7d20\u662f\u4f18\u5316<strong>\u6807\u8bb0\u5316<\/strong>\u6d41\u7a0b\u5e76\u8c03\u6574<strong>\u6279\u91cf\u5927\u5c0f<\/strong>\u548c<strong>\u6807\u8bb0\u957f\u5ea6<\/strong>\u3002\u901a\u8fc7\u589e\u52a0\u6279\u91cf\u5927\u5c0f\uff08batch_size = 16\uff09\u548c\u51cf\u5c11\u4ee4\u724c\u957f\u5ea6\uff08max_length = 512\uff09\uff0c\u6211\u4eec\u53ef\u4ee5\u5e76\u884c\u5316\u4ee4\u724c\u5316\u5e76\u51cf\u5c11\u91cd\u590d\u64cd\u4f5c\u7684\u5f00\u9500\u3002\u5728\u6211\u4eec\u7684\u5b9e\u9a8c\u4e2d\uff0c\u6211\u4eec\u53d1\u73b0 16 \u5230 64 \u4e4b\u95f4\u7684 batch_size \u6548\u679c\u5f88\u597d\uff0c\u4efb\u4f55\u66f4\u5927\u7684\u503c\u90fd\u4f1a\u964d\u4f4e\u6027\u80fd\u3002\u540c\u6837\uff0c\u6211\u4eec\u5c06 max_length \u8bbe\u7f6e\u4e3a 128\uff0c\u5982\u679c\u62a5\u544a\u7684\u5e73\u5747\u957f\u5ea6\u76f8\u5bf9\u8f83\u77ed\uff0c\u5219\u8be5\u503c\u662f\u53ef\u884c\u7684\u3002\u901a\u8fc7\u8fd9\u4e9b\u66f4\u6539\uff0c\u6211\u4eec\u5b9e\u73b0\u4e86 8 \u500d\u7684\u6574\u4f53\u52a0\u901f\uff0c\u5c06\u91cd\u65b0\u6392\u540d\u65f6\u95f4\u7f29\u77ed\u81f3 1 \u79d2\u4ee5\u4e0b\uff0c\u5373\u4f7f\u5728 cpu \u4e0a\u4e5f\u662f\u5982\u6b64\u3002<\/p>\n<p>\u5728\u5b9e\u8df5\u4e2d\uff0c\u8fd9\u610f\u5473\u7740\u5c1d\u8bd5\u4e0d\u540c\u7684\u6279\u91cf\u5927\u5c0f\u548c\u4ee4\u724c\u957f\u5ea6\uff0c\u4ee5\u627e\u5230\u6570\u636e\u901f\u5ea6\u548c\u51c6\u786e\u6027\u4e4b\u95f4\u7684\u9002\u5f53\u5e73\u8861\u3002\u901a\u8fc7\u8fd9\u6837\u505a\uff0c\u6211\u4eec\u770b\u5230\u4e86\u54cd\u5e94\u65f6\u95f4\u7684\u663e\u7740\u6539\u8fdb\uff0c\u4f7f\u5f97\u641c\u7d22\u7cfb\u7edf\u5373\u4f7f\u6709 1,000 \u591a\u4e2a\u62a5\u544a\u4e5f\u53ef\u6269\u5c55\u3002<\/p>\n<p>\u901a\u8fc7\u4f7f\u7528 openvino \u5e76\u4f18\u5316\u6807\u8bb0\u5316\u548c\u6279\u5904\u7406\uff0c\u6211\u4eec\u80fd\u591f\u6784\u5efa\u4e00\u4e2a\u9ad8\u6027\u80fd\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf\uff0c\u6ee1\u8db3\u4ec5 cpu \u8bbe\u7f6e\u7684\u5b9e\u65f6\u8981\u6c42\u3002\u4e8b\u5b9e\u4e0a\uff0c\u6211\u4eec\u7684\u6574\u4f53\u901f\u5ea6\u63d0\u5347\u4e86 8 \u500d\u3002\u4f7f\u7528\u53e5\u5b50\u8f6c\u6362\u5668\u8fdb\u884c\u68c0\u7d22\u4e0e\u4f7f\u7528\u4ea4\u53c9\u7f16\u7801\u5668\u6a21\u578b\u8fdb\u884c\u91cd\u65b0\u6392\u540d\u76f8\u7ed3\u5408\uff0c\u521b\u9020\u4e86\u5f3a\u5927\u7684\u3001\u7528\u6237\u53cb\u597d\u7684\u641c\u7d22\u4f53\u9a8c\u3002<\/p>\n<p>\u5982\u679c\u60a8\u6b63\u5728\u6784\u5efa\u54cd\u5e94\u65f6\u95f4\u548c\u8ba1\u7b97\u8d44\u6e90\u53d7\u5230\u9650\u5236\u7684\u7c7b\u4f3c\u7cfb\u7edf\uff0c\u6211\u5f3a\u70c8\u5efa\u8bae\u60a8\u63a2\u7d22 openvino \u548c\u667a\u80fd\u6279\u5904\u7406\u4ee5\u91ca\u653e\u66f4\u597d\u7684\u6027\u80fd\u3002<\/p>\n<p>\u5e0c\u671b\u60a8\u559c\u6b22\u8fd9\u7bc7\u6587\u7ae0\u3002\u5982\u679c\u60a8\u89c9\u5f97\u8fd9\u7bc7\u6587\u7ae0\u6709\u7528\uff0c\u8bf7\u7ed9\u6211\u4e00\u4e2a\u8d5e\uff0c\u4ee5\u4fbf\u5176\u4ed6\u4eba\u4e5f\u53ef\u4ee5\u627e\u5230\u5b83\uff0c\u5e76\u4e0e\u60a8\u7684\u670b\u53cb\u5206\u4eab\u3002\u5728 linkedin \u4e0a\u5173\u6ce8\u6211\uff0c\u4e86\u89e3\u6211\u7684\u6700\u65b0\u5de5\u4f5c\u3002\u611f\u8c22\u60a8\u7684\u9605\u8bfb\uff01<\/p>\n<p>\u597d\u4e86\uff0c\u672c\u6587\u5230\u6b64\u7ed3\u675f\uff0c\u5e26\u5927\u5bb6\u4e86\u89e3\u4e86\u300a\u4f7f\u7528 OpenVINO \u548c Postgres \u6784\u5efa\u5feb\u901f\u9ad8\u6548\u7684\u8bed\u4e49\u641c\u7d22\u7cfb\u7edf\u300b\uff0c\u5e0c\u671b\u672c\u6587\u5bf9\u4f60\u6709\u6240\u5e2e\u52a9\uff01\u5173\u6ce8\u516c\u4f17\u53f7\uff0c\u7ed9\u5927\u5bb6\u5206\u4eab\u66f4\u591a\u6587\u7ae0\u77e5\u8bc6\uff01<\/p>\n<p>      \u7248\u672c\u58f0\u660e \u672c\u6587\u8f6c\u8f7d\u4e8e\uff1adev.to \u5982\u6709\u4fb5\u72af\uff0c\u8bf7\u8054\u7cfb\u5220\u9664<\/p>\n","protected":false},"excerpt":{"rendered":"<p>\u4f7f\u7528 OpenVINO \u548c Po&#46;&#46;&#46;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[4925],"tags":[],"class_list":["post-204039","post","type-post","status-publish","format-standard","hentry","category-4925"],"_links":{"self":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts\/204039","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/comments?post=204039"}],"version-history":[{"count":0,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/posts\/204039\/revisions"}],"wp:attachment":[{"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/media?parent=204039"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/categories?post=204039"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/server.hk\/cnblog\/wp-json\/wp\/v2\/tags?post=204039"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}