| <!doctype html> |
| <html lang="en"> |
| <head> |
| <meta charset="utf-8" /> |
| <meta name="viewport" content="width=device-width, initial-scale=1" /> |
| <meta name="color-scheme" content="light dark" /> |
| <title>Lernex</title> |
| <style> |
| * { box-sizing: border-box; } |
| |
| body { |
| margin: 0; |
| background: transparent; |
| color: #d7dde8; |
| font-family: ui-sans-serif, system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif; |
| letter-spacing: 0; |
| } |
| |
| .org-card { |
| width: 100%; |
| max-width: 100%; |
| margin: 0; |
| padding: 0 0 8px; |
| } |
| |
| .hero { |
| display: block; |
| width: 100%; |
| height: auto; |
| margin: 0 0 40px; |
| border-radius: 2px; |
| } |
| |
| p { |
| margin: 0 0 18px; |
| color: #c6cedb; |
| font-size: clamp(18px, 2.1vw, 22px); |
| line-height: 1.55; |
| } |
| |
| .intro { |
| color: #e7ebf2; |
| font-style: italic; |
| } |
| |
| a { |
| color: inherit; |
| font-weight: 700; |
| text-decoration: underline; |
| text-underline-offset: 3px; |
| } |
| |
| h2 { |
| margin: 52px 0 20px; |
| color: #eef2f8; |
| font-size: clamp(24px, 3vw, 30px); |
| line-height: 1.2; |
| font-weight: 800; |
| } |
| |
| ul { |
| margin: 0 0 12px; |
| padding-left: 28px; |
| color: #c6cedb; |
| font-size: clamp(18px, 2.1vw, 22px); |
| line-height: 1.55; |
| } |
| |
| li { |
| margin: 0 0 16px; |
| } |
| |
| strong { |
| color: #eef2f8; |
| font-weight: 800; |
| } |
| |
| .resources { |
| margin-bottom: 0; |
| } |
| |
| @media (prefers-color-scheme: light) { |
| body { color: #202733; } |
| p, ul { color: #475467; } |
| .intro, h2, strong { color: #111827; } |
| } |
| |
| @media (max-width: 700px) { |
| .hero { margin-bottom: 28px; } |
| h2 { margin-top: 36px; } |
| ul { padding-left: 22px; } |
| } |
| </style> |
| </head> |
| <body> |
| <main class="org-card"> |
| <img |
| class="hero" |
| src="https://huggingface.co/spaces/Lernex/README/resolve/main/assets/lernex-org-banner.jpg" |
| alt="Lernex turns learning material into adaptive study paths and compact AI systems" |
| /> |
|
|
| <p class="intro">Welcome to the official Lernex organization on Hugging Face.</p> |
|
|
| <p> |
| <a href="https://www.lernex.net/" target="_blank" rel="noreferrer">Lernex</a> |
| is an AI-personalized learning platform built around a simple idea: |
| studying should adapt to the learner, the material, and the moment. |
| Lernex turns notes, PDFs, slides, classroom material, questions, and |
| learner goals into lessons, quizzes, flashcards, explanations, study |
| paths, feedback, and memory-aware tutoring. |
| </p> |
|
|
| <p> |
| This organization is where Lernex publishes the Metis model line and |
| related open research artifacts. Metis models are compact, efficient |
| language models built for the specific jobs inside a learning product: |
| tutoring, explanations, writing support, practice generation, concise |
| reasoning, and fast product loops that cannot depend on oversized |
| systems for every interaction. |
| </p> |
|
|
| <h2>What Lernex Is Building</h2> |
| <ul> |
| <li><strong>A continuous learning companion</strong>: Lernex is designed so a learner does not restart from zero every time they open a different class, generated lesson, quiz, flashcard set, or study path.</li> |
| <li><strong>Adaptive study material</strong>: source material can become structured lessons, checks for understanding, retrieval-friendly notes, and targeted practice.</li> |
| <li><strong>Personalized learning paths</strong>: Lernex uses learner goals, progress, confusion signals, and prior activity to shape what should come next.</li> |
| <li><strong>Fast learning workflows</strong>: learning tools should feel responsive enough for everyday studying, not like heavyweight one-off AI demos.</li> |
| </ul> |
|
|
| <h2>Why We Build Models</h2> |
| <p> |
| Education products need a different model shape than general chat |
| products. Some proprietary models are powerful but expensive or slow |
| for high-frequency learning loops. Some open models are affordable but |
| miss the tutoring, instruction-following, reasoning, or product-specific |
| behavior Lernex needs. |
| </p> |
| <p> |
| Metis is our answer to that mismatch: build compact models that are |
| useful where learning quality matters, efficient enough to run often, |
| and honest about the tradeoffs between size, reasoning, latency, and |
| cost. |
| </p> |
|
|
| <h2>Featured Models</h2> |
| <ul> |
| <li><strong><a href="https://huggingface.co/Lernex/Metis-1.4-base" target="_blank" rel="noreferrer">Metis-1.4 Base</a></strong> - current base release for the Metis-1.4 line.</li> |
| <li><strong><a href="https://huggingface.co/Lernex/Metis-1.4-chat" target="_blank" rel="noreferrer">Metis-1.4 Chat</a></strong> - conversational variant for tutoring, writing, explanations, and learning support.</li> |
| <li><strong><a href="https://huggingface.co/Lernex/Metis-1.4-think" target="_blank" rel="noreferrer">Metis-1.4 Think</a></strong> - reasoning-oriented variant for compact step-by-step problem solving.</li> |
| <li><strong><a href="https://huggingface.co/Lernex/Metis-1.3-base" target="_blank" rel="noreferrer">Metis-1.3 Base</a></strong>, <strong><a href="https://huggingface.co/Lernex/Metis-1.3-chat" target="_blank" rel="noreferrer">Chat</a></strong>, and <strong><a href="https://huggingface.co/Lernex/Metis-1.3-think" target="_blank" rel="noreferrer">Think</a></strong> - earlier hybrid Mamba-attention research releases.</li> |
| </ul> |
|
|
| <h2>Research Direction</h2> |
| <ul> |
| <li><strong>Compact tutoring models</strong>: smaller models that can still explain clearly, generate useful practice, and behave consistently in learning contexts.</li> |
| <li><strong>Efficient reasoning</strong>: model variants that spend compute where step-by-step problem solving matters instead of making every request expensive.</li> |
| <li><strong>Product-shaped evaluation</strong>: evaluating models on whether they help learners understand, practice, revise, and move forward, not only on broad benchmark prestige.</li> |
| <li><strong>Deployment practicality</strong>: keeping model size, serving cost, latency, and memory footprint in view from the start.</li> |
| </ul> |
|
|
| <h2>How This Connects to Lernex</h2> |
| <p> |
| Lernex uses AI to help learners bring messy material into a system that |
| can teach from it. The long-term goal is not just a better chatbot; it |
| is one learning relationship that carries context across generated |
| lessons, study paths, practice, feedback, and review. |
| </p> |
| <p> |
| The Metis line supports that goal by giving Lernex a model family it can |
| tune around real learning workflows instead of treating every AI job as |
| the same generic chat problem. |
| </p> |
|
|
| <h2>Resources</h2> |
| <ul class="resources"> |
| <li>Website: <a href="https://www.lernex.net/" target="_blank" rel="noreferrer">lernex.net</a></li> |
| <li>Company page: <a href="https://www.lernex.net/company" target="_blank" rel="noreferrer">lernex.net/company</a></li> |
| <li>Product feedback: <a href="mailto:feedback@lernex.net">feedback@lernex.net</a></li> |
| <li>Business contact: <a href="mailto:business@lernex.net">business@lernex.net</a></li> |
| </ul> |
| </main> |
| </body> |
| </html> |
|
|
