Partnering with Ineffable Intelligence: A Superlearner for the Era of Experience

Sequoia · 5 months ago · 42 ● 2 sources

Sequoia announced a partnership with David Silver and the London lab Ineffable Intelligence to pursue reinforcement-learning “superlearner” research aimed at learning from experience rather than human data or imitation. The update cites a jump in AlphaGo Zero’s ELO rating from about 3,700 to 5,000+ after removing human pre-training. This backing expands Ineffable’s effort to scale reinforcement learning from a clean start, with Sequoia co-leading its first funding round.