/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Facebook AI Research, DeepMind, NYU, and University of Washington debut SuperGLUE, a series of AI benchmarks to measure natural language processing performance

Khari Johnson / VentureBeat :

VentureBeat Khari Johnson

Context & Ripple Effects

SuperGLUE lands two months after MLPerf's 40-company consortium released system-level AI benchmarks — same coordination play, but aimed at natural language capability rather than hardware throughput. For Facebook AI Research, it extends an open-research posture dating back to its 2015 pledge to start building things in the open, now exercised jointly with rival lab DeepMind and academic partners NYU and University of Washington.

First-order effects

  • NLP researchers get a shared yardstick from four institutions spanning industry and academia, and the named labs gain agenda-setting power over which language capabilities count as progress.

Second-order effects

  • Benchmark-led coordination compounds quickly at FAIR: within weeks of SuperGLUE, Facebook follows with the AI Language Research Consortium, converting the benchmark audience into a standing partner community.

Third-order effects

  • If rivals keep co-authoring measurement standards the way FAIR and DeepMind do here, NLP progress reporting consolidates around consortium-defined benchmarks rather than any single lab's leaderboard — the same pattern MLPerf established for systems performance.

The trend: AI evaluation is moving from individual-lab leaderboards to multi-institution benchmark consortia, with Facebook positioning itself as the convener.

Discussion

  • @ylecun Yann LeCun on x
    A nice blog post explaining recent progress in NLP and translation and some Facebook AI's contributions. BERT, RoBERTa, backtranslation, GLUE, SuperGLUE, WMT explained.... https://ai.facebook.com/...
  • @facebookai @facebookai on x
    We're detailing our recent #NLP breakthroughs: surpassed existing benchmarks on key tasks; led the #WMT2019 competition ranking higher than humans, per organizers; & partnered with @nyuniversity, @uw & @deepmindAI on SuperGLUE, a new harder benchmark: https://ai.facebook.com/... …