RAM Score Proposed to Quantify Relative HuggingFace Model Adoption
natolambert · x · 2026-08-20
Nathan Lambert introduced the RAM (Relative Adoption Metric) score, designed with xeophon for ATOM work to measure model downloads at a specific point in time relative to the top 10 all-time downloads in that size class.
- Formula: RAM score @ t = (Model's cumulative downloads @ t) / (Top-10 downloads for size class @ t)
- Interpretation: A score above 1 indicates the model is on track to become a top 10 downloaded model in its size category.
- Validity: Most meaningful for 50B+ models. Large MoE models rarely get high adoption without being solid.
More from Research
- Yoav Goldberg on new retrieval mechanism vs original RAG — yoavgo · 2026-08-20
- FDA cleared 1,357 medical AI devices, only 3 tested on patient outcomes — EricTopol · 2026-08-20
- Yoav Artzi: Training from scratch, not post-hoc RAG augmentation — yoavartzi · 2026-08-20
- Codex asks researcher to blind treatment status from itself during data analysis — Afinetheorem · 2026-08-20
- UC Berkeley releases open-source humanoid robot for under $5,000 — lukas_m_ziegler · 2026-08-20
- UW–Madison Hosts Brain Decoding Challenge for #MLM26 — Pseudomanifold · 2026-08-20