ARC — AI & Repository Chronicle

GitHub Engineering Gravity CODE ★

—

Cumulative stars across active lens

HF 30-Day Production Pulls WEIGHTS ↓

—

All-time cumulative: —

Top Production Workhorse (↓ / ★) MULTIPLIER

—

Monthly HF pulls per GitHub star

Research-to-Code Conversion ARXIV → GH → HF

—

Daily papers shipping with GitHub code

Log-Log Quadrant: GitHub Stars (X) vs. HF 30d Downloads (Y) Click any node to open cross-platform dossier

Silent Workhorses (High ↓/★ ratio) Titan Infrastructure (High ★ & High ↓) Engine / Code Frameworks Bubble size = All-Time HF Pulls

Coupled Ecosystem Flagships (Codebase ↔ Canonical Model) 16 verified cross-platform pairs

Cross-platform coupling between GitHub repositories and Hugging Face models
Project / Lab & Coupled Pair	GH Stars (★)	HF 30d (↓)	HF All-Time	Ratio (↓/★)

GitHub Repositories Live search

Live GitHub AI and ML repository rankings
Repository	Stack	Stars (★)	★/Day	Forks

Hugging Face Models Expanded telemetry

|

Live Hugging Face models and datasets rankings
Model / Artifact ID	Spec / Task	30d Pulls (↓)	All-Time (↓)	Likes (♥)

ARC Telemetry & API Configuration

1. Hugging Face Hub REST API

Queries https://huggingface.co/api/models, /api/datasets, and /api/daily_papers directly with bulk field expansion (expand[]=downloads, downloadsAllTime, likes, trendingScore, safetensors, pipeline_tag, library_name). Clicking any item queries /api/models/{id} or /api/papers/{arxiv_id} for parameter precision, community fine-tune counts, and linked Spaces.

2. GitHub REST API & Rate-Limit Shield

Queries https://api.github.com/search/repositories and /repos/{owner}/{repo} with client-side session caching and verified fallback snapshots. Unauthenticated requests allow 60 req/hr per IP; you can optionally provide a read-only GitHub Personal Access Token below to unlock 5,000 req/hr in your browser.

3. Cross-Platform Production Ratio (↓ / ★)

Contrasts Hugging Face 30-day downloads against GitHub repository stars. High ratios (>100 ↓/★) identify silent infrastructure workhorses (such as embedding & reranking models pulled millions of times monthly into CI/CD pipelines), whereas low ratios identify code-first engines and developer frameworks.