Polish OpenSAE public onboarding, readability and tutorial navigation

Source-verified updates to existing public Space files; no experiment data or papers changed.

README.md

@@ -1,39 +1,39 @@
----
-title: OpenSAE
-emoji: 🧱
-colorFrom: gray
-colorTo: blue
-sdk: static
-app_file: index.html
-short_description: Research infrastructure for understanding neural networks.
----
-
-<div align="center">
-  <img src="https://raw.githubusercontent.com/OpenSAE/.github/main/opensae_banner.png" alt="OpenSAE" width="100%" />
-  <h2>Open foundations for understanding, evaluating, and intervening in neural networks.</h2>
-  <p>OpenSAE develops open research methods and infrastructure for studying neural representations, SAE features, and their effects on model behavior.</p>
-</div>
-
----
-
-### Research infrastructure
-
-**[🧱 SAEBase — Open the interactive research lab](https://huggingface.co/spaces/OpenSAE/SAEBase)**  
-Capture reusable activations, intervene on SAE features, measure downstream effects, and evaluate the reliability of experiments. The published software includes 18 model–SAE configurations and 48 verified model/layer sites across models from 270M to 9B parameters.  
-[GitHub](https://github.com/OpenSAE/SAEBase) · [v0.3.0 Release](https://github.com/OpenSAE/SAEBase/releases/tag/v0.3.0) · [Evaluation Data](https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation)
-
-### Research projects
-
-| Project | Research question | Explore |
-| --- | --- | --- |
-| **SAR — Behavioral Compositionality** | How do SAE features interact to shape model behavior? | [Research and tutorial](https://github.com/OpenSAE/SAE-Behavioral-Compositionality) |
-| **RSS — Representation Redistribution** | How does feature activity change across prompt presentations? | [Research and tutorial](https://github.com/OpenSAE/Redistribution-Sensitivity-Score) |
-| **SAE Input Localization** | Which controlled inputs are required to explain SAE features? | [Research and tutorial](https://github.com/OpenSAE/SAE-Input-Localization) |
-
-### Open science
-
-We share software, evidence, reproducible evaluation procedures and tutorials. Original pretrained weights remain with their publishers. Our [benchmark evidence](https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation) is freely browsable, with links to the original run records.
-
-**Explore:** [OpenSAE GitHub](https://github.com/OpenSAE) · [SAEBase](https://huggingface.co/spaces/OpenSAE/SAEBase) · [Contribute](https://github.com/OpenSAE/SAEBase/blob/main/CONTRIBUTING.md)  
-**Contact:** hma232@wisc.edu
-
+---
+title: OpenSAE
+emoji: 🧱
+colorFrom: gray
+colorTo: blue
+sdk: static
+app_file: index.html
+short_description: Open infrastructure for mechanistic interpretability.
+---
+
+<div align="center">
+  <img src="https://huggingface.co/spaces/OpenSAE/README/resolve/main/assets/opensae_banner.png" alt="OpenSAE" width="450" />
+</div>
+
+# From learned features to testable explanations
+
+OpenSAE develops open research methods and infrastructure for studying how neural networks represent information and how their internal components shape model behavior.
+
+## Research infrastructure
+
+**[🧱 SAEBase Research Lab](https://huggingface.co/spaces/OpenSAE/SAEBase)** brings together reusable activation capture, SAE feature interventions, benchmarking, and reliability analysis.
+
+- [Interactive 5-minute walkthrough](https://huggingface.co/spaces/OpenSAE/SAEBase)
+- [SAEBase v0.3.0 on GitHub](https://github.com/OpenSAE/SAEBase/releases/tag/v0.3.0)
+- [SAEBase Evaluation Dataset](https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation) — 18 model–SAE configurations and 48 verified model/layer sites, linked to original reports
+
+## Research methods
+
+| Project | Focus | Explore |
+| --- | --- | --- |
+| **SAR** | Interactions among SAE features and their behavioral effects | [Project & tutorial](https://github.com/OpenSAE/SAE-Behavioral-Compositionality) |
+| **RSS** | Representation redistribution across prompt transformations | [Project & tutorial](https://github.com/OpenSAE/Redistribution-Sensitivity-Score) |
+| **Input Localization** | Input dependencies of SAE feature activations | [Project & tutorial](https://github.com/OpenSAE/SAE-Input-Localization) |
+
+## Open science
+
+We share research code, protocols, tutorials, and traceable evaluation records. External pretrained model and SAE weights remain with their respective publishers.
+
+[GitHub organization](https://github.com/OpenSAE) · [Contribute](https://github.com/OpenSAE/SAEBase/blob/main/CONTRIBUTING.md) · Contact: hma232@wisc.edu

index.html

@@ -1 +1,68 @@
-<!doctype html><html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width,initial-scale=1"><title>OpenSAE — Open Research Infrastructure</title><style>body{margin:0;background:#f5f9fa;color:#18313c;font:16px/1.65 system-ui,sans-serif}main{max-width:920px;margin:4vh auto;padding:25px}.banner{width:100%;height:auto;border-radius:16px;display:block;background:#fff;object-fit:contain}h1{font-size:44px;letter-spacing:-.06em;margin:22px 0 5px}p{color:#5e7883}.cards{display:grid;grid-template-columns:repeat(2,1fr);gap:12px;margin:23px 0}.card{padding:22px;border-radius:13px;background:white;border:1px solid #dfebee;color:#143342;text-decoration:none}.card strong{display:block;margin-bottom:6px}.card span{font-size:13px;color:#5d7b84}a{color:#097e79}@media(max-width:650px){.cards{grid-template-columns:1fr}h1{font-size:31px}}</style></head><body><main><img class="banner" src="https://huggingface.co/spaces/OpenSAE/README/resolve/main/assets/opensae_banner.png" alt="OpenSAE"><h1>OpenSAE</h1><p>Open foundations for understanding, evaluating, and intervening in neural networks.</p><div class="cards"><a class="card" href="https://huggingface.co/spaces/OpenSAE/SAEBase" target="_blank"><strong>🧱 SAEBase Research Lab ↗</strong><span>Explore original checkpoint and feature-intervention evidence.</span></a><a class="card" href="https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation" target="_blank"><strong>▤ Evaluation Dataset ↗</strong><span>Browse 48 verified layer sites and their measurements.</span></a><a class="card" href="https://github.com/OpenSAE/SAE-Behavioral-Compositionality" target="_blank"><strong>🧩 SAR</strong><span>Behavioral interactions of SAE features.</span></a><a class="card" href="https://github.com/OpenSAE/Redistribution-Sensitivity-Score" target="_blank"><strong>🔀 RSS</strong><span>Representation redistribution across prompt changes.</span></a></div><p><a href="https://github.com/OpenSAE">GitHub organization ↗</a> · <a href="mailto:hma232@wisc.edu">Contact</a></p></main></body></html>
+<!doctype html>
+<html lang="en">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <meta name="theme-color" content="#f5f9fa">
+  <meta name="description" content="OpenSAE builds open research infrastructure for mechanistic interpretability, feature interventions, and reliable evaluation.">
+  <title>OpenSAE · Open Research Infrastructure</title>
+  <style>
+    :root { --ink: #17313d; --muted: #48616d; --line: #dce8e9; --teal: #087f78; --light: #eef8f6; }
+    * { box-sizing: border-box; }
+    body { margin: 0; background: #f6fafb; color: var(--ink); font: 15px/1.55 Inter, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif; }
+    a { color: var(--teal); }
+    a:focus-visible { outline: 3px solid #169888; outline-offset: 3px; border-radius: 6px; }
+    main { max-width: 880px; margin: 0 auto; padding: 12px 26px 28px; }
+    .brand { text-align: center; padding: 3px 0 8px; }
+    .brand img { width: min(360px, 80%); height: auto; display: block; margin: auto; object-fit: contain; }
+    .eyebrow { font-size: 10px; color: #18786f; letter-spacing: .16em; font-weight: 800; text-transform: uppercase; margin: 11px 0 7px; }
+    h1 { font-size: clamp(22px, 3.8vw, 31px); font-weight: 780; letter-spacing: -.035em; line-height: 1.16; margin: 0 0 9px; }
+    .intro { color: var(--muted); max-width: 690px; margin: 0 0 20px; font-size: 14px; }
+    .links { display: flex; gap: 10px; flex-wrap: wrap; margin-bottom: 20px; }
+    .action { font-size: 12px; font-weight: 730; text-decoration: none; border: 1px solid #bbded7; padding: 10px 14px; border-radius: 9px; background: white; color: #176159; }
+    .action.primary { background: #0b504d; border-color: #0b504d; color: #f1fffb; }
+    .action:hover { box-shadow: 0 4px 16px #0d47451a; }
+    h2 { font-size: 14px; letter-spacing: -.015em; margin: 18px 0 9px; }
+    .cards { display: grid; grid-template-columns: repeat(2, minmax(0, 1fr)); gap: 11px; }
+    .card { background: white; border: 1px solid var(--line); border-radius: 12px; padding: 16px; display: block; text-decoration: none; color: var(--ink); transition: border-color .15s, transform .15s; }
+    .card:hover { border-color: #5caca2; transform: translateY(-1px); }
+    .card strong { display: block; font-size: 14px; margin-bottom: 5px; }
+    .card span { color: var(--muted); display: block; font-size: 12px; line-height: 1.5; }
+    .card.featured { border-color: #b8dcd5; background: linear-gradient(135deg, #f2fbf9, white); }
+    .small-grid { display: grid; grid-template-columns: repeat(3, minmax(0, 1fr)); gap: 11px; }
+    .small-grid .card strong { font-size: 13px; }
+    .small-grid .card span { font-size: 11px; }
+    footer { border-top: 1px solid var(--line); padding-top: 13px; margin-top: 23px; color: #617984; font-size: 12px; }
+    footer a { text-decoration: none; font-weight: 700; }
+    @media(max-width:620px) { main { padding: 12px 15px 25px; } .brand img { width: min(310px, 85%); } .cards, .small-grid { grid-template-columns: 1fr; } }
+    @media(prefers-reduced-motion: reduce) { .card { transition: none; } .card:hover { transform: none; } }
+  </style>
+</head>
+<body>
+  <main>
+    <div class="brand">
+      <img src="https://huggingface.co/spaces/OpenSAE/README/resolve/main/assets/opensae_banner.png" alt="OpenSAE official logo" loading="eager">
+    </div>
+    <div class="eyebrow">Open science · Mechanistic interpretability</div>
+    <h1>From learned features to testable explanations.</h1>
+    <p class="intro">OpenSAE builds open research methods and infrastructure to study neural representations, intervene on model behavior, and evaluate how reliably our measurements explain it.</p>
+    <div class="links">
+      <a class="action primary" href="https://huggingface.co/spaces/OpenSAE/SAEBase">Explore SAEBase ↗</a>
+      <a class="action" href="https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation">Browse evaluation data ↗</a>
+      <a class="action" href="https://github.com/OpenSAE">OpenSAE on GitHub ↗</a>
+    </div>
+    <h2>Research infrastructure</h2>
+    <div class="cards">
+      <a class="card featured" href="https://huggingface.co/spaces/OpenSAE/SAEBase"><strong>🧱 SAEBase Research Lab ↗</strong><span>Capture activations, intervene on SAE features, benchmark effects, and audit reliability.</span></a>
+      <a class="card" href="https://huggingface.co/datasets/OpenSAE/SAEBase-Evaluation"><strong>▤ SAEBase Evaluation ↗</strong><span>Explore 18 model–SAE configurations and 48 verified model/layer sites with traceable reports.</span></a>
+    </div>
+    <h2>Research methods and tutorials</h2>
+    <div class="small-grid">
+      <a class="card" href="https://github.com/OpenSAE/SAE-Behavioral-Compositionality"><strong>🧩 SAR ↗</strong><span>How SAE features interact to shape downstream model behavior.</span></a>
+      <a class="card" href="https://github.com/OpenSAE/Redistribution-Sensitivity-Score"><strong>🔀 RSS ↗</strong><span>How feature activity changes across prompt presentations.</span></a>
+      <a class="card" href="https://github.com/OpenSAE/SAE-Input-Localization"><strong>🧭 Input Localization ↗</strong><span>Which input variables are needed to explain a feature's activation.</span></a>
+    </div>
+    <footer>Open research, reproducible evidence, shared tools. &nbsp; <a href="https://github.com/OpenSAE">Contribute ↗</a> &nbsp;·&nbsp; <a href="mailto:hma232@wisc.edu">Contact</a></footer>
+  </main>
+</body>
+</html>