Infograf\u00eda: Familia de Modelos Google Gemini<\/title>\n <script src=\"https:\/\/cdn.tailwindcss.com\"><\/script>\n <script src=\"https:\/\/cdn.jsdelivr.net\/npm\/chart.js\"><\/script>\n <link rel=\"preconnect\" href=\"https:\/\/fonts.googleapis.com\">\n <link rel=\"preconnect\" href=\"https:\/\/fonts.gstatic.com\" crossorigin>\n <link href=\"https:\/\/fonts.googleapis.com\/css2?family=Inter:wght@400;600;700;900&display=swap\" rel=\"stylesheet\">\n <style>\n body {\n font-family: 'Inter', sans-serif;\n background-color: #073B4C;\n }\n .chart-container {\n position: relative;\n width: 100%;\n margin-left: auto;\n margin-right: auto;\n max-height: 400px;\n }\n .card {\n background-color: #112a3d;\n transition: transform 0.3s ease, box-shadow 0.3s ease;\n }\n .card:hover {\n transform: translateY(-5px);\n box-shadow: 0 10px 15px -3px rgba(6, 214, 160, 0.1), 0 4px 6px -2px rgba(6, 214, 160, 0.05);\n }\n .timeline-item::before {\n content: '';\n position: absolute;\n left: -13px;\n top: 0;\n bottom: 0;\n width: 2px;\n background-color: #118AB2;\n }\n .timeline-dot {\n position: absolute;\n left: -25px;\n top: 4px;\n height: 24px;\n width: 24px;\n border-radius: 50%;\n background-color: #FFD166;\n border: 4px solid #073B4C;\n }\n <\/style>\n<\/head>\n<body class=\"text-white\">\n \n <div class=\"w-full p-4 sm:p-8\">\n\n <header class=\"text-center mb-12\">\n <h1 class=\"text-4xl md:text-6xl font-black tracking-tighter mb-2\">FAMILIA DE MODELOS GEMINI<\/h1>\n <p class=\"text-lg md:text-xl text-gray-300 max-w-3xl mx-auto\">Una mirada profunda a la arquitectura, rendimiento y capacidades de la IA multimodal de Google.<\/p>\n <\/header>\n\n \n <main class=\"grid grid-cols-1 gap-8\">\n \n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Descripci\u00f3n General: El Cerebro Multimodal Nativo<\/h2>\n <p class=\"text-gray-300 mb-6\">A diferencia de modelos que unen componentes, Gemini fue dise\u00f1ado desde cero para comprender y razonar fluidamente a trav\u00e9s de diversas modalidades de informaci\u00f3n de forma simult\u00e1nea. Es una \u00fanica IA cohesiva, no una colecci\u00f3n de partes.<\/p>\n <div class=\"grid grid-cols-2 sm:grid-cols-5 gap-4 text-center max-w-4xl mx-auto\">\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\udcdd<\/span>\n <p class=\"mt-2 font-semibold\">Texto<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\uddbc\ufe0f<\/span>\n <p class=\"mt-2 font-semibold\">Im\u00e1genes<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\udd0a<\/span>\n <p class=\"mt-2 font-semibold\">Audio<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83c\udfac<\/span>\n <p class=\"mt-2 font-semibold\">Video<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg col-span-2 sm:col-span-1\">\n <span class=\"text-4xl\">\ud83d\udcbb<\/span>\n <p class=\"mt-2 font-semibold\">C\u00f3digo<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Hitos Clave<\/h2>\n <p class=\"text-gray-300 mb-6\">La evoluci\u00f3n de Gemini ha sido r\u00e1pida, introduciendo mejoras significativas en arquitectura y capacidad en un corto per\u00edodo.<\/p>\n <div class=\"relative pl-8 max-w-lg mx-auto\">\n <div class=\"relative mb-8 timeline-item\">\n <div class=\"timeline-dot\"><\/div>\n <p class=\"font-bold text-lg text-gray-200\">Diciembre 2023<\/p>\n <p class=\"text-gray-400\">Lanzamiento de Gemini 1.0 (Pro, Ultra, Nano).<\/p>\n <\/div>\n <div class=\"relative timeline-item\">\n <div class=\"timeline-dot\"><\/div>\n <p class=\"font-bold text-lg text-gray-200\">Febrero 2024<\/p>\n <p class=\"text-gray-400\">Lanzamiento de Gemini 1.5 Pro con arquitectura MoE.<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Rendimiento de Vanguardia: Superando L\u00edmites<\/h2>\n <p class=\"text-gray-300 mb-4\">Gemini 1.0 Ultra estableci\u00f3 un nuevo est\u00e1ndar en la comprensi\u00f3n masiva de lenguajes multitarea (MMLU), una m\u00e9trica clave que eval\u00faa el conocimiento y la capacidad de resoluci\u00f3n de problemas.<\/p>\n <div class=\"grid grid-cols-1 sm:grid-cols-2 gap-6 items-center max-w-3xl mx-auto\">\n <div>\n <div class=\"chart-container h-64 sm:h-auto\">\n <canvas id=\"mmluChart\"><\/canvas>\n <\/div>\n <\/div>\n <div class=\"text-center sm:text-left\">\n <p class=\"text-7xl font-black text-[#06D6A0]\">90.0%<\/p>\n <p class=\"text-lg font-bold\">Puntuaci\u00f3n en MMLU<\/p>\n <p class=\"text-gray-400\">Primer modelo en superar el rendimiento a nivel de experto humano.<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">El Salto Cu\u00e1ntico en la Ventana de Contexto<\/h2>\n <p class=\"text-gray-300 mb-6\">La ventana de contexto define cu\u00e1nta informaci\u00f3n puede procesar un modelo en una sola consulta. Gemini 1.5 Pro, con su arquitectura de Mezcla de Expertos (MoE), representa un avance monumental, permitiendo el an\u00e1lisis de bases de c\u00f3digo completas, libros enteros o largas grabaciones de video de una sola vez.<\/p>\n <div class=\"chart-container h-96 max-w-4xl mx-auto\">\n <canvas id=\"contextWindowChart\"><\/canvas>\n <\/div>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Dominio en Benchmarks<\/h2>\n <p class=\"text-gray-300 mb-6\">El modelo ha demostrado un rendimiento de vanguardia (State-of-the-Art) en la gran mayor\u00eda de los benchmarks acad\u00e9micos m\u00e1s utilizados para evaluar LLMs. De 32 pruebas clave, es el l\u00edder en 30.<\/p>\n <div class=\"chart-container h-72 w-full max-w-lg mx-auto\">\n <canvas id=\"benchmarkChart\"><\/canvas>\n <\/div>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Arquitectura y Eficiencia<\/h2>\n <ul class=\"space-y-4 text-gray-300\">\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\u2699\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Base Transformer:<\/span> Optimizada para la m\u00e1xima escalabilidad y eficiencia.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\u26a1\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Infraestructura TPU:<\/span> Co-dise\u00f1ado para ejecutarse en los Tensor Processing Units (TPUs) de Google, logrando mayor velocidad y menor coste.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83c\udf0d<\/span>\n <div><span class=\"font-semibold text-white\">Huella de Carbono Reducida:<\/span> Entrenado en centros de datos que operan con un alto porcentaje de energ\u00eda libre de carbono.<\/div>\n <\/li>\n <\/ul>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Seguridad y \u00c9tica<\/h2>\n <ul class=\"space-y-4 text-gray-300\">\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udee1\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Evaluaciones Exhaustivas:<\/span> Pruebas adversarias (\"red teaming\") para identificar y mitigar riesgos de sesgos y toxicidad.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udeab<\/span>\n <div><span class=\"font-semibold text-white\">Clasificadores de Seguridad:<\/span> Filtros activos para prevenir la generaci\u00f3n de contenido que viole las pol\u00edticas de uso.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udca7<\/span>\n <div><span class=\"font-semibold text-white\">SynthID Watermarking:<\/span> Incrusta una marca de agua digital imperceptible en im\u00e1genes generadas para identificarlas como creadas por IA.<\/div>\n <\/li>\n <\/ul>\n <\/section>\n <\/main>\n \n <footer class=\"text-center mt-12 text-gray-500 text-sm\">\n <p>Infograf\u00eda generada a partir de la ficha t\u00e9cnica del modelo Google Gemini. Visualizaciones creadas con Chart.js y Tailwind CSS.<\/p>\n <\/footer>\n\n <\/div>\n\n <script>\n document.addEventListener('DOMContentLoaded', () => {\n const FONT_COLOR = '#E5E7EB';\n const GRID_COLOR = 'rgba(255, 255, 255, 0.1)';\n\n function wrapLabel(label) {\n const max_width = 16;\n if (label.length <= max_width) {\n return label;\n }\n const words = label.split(' ');\n let lines = [];\n let current_line = '';\n for (const word of words) {\n if ((current_line + ' ' + word).length > max_width) {\n lines.push(current_line);\n current_line = word;\n } else {\n current_line += (current_line.length === 0 ? '' : ' ') + word;\n }\n }\n lines.push(current_line);\n return lines;\n }\n \n const tooltipTitleCallback = (tooltipItems) => {\n const item = tooltipItems[0];\n let label = item.chart.data.labels[item.dataIndex];\n if (Array.isArray(label)) {\n return label.join(' ');\n } else {\n return label;\n }\n };\n \n const mmluCtx = document.getElementById('mmluChart');\n if(mmluCtx) {\n new Chart(mmluCtx, {\n type: 'doughnut',\n data: {\n labels: ['Puntuaci\u00f3n Gemini Ultra', 'Restante para 100%'],\n datasets: [{\n label: 'Puntuaci\u00f3n MMLU',\n data: [90, 10],\n backgroundColor: ['#06D6A0', '#118AB2'],\n borderColor: '#112a3d',\n borderWidth: 4,\n hoverOffset: 8\n }]\n },\n options: {\n responsive: true,\n maintainAspectRatio: false,\n cutout: '70%',\n plugins: {\n legend: { display: false },\n tooltip: {\n callbacks: { title: tooltipTitleCallback }\n }\n }\n }\n });\n }\n\n const contextWindowCtx = document.getElementById('contextWindowChart');\n if(contextWindowCtx) {\n new Chart(contextWindowCtx, {\n type: 'bar',\n data: {\n labels: ['Gemini 1.0 Pro', 'Gemini 1.5 Pro (Est\u00e1ndar)', ['Gemini 1.5 Pro', '(Experimental)']],\n datasets: [{\n label: 'Tokens de Contexto',\n data: [32000, 128000, 1000000],\n backgroundColor: ['#FF6B6B', '#FFD166', '#06D6A0'],\n borderColor: ['#FF6B6B', '#FFD166', '#06D6A0'],\n borderWidth: 1,\n borderRadius: 5\n }]\n },\n options: {\n indexAxis: 'y',\n responsive: true,\n maintainAspectRatio: false,\n scales: {\n x: {\n type: 'logarithmic',\n ticks: {\n color: FONT_COLOR,\n callback: function(value, index, values) {\n if (value === 1000 || value === 10000 || value === 100000 || value === 1000000) {\n return value.toLocaleString() + ' tokens';\n }\n }\n },\n grid: { color: GRID_COLOR }\n },\n y: {\n beginAtZero: true,\n ticks: { color: FONT_COLOR, font: { size: 14 } },\n grid: { display: false }\n }\n },\n plugins: {\n legend: { display: false },\n tooltip: {\n callbacks: {\n title: tooltipTitleCallback,\n label: function(context) {\n return ` ${context.raw.toLocaleString()} tokens`;\n }\n }\n }\n }\n }\n });\n }\n\n const benchmarkCtx = document.getElementById('benchmarkChart');\n if(benchmarkCtx) {\n new Chart(benchmarkCtx, {\n type: 'bar',\n data: {\n labels: ['Rendimiento SOTA', 'Otros Benchmarks'],\n datasets: [{\n label: 'N\u00famero de Benchmarks',\n data: [30, 2],\n backgroundColor: ['#118AB2', '#FF6B6B'],\n borderRadius: 5\n }]\n },\n options: {\n responsive: true,\n maintainAspectRatio: false,\n scales: {\n y: {\n beginAtZero: true,\n max: 32,\n ticks: { color: FONT_COLOR },\n grid: { color: GRID_COLOR }\n },\n x: {\n ticks: { color: FONT_COLOR },\n grid: { display: false }\n }\n },\n plugins: {\n legend: { display: false },\n tooltip: { callbacks: { title: tooltipTitleCallback } }\n }\n }\n });\n }\n });\n <\/script>\n<\/body>\n<\/html><\/div><\/div><\/div><\/div>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":2461,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"image","meta":{"_acf_changed":false,"footnotes":""},"categories":[41],"tags":[42,43,44],"class_list":["post-2159","post","type-post","status-publish","format-image","has-post-thumbnail","hentry","category-modelos-ia","tag-gemini","tag-inteligencia-artificial","tag-modelos-ia","post_format-post-format-image"],"acf":[],"_links":{"self":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/comments?post=2159"}],"version-history":[{"count":6,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159\/revisions"}],"predecessor-version":[{"id":2571,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159\/revisions\/2571"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/media\/2461"}],"wp:attachment":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/media?parent=2159"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/categories?post=2159"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/tags?post=2159"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}

\n\n\n \n \n Infograf\u00eda: Familia de Modelos Google Gemini<\/title>\n <script src=\"https:\/\/cdn.tailwindcss.com\"><\/script>\n <script src=\"https:\/\/cdn.jsdelivr.net\/npm\/chart.js\"><\/script>\n <link rel=\"preconnect\" href=\"https:\/\/fonts.googleapis.com\">\n <link rel=\"preconnect\" href=\"https:\/\/fonts.gstatic.com\" crossorigin>\n <link href=\"https:\/\/fonts.googleapis.com\/css2?family=Inter:wght@400;600;700;900&display=swap\" rel=\"stylesheet\">\n <style>\n body {\n font-family: 'Inter', sans-serif;\n background-color: #073B4C;\n }\n .chart-container {\n position: relative;\n width: 100%;\n margin-left: auto;\n margin-right: auto;\n max-height: 400px;\n }\n .card {\n background-color: #112a3d;\n transition: transform 0.3s ease, box-shadow 0.3s ease;\n }\n .card:hover {\n transform: translateY(-5px);\n box-shadow: 0 10px 15px -3px rgba(6, 214, 160, 0.1), 0 4px 6px -2px rgba(6, 214, 160, 0.05);\n }\n .timeline-item::before {\n content: '';\n position: absolute;\n left: -13px;\n top: 0;\n bottom: 0;\n width: 2px;\n background-color: #118AB2;\n }\n .timeline-dot {\n position: absolute;\n left: -25px;\n top: 4px;\n height: 24px;\n width: 24px;\n border-radius: 50%;\n background-color: #FFD166;\n border: 4px solid #073B4C;\n }\n <\/style>\n<\/head>\n<body class=\"text-white\">\n \n <div class=\"w-full p-4 sm:p-8\">\n\n <header class=\"text-center mb-12\">\n <h1 class=\"text-4xl md:text-6xl font-black tracking-tighter mb-2\">FAMILIA DE MODELOS GEMINI<\/h1>\n <p class=\"text-lg md:text-xl text-gray-300 max-w-3xl mx-auto\">Una mirada profunda a la arquitectura, rendimiento y capacidades de la IA multimodal de Google.<\/p>\n <\/header>\n\n \n <main class=\"grid grid-cols-1 gap-8\">\n \n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Descripci\u00f3n General: El Cerebro Multimodal Nativo<\/h2>\n <p class=\"text-gray-300 mb-6\">A diferencia de modelos que unen componentes, Gemini fue dise\u00f1ado desde cero para comprender y razonar fluidamente a trav\u00e9s de diversas modalidades de informaci\u00f3n de forma simult\u00e1nea. Es una \u00fanica IA cohesiva, no una colecci\u00f3n de partes.<\/p>\n <div class=\"grid grid-cols-2 sm:grid-cols-5 gap-4 text-center max-w-4xl mx-auto\">\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\udcdd<\/span>\n <p class=\"mt-2 font-semibold\">Texto<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\uddbc\ufe0f<\/span>\n <p class=\"mt-2 font-semibold\">Im\u00e1genes<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83d\udd0a<\/span>\n <p class=\"mt-2 font-semibold\">Audio<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg\">\n <span class=\"text-4xl\">\ud83c\udfac<\/span>\n <p class=\"mt-2 font-semibold\">Video<\/p>\n <\/div>\n <div class=\"bg-[#073B4C] p-4 rounded-lg col-span-2 sm:col-span-1\">\n <span class=\"text-4xl\">\ud83d\udcbb<\/span>\n <p class=\"mt-2 font-semibold\">C\u00f3digo<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Hitos Clave<\/h2>\n <p class=\"text-gray-300 mb-6\">La evoluci\u00f3n de Gemini ha sido r\u00e1pida, introduciendo mejoras significativas en arquitectura y capacidad en un corto per\u00edodo.<\/p>\n <div class=\"relative pl-8 max-w-lg mx-auto\">\n <div class=\"relative mb-8 timeline-item\">\n <div class=\"timeline-dot\"><\/div>\n <p class=\"font-bold text-lg text-gray-200\">Diciembre 2023<\/p>\n <p class=\"text-gray-400\">Lanzamiento de Gemini 1.0 (Pro, Ultra, Nano).<\/p>\n <\/div>\n <div class=\"relative timeline-item\">\n <div class=\"timeline-dot\"><\/div>\n <p class=\"font-bold text-lg text-gray-200\">Febrero 2024<\/p>\n <p class=\"text-gray-400\">Lanzamiento de Gemini 1.5 Pro con arquitectura MoE.<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Rendimiento de Vanguardia: Superando L\u00edmites<\/h2>\n <p class=\"text-gray-300 mb-4\">Gemini 1.0 Ultra estableci\u00f3 un nuevo est\u00e1ndar en la comprensi\u00f3n masiva de lenguajes multitarea (MMLU), una m\u00e9trica clave que eval\u00faa el conocimiento y la capacidad de resoluci\u00f3n de problemas.<\/p>\n <div class=\"grid grid-cols-1 sm:grid-cols-2 gap-6 items-center max-w-3xl mx-auto\">\n <div>\n <div class=\"chart-container h-64 sm:h-auto\">\n <canvas id=\"mmluChart\"><\/canvas>\n <\/div>\n <\/div>\n <div class=\"text-center sm:text-left\">\n <p class=\"text-7xl font-black text-[#06D6A0]\">90.0%<\/p>\n <p class=\"text-lg font-bold\">Puntuaci\u00f3n en MMLU<\/p>\n <p class=\"text-gray-400\">Primer modelo en superar el rendimiento a nivel de experto humano.<\/p>\n <\/div>\n <\/div>\n <\/section>\n\n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">El Salto Cu\u00e1ntico en la Ventana de Contexto<\/h2>\n <p class=\"text-gray-300 mb-6\">La ventana de contexto define cu\u00e1nta informaci\u00f3n puede procesar un modelo en una sola consulta. Gemini 1.5 Pro, con su arquitectura de Mezcla de Expertos (MoE), representa un avance monumental, permitiendo el an\u00e1lisis de bases de c\u00f3digo completas, libros enteros o largas grabaciones de video de una sola vez.<\/p>\n <div class=\"chart-container h-96 max-w-4xl mx-auto\">\n <canvas id=\"contextWindowChart\"><\/canvas>\n <\/div>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Dominio en Benchmarks<\/h2>\n <p class=\"text-gray-300 mb-6\">El modelo ha demostrado un rendimiento de vanguardia (State-of-the-Art) en la gran mayor\u00eda de los benchmarks acad\u00e9micos m\u00e1s utilizados para evaluar LLMs. De 32 pruebas clave, es el l\u00edder en 30.<\/p>\n <div class=\"chart-container h-72 w-full max-w-lg mx-auto\">\n <canvas id=\"benchmarkChart\"><\/canvas>\n <\/div>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Arquitectura y Eficiencia<\/h2>\n <ul class=\"space-y-4 text-gray-300\">\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\u2699\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Base Transformer:<\/span> Optimizada para la m\u00e1xima escalabilidad y eficiencia.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\u26a1\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Infraestructura TPU:<\/span> Co-dise\u00f1ado para ejecutarse en los Tensor Processing Units (TPUs) de Google, logrando mayor velocidad y menor coste.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83c\udf0d<\/span>\n <div><span class=\"font-semibold text-white\">Huella de Carbono Reducida:<\/span> Entrenado en centros de datos que operan con un alto porcentaje de energ\u00eda libre de carbono.<\/div>\n <\/li>\n <\/ul>\n <\/section>\n \n <section class=\"card rounded-xl shadow-2xl p-6\">\n <h2 class=\"text-2xl font-bold mb-4 text-[#FFD166]\">Seguridad y \u00c9tica<\/h2>\n <ul class=\"space-y-4 text-gray-300\">\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udee1\ufe0f<\/span>\n <div><span class=\"font-semibold text-white\">Evaluaciones Exhaustivas:<\/span> Pruebas adversarias (\"red teaming\") para identificar y mitigar riesgos de sesgos y toxicidad.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udeab<\/span>\n <div><span class=\"font-semibold text-white\">Clasificadores de Seguridad:<\/span> Filtros activos para prevenir la generaci\u00f3n de contenido que viole las pol\u00edticas de uso.<\/div>\n <\/li>\n <li class=\"flex items-start\">\n <span class=\"text-[#118AB2] mr-3 mt-1 text-xl\">\ud83d\udca7<\/span>\n <div><span class=\"font-semibold text-white\">SynthID Watermarking:<\/span> Incrusta una marca de agua digital imperceptible en im\u00e1genes generadas para identificarlas como creadas por IA.<\/div>\n <\/li>\n <\/ul>\n <\/section>\n <\/main>\n \n <footer class=\"text-center mt-12 text-gray-500 text-sm\">\n <p>Infograf\u00eda generada a partir de la ficha t\u00e9cnica del modelo Google Gemini. Visualizaciones creadas con Chart.js y Tailwind CSS.<\/p>\n <\/footer>\n\n <\/div>\n\n <script>\n document.addEventListener('DOMContentLoaded', () => {\n const FONT_COLOR = '#E5E7EB';\n const GRID_COLOR = 'rgba(255, 255, 255, 0.1)';\n\n function wrapLabel(label) {\n const max_width = 16;\n if (label.length <= max_width) {\n return label;\n }\n const words = label.split(' ');\n let lines = [];\n let current_line = '';\n for (const word of words) {\n if ((current_line + ' ' + word).length > max_width) {\n lines.push(current_line);\n current_line = word;\n } else {\n current_line += (current_line.length === 0 ? '' : ' ') + word;\n }\n }\n lines.push(current_line);\n return lines;\n }\n \n const tooltipTitleCallback = (tooltipItems) => {\n const item = tooltipItems[0];\n let label = item.chart.data.labels[item.dataIndex];\n if (Array.isArray(label)) {\n return label.join(' ');\n } else {\n return label;\n }\n };\n \n const mmluCtx = document.getElementById('mmluChart');\n if(mmluCtx) {\n new Chart(mmluCtx, {\n type: 'doughnut',\n data: {\n labels: ['Puntuaci\u00f3n Gemini Ultra', 'Restante para 100%'],\n datasets: [{\n label: 'Puntuaci\u00f3n MMLU',\n data: [90, 10],\n backgroundColor: ['#06D6A0', '#118AB2'],\n borderColor: '#112a3d',\n borderWidth: 4,\n hoverOffset: 8\n }]\n },\n options: {\n responsive: true,\n maintainAspectRatio: false,\n cutout: '70%',\n plugins: {\n legend: { display: false },\n tooltip: {\n callbacks: { title: tooltipTitleCallback }\n }\n }\n }\n });\n }\n\n const contextWindowCtx = document.getElementById('contextWindowChart');\n if(contextWindowCtx) {\n new Chart(contextWindowCtx, {\n type: 'bar',\n data: {\n labels: ['Gemini 1.0 Pro', 'Gemini 1.5 Pro (Est\u00e1ndar)', ['Gemini 1.5 Pro', '(Experimental)']],\n datasets: [{\n label: 'Tokens de Contexto',\n data: [32000, 128000, 1000000],\n backgroundColor: ['#FF6B6B', '#FFD166', '#06D6A0'],\n borderColor: ['#FF6B6B', '#FFD166', '#06D6A0'],\n borderWidth: 1,\n borderRadius: 5\n }]\n },\n options: {\n indexAxis: 'y',\n responsive: true,\n maintainAspectRatio: false,\n scales: {\n x: {\n type: 'logarithmic',\n ticks: {\n color: FONT_COLOR,\n callback: function(value, index, values) {\n if (value === 1000 || value === 10000 || value === 100000 || value === 1000000) {\n return value.toLocaleString() + ' tokens';\n }\n }\n },\n grid: { color: GRID_COLOR }\n },\n y: {\n beginAtZero: true,\n ticks: { color: FONT_COLOR, font: { size: 14 } },\n grid: { display: false }\n }\n },\n plugins: {\n legend: { display: false },\n tooltip: {\n callbacks: {\n title: tooltipTitleCallback,\n label: function(context) {\n return ` ${context.raw.toLocaleString()} tokens`;\n }\n }\n }\n }\n }\n });\n }\n\n const benchmarkCtx = document.getElementById('benchmarkChart');\n if(benchmarkCtx) {\n new Chart(benchmarkCtx, {\n type: 'bar',\n data: {\n labels: ['Rendimiento SOTA', 'Otros Benchmarks'],\n datasets: [{\n label: 'N\u00famero de Benchmarks',\n data: [30, 2],\n backgroundColor: ['#118AB2', '#FF6B6B'],\n borderRadius: 5\n }]\n },\n options: {\n responsive: true,\n maintainAspectRatio: false,\n scales: {\n y: {\n beginAtZero: true,\n max: 32,\n ticks: { color: FONT_COLOR },\n grid: { color: GRID_COLOR }\n },\n x: {\n ticks: { color: FONT_COLOR },\n grid: { display: false }\n }\n },\n plugins: {\n legend: { display: false },\n tooltip: { callbacks: { title: tooltipTitleCallback } }\n }\n }\n });\n }\n });\n <\/script>\n<\/body>\n<\/html><\/div><\/div><\/div><\/div>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":2461,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"image","meta":{"_acf_changed":false,"footnotes":""},"categories":[41],"tags":[42,43,44],"class_list":["post-2159","post","type-post","status-publish","format-image","has-post-thumbnail","hentry","category-modelos-ia","tag-gemini","tag-inteligencia-artificial","tag-modelos-ia","post_format-post-format-image"],"acf":[],"_links":{"self":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/comments?post=2159"}],"version-history":[{"count":6,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159\/revisions"}],"predecessor-version":[{"id":2571,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/posts\/2159\/revisions\/2571"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/media\/2461"}],"wp:attachment":[{"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/media?parent=2159"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/categories?post=2159"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/edukia.org\/en\/wp-json\/wp\/v2\/tags?post=2159"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}