USAMO 2026
PulseAugur coverage of USAMO 2026 — every cluster mentioning USAMO 2026 across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
New AI framework boosts theorem proving with mutation-enhanced skill evolution
Researchers have developed SkillEvoLean, a novel framework for enhancing large language model agents in formal theorem proving. This method employs mutation-enhanced skill evolution, which updates both a high-level solv…
-
ByteDance's 10T Parameter Model: Scale No Longer Guarantees Performance
ByteDance is reportedly training a large language model with up to 10 trillion parameters, a scale that rivals Anthropic's Mythos 5. However, experts argue that parameter count alone is no longer a primary indicator of …
-
Tencent releases Hy3, an open 295B MoE model with 256K context
Tencent has released Hy3, an open-source 295 billion parameter Mixture-of-Experts (MoE) model designed for complex reasoning, agentic workflows, and long-context tasks. The model activates only 21 billion parameters per…