poolside researchers on turning tens of thousands of data-mix experiments into frontier-moving models
arena · x · 2026-08-21
LMArena's Conversations series published a video where Peter Gostev talks with poolside researchers Connor Adams and Ravi Son about learning through fast iteration: how tens of thousands of experiments across data mixes translate into models that move the frontier, covering massive pre-training runs and RL decisions. Full conversation on LMArena's YouTube channel.
Related event: poolside Researchers on Scaling Data Mixes for Frontier Models(2 posts)→
More from Companies & People
- Synthesia CEO: RIP 'Tokenmaxxing', Measure AI Adoption by Outcomes Not Usage — synthesiaIO · 2026-08-21
- Decentralizing OpenAI's power could be popular if backed by written commitments — iamtrask · 2026-08-21
- OpenAI Staff Responds to Criticism: Balancing Mission vs. Capitalist Perception — iamtrask · 2026-08-21
- AI practitioner shares: Getting stronger yet lazier, still trusts human reviews — vista8 · 2026-08-21
- Remote Co-founder Builds Agent Icarus to Automate Tacit Company Knowledge — dscape · 2026-08-21
- Analyzing AGENTS.md in Top 100 Repos: Strict Rules and 'Don't Claim Timed-out Tests Passed' — ohansemmanuel · 2026-08-21