{"id":134,"date":"2024-12-01T15:09:08","date_gmt":"2024-12-01T15:09:08","guid":{"rendered":"https:\/\/pacific.ai\/staging\/3667\/?post_type=peer-reviews-paper&#038;p=134"},"modified":"2026-02-19T11:14:34","modified_gmt":"2026-02-19T11:14:34","slug":"holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications","status":"publish","type":"peer-reviews-paper","link":"https:\/\/pacific.ai\/staging\/3667\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/","title":{"rendered":"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications"},"content":{"rendered":"<div id=\"bsf_rt_marker\"><\/div>\n<p>As the adoption of Large Language Models (LLMs) accelerates, their performance evaluation becomes increasingly critical\u2014not just in terms of accuracy but across a broader spectrum of metrics like robustness and toxicity. <\/p>\n<p>A recent paper titled <strong>\u201cHolistic Evaluation of Large Language Models\u201d<\/strong>, published by researchers at John Snow Labs, introduces an innovative approach to assessing LLMs. This framework evaluates models comprehensively, offering insights that go beyond traditional metrics.<\/p>\n<h4>Key Objectives of the Holistic Evaluation Framework<\/h4>\n<h6>1. Broadening the Scope of Evaluation Metrics<\/h6>\n<p>Traditional metrics like BLEU or ROUGE focus narrowly on accuracy, which is insufficient for real-world applications. The new framework includes:<\/p>\n\t<ul>\n        <li>Robustness: How well a model handles text perturbations like typos, slang, or casing changes.<\/li>\n        <li>Toxicity: The ability of a model to avoid generating harmful or offensive outputs.<\/li>\n        <li>Accuracy: Traditional metrics still play a role, offering a baseline for comparison.<\/li>\n    <\/ul>\n<h6>2. LangTest Toolkit and Leaderboard<\/h6>\n<p>A key feature of this framework is the LangTest toolkit, an open-source Python library that enables testing of LLMs for a variety of parameters. Paired with the LangTest Leaderboard, it allows stakeholders to compare models on multiple metrics dynamically.<\/p>\n<h4>Evaluation Metrics and Findings<\/h4>\n<h6>1.\tRobustness<\/h6>\n<p>Robustness tests simulate real-world conditions by introducing minor changes to the input text, such as typos or slang. Models are evaluated based on their ability to maintain accurate predictions under these conditions.<\/p>\n<p><strong>Key Finding:<\/strong> GPT-4 scored highest in robustness, demonstrating resilience against text perturbations. Smaller models like Open Chat 3.5 also performed competitively, highlighting that parameter size isn\u2019t the sole determinant of robustness.<\/p>\n<h6>2. Accuracy<\/h6>\n<p>Accuracy remains a vital metric, focusing on how well models perform tasks like question answering and summarization.<\/p>\n<p><strong>Key Finding:<\/strong> GPT-4 led in accuracy, while smaller models like Mixtral 8x7B and Neural Chat 7B showed commendable performance relative to their size.<\/p>\n<h6>3. Toxicity<\/h6>\n<p>Toxicity evaluation is crucial for applications in sensitive domains. This metric assesses whether models generate harmful or offensive content.<\/p>\n<p><strong>Key Finding:<\/strong> Llama 2 7B excelled in avoiding toxic outputs, making it a preferred choice for applications where content sensitivity is paramount.<\/p>\n<h4>Implications for Real-World Applications<\/h4>\n<p>The framework\u2019s holistic approach is particularly valuable for domain-specific applications like healthcare, legal, and finance. Traditional evaluation metrics often fail to capture the nuances required in these fields, such as compliance with ethical standards or sensitivity to misinformation.<\/p>\n<h6>1. Healthcare<\/h6>\n<p>Models need to provide accurate, bias-free, and clinically relevant outputs to support medical decision-making. Future iterations of the LangTest toolkit will include domain-specific evaluations tailored for healthcare use cases.<\/p>\n<h6>2. Legal and Financial Domains<\/h6>\n<p>Applications in legal and financial fields demand a higher degree of accuracy and transparency. The ability to measure robustness and toxicity ensures that LLMs can operate reliably in these high-stakes environments.<\/p>\n<h4>Innovations in Evaluation<\/h4>\n<p>What sets the LangTest framework apart is its dynamic nature. Unlike static datasets, which become outdated as models evolve, this framework allows continuous updates with new datasets, metrics, and tests. This adaptability ensures that evaluations remain relevant and comprehensive over time.<\/p>\n<p>Additionally, the toolkit supports data augmentation, enabling users to create tailored datasets to address specific weaknesses in model performance.<\/p>\n<h4>Looking Ahead<\/h4>\n<p>The study\u2019s findings emphasize that no single model excels across all metrics. While GPT-4 dominates in accuracy and robustness, smaller open-source models like Neural Chat 7B provide a cost-effective alternative with respectable performance. Similarly, Llama 2 7B\u2019s excellence in toxicity avoidance suggests that specific models can be optimized for unique applications.<\/p>\n<h4>Conclusion<\/h4>\n<p>As AI continues to influence diverse industries, ensuring its reliability, safety, and fairness is more critical than ever. The holistic evaluation framework proposed in this paper offers a robust methodology for assessing LLMs across multiple dimensions, paving the way for more informed decisions in model selection.<\/p>\n<p>The LangTest toolkit and leaderboard stand as valuable resources for researchers, developers, and businesses alike, fostering transparency and accountability in AI development.<\/p>\n<p>For more information, explore the LangTest Leaderboard: <a target=\"_blank\" rel=\"noopener\" href=\"https:\/\/langtest.org\/leaderboard\/llm\">LangTest Leaderboard<\/a>.<\/p>\n\n\n<h2>FAQ<\/h2>\n<p><strong>How does the Holistic Evaluation framework assess LLM performance?<\/strong><\/p>\n<p>The framework\u2014using LangTest\u2014measures performance across robustness, accuracy, and toxicity, enabling evaluation under real-world conditions by including tests for adversarial prompts, correctness, and harmful content.<\/p>\n<p><strong>Which LLMs perform best on robustness and toxicity in the study?<\/strong><\/p>\n<p>GPT\u20114 tops both robustness and accuracy tests, Llama 2 leads toxicity evaluation, while Mixtral 8x7B and certain open\u2011source models (OpenChat 3.5, NeuralChat 7B) perform well across all three dimensions.<\/p>\n<p><strong>What is LangTest and how does it support LLM evaluation?<\/strong><\/p>\n<p>LangTest is an open-source Python toolkit offering 60+ tests\u2014covering robustness, fairness, toxicity, accuracy, and more\u2014to comprehensively evaluate LLMs and NLP models for real-world applications.<\/p>\n<p><strong>Why is a holistic evaluation necessary beyond accuracy metrics?<\/strong><\/p>\n<p>Accuracy alone misses critical issues: robustness ensures models withstand noisy or adversarial inputs, and toxicity checks prevent harmful outputs. Holistic evaluation highlights trade-offs between these dimensions.<\/p>\n<p><strong>How can developers use evaluation results to improve LLM deployment?<\/strong><\/p>\n<p>Developers can consult the LangTest leaderboard to compare model strengths, select models suited to specific use cases, identify weaknesses (e.g., toxicity), and guide fine-tuning or safety interventions.<\/p>\n\n\n<script type=\"application\/ld+json\">\n{\n  \"@context\": \"https:\/\/schema.org\",\n  \"@type\": \"FAQPage\",\n  \"mainEntity\": [\n    {\n      \"@type\": \"Question\",\n      \"name\": \"How does the Holistic Evaluation framework assess LLM performance?\",\n      \"acceptedAnswer\": {\n        \"@type\": \"Answer\",\n        \"text\": \"The framework\u2014using LangTest\u2014measures performance across robustness, accuracy, and toxicity, enabling evaluation under real-world conditions by including tests for adversarial prompts, correctness, and harmful content.\"\n      }\n    },\n    {\n      \"@type\": \"Question\",\n      \"name\": \"Which LLMs perform best on robustness and toxicity in the study?\",\n      \"acceptedAnswer\": {\n        \"@type\": \"Answer\",\n        \"text\": \"GPT-4 tops both robustness and accuracy tests, Llama 2 leads toxicity evaluation, while Mixtral 8x7B and certain open-source models (OpenChat 3.5, NeuralChat 7B) perform well across all three dimensions.\"\n      }\n    },\n    {\n      \"@type\": \"Question\",\n      \"name\": \"What is LangTest and how does it support LLM evaluation?\",\n      \"acceptedAnswer\": {\n        \"@type\": \"Answer\",\n        \"text\": \"LangTest is an open-source Python toolkit offering 60+ tests\u2014covering robustness, fairness, toxicity, accuracy, and more\u2014to comprehensively evaluate LLMs and NLP models for real-world applications.\"\n      }\n    },\n    {\n      \"@type\": \"Question\",\n      \"name\": \"Why is a holistic evaluation necessary beyond accuracy metrics?\",\n      \"acceptedAnswer\": {\n        \"@type\": \"Answer\",\n        \"text\": \"Accuracy alone misses critical issues: robustness ensures models withstand noisy or adversarial inputs, and toxicity checks prevent harmful outputs. Holistic evaluation highlights trade-offs between these dimensions.\"\n      }\n    },\n    {\n      \"@type\": \"Question\",\n      \"name\": \"How can developers use evaluation results to improve LLM deployment?\",\n      \"acceptedAnswer\": {\n        \"@type\": \"Answer\",\n        \"text\": \"Developers can consult the LangTest leaderboard to compare model strengths, select models suited to specific use cases, identify weaknesses (e.g., toxicity), and guide fine-tuning or safety interventions.\"\n      }\n    }\n  ]\n}\n<\/script>\n","protected":false},"featured_media":921,"template":"","meta":{"_acf_changed":true,"nf_dc_page":"","content-type":"","inline_featured_image":false},"class_list":["post-134","peer-reviews-paper","type-peer-reviews-paper","status-publish","has-post-thumbnail","hentry"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.3 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI<\/title>\n<meta name=\"description\" content=\"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability\" \/>\n<meta name=\"robots\" content=\"noindex, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<meta property=\"og:locale\" content=\"en_US\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI\" \/>\n<meta property=\"og:description\" content=\"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability\" \/>\n<meta property=\"og:url\" content=\"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/\" \/>\n<meta property=\"og:site_name\" content=\"Pacific AI\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/people\/Pacific-AI\/61566807347567\/\" \/>\n<meta property=\"article:modified_time\" content=\"2026-02-19T11:14:34+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/pacific.ai\/wp-content\/uploads\/2024\/12\/3-6.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"550\" \/>\n\t<meta property=\"og:image:height\" content=\"440\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"Est. reading time\" \/>\n\t<meta name=\"twitter:data1\" content=\"4 minutes\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/\",\"url\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/\",\"name\":\"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/wp-content\\\/uploads\\\/2024\\\/12\\\/3-6.webp\",\"datePublished\":\"2024-12-01T15:09:08+00:00\",\"dateModified\":\"2026-02-19T11:14:34+00:00\",\"description\":\"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/#breadcrumb\"},\"inLanguage\":\"en\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"en\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/#primaryimage\",\"url\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/wp-content\\\/uploads\\\/2024\\\/12\\\/3-6.webp\",\"contentUrl\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/wp-content\\\/uploads\\\/2024\\\/12\\\/3-6.webp\",\"width\":550,\"height\":440,\"caption\":\"Holistic evaluation of large language models, illustrating an AI profile formed from data streams and neural networks to represent robustness testing, accuracy assessment, and toxicity analysis for real-world LLM applications.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/peer-reviews-paper\\\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/pacific.ai\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#website\",\"url\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/\",\"name\":\"Pacific AI\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"en\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#organization\",\"name\":\"Pacific AI\",\"url\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"en-US\",\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/wp-content\\\/uploads\\\/2025\\\/06\\\/site_logo.svg\",\"contentUrl\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/wp-content\\\/uploads\\\/2025\\\/06\\\/site_logo.svg\",\"width\":182,\"height\":41,\"caption\":\"Pacific AI\"},\"image\":{\"@id\":\"https:\\\/\\\/pacific.ai\\\/staging\\\/3667\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/people\\\/Pacific-AI\\\/61566807347567\\\/\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/pacific-ai\\\/\"]}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI","description":"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability","robots":{"index":"noindex","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"og_locale":"en_US","og_type":"article","og_title":"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI","og_description":"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability","og_url":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/","og_site_name":"Pacific AI","article_publisher":"https:\/\/www.facebook.com\/people\/Pacific-AI\/61566807347567\/","article_modified_time":"2026-02-19T11:14:34+00:00","og_image":[{"width":550,"height":440,"url":"https:\/\/pacific.ai\/wp-content\/uploads\/2024\/12\/3-6.webp","type":"image\/webp"}],"twitter_card":"summary_large_image","twitter_misc":{"Est. reading time":"4 minutes"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/","url":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/","name":"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications - Pacific AI","isPartOf":{"@id":"https:\/\/pacific.ai\/staging\/3667\/#website"},"primaryImageOfPage":{"@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/#primaryimage"},"image":{"@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/#primaryimage"},"thumbnailUrl":"https:\/\/pacific.ai\/staging\/3667\/wp-content\/uploads\/2024\/12\/3-6.webp","datePublished":"2024-12-01T15:09:08+00:00","dateModified":"2026-02-19T11:14:34+00:00","description":"Peer-reviewed study introduces a holistic framework to evaluate LLMs on accuracy, robustness, and toxicity, using LangTest for real-world AI reliability","breadcrumb":{"@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/#breadcrumb"},"inLanguage":"en","potentialAction":[{"@type":"ReadAction","target":["https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/"]}]},{"@type":"ImageObject","inLanguage":"en","@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/#primaryimage","url":"https:\/\/pacific.ai\/staging\/3667\/wp-content\/uploads\/2024\/12\/3-6.webp","contentUrl":"https:\/\/pacific.ai\/staging\/3667\/wp-content\/uploads\/2024\/12\/3-6.webp","width":550,"height":440,"caption":"Holistic evaluation of large language models, illustrating an AI profile formed from data streams and neural networks to represent robustness testing, accuracy assessment, and toxicity analysis for real-world LLM applications."},{"@type":"BreadcrumbList","@id":"https:\/\/pacific.ai\/peer-reviews-paper\/holistic-evaluation-of-large-language-models-assessing-robustness-accuracy-and-toxicity-for-real-world-applications\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/pacific.ai\/"},{"@type":"ListItem","position":2,"name":"Holistic Evaluation of Large Language Models: Assessing Robustness, Accuracy, and Toxicity for Real-World Applications"}]},{"@type":"WebSite","@id":"https:\/\/pacific.ai\/staging\/3667\/#website","url":"https:\/\/pacific.ai\/staging\/3667\/","name":"Pacific AI","description":"","publisher":{"@id":"https:\/\/pacific.ai\/staging\/3667\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/pacific.ai\/staging\/3667\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"en"},{"@type":"Organization","@id":"https:\/\/pacific.ai\/staging\/3667\/#organization","name":"Pacific AI","url":"https:\/\/pacific.ai\/staging\/3667\/","logo":{"@type":"ImageObject","inLanguage":"en-US","@id":"https:\/\/pacific.ai\/staging\/3667\/#\/schema\/logo\/image\/","url":"https:\/\/pacific.ai\/staging\/3667\/wp-content\/uploads\/2025\/06\/site_logo.svg","contentUrl":"https:\/\/pacific.ai\/staging\/3667\/wp-content\/uploads\/2025\/06\/site_logo.svg","width":182,"height":41,"caption":"Pacific AI"},"image":{"@id":"https:\/\/pacific.ai\/staging\/3667\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/people\/Pacific-AI\/61566807347567\/","https:\/\/www.linkedin.com\/company\/pacific-ai\/"]}]}},"_links":{"self":[{"href":"https:\/\/pacific.ai\/staging\/3667\/wp-json\/wp\/v2\/peer-reviews-paper\/134","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/pacific.ai\/staging\/3667\/wp-json\/wp\/v2\/peer-reviews-paper"}],"about":[{"href":"https:\/\/pacific.ai\/staging\/3667\/wp-json\/wp\/v2\/types\/peer-reviews-paper"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/pacific.ai\/staging\/3667\/wp-json\/wp\/v2\/media\/921"}],"wp:attachment":[{"href":"https:\/\/pacific.ai\/staging\/3667\/wp-json\/wp\/v2\/media?parent=134"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}