Why throw away losing agent workflows? They may solve what the winner misses
furongh · x · 2026-10-05
Part 2 of the FlowBank thread: many workflow optimizers keep one winner; others regenerate per query. A workflow that loses on average may solve exactly the queries the winner misses. The harness's workflow layer—task decomposition, agent roles, sharing, output checks—changes performance with the model held fixed.
Related event: FlowBank: Precomputed Workflow Libraries for Smarter Agent Reuse(9 posts)→
More from Research
- SciUtopia: Closed-Loop LLM Simulation of 40,000 Researchers Reveals How Academic Ecosystems Evolve — GeorgiaTech · 2026-10-05
- MIT's Triadic Linear Attention Expands RNN State to 3D Tensors for Better Long-Context Recall — Massachusetts-Institute-of-Technology · 2026-10-05
- LOOM Scales Looped MoE to 9-12 Loops, Cutting 1.7B Model Perplexity From 9.62 to 7.77 — Intelligent-Systems · 2026-10-05
- Tencent Hunyuan's RSR rewrites expert-harness trajectories for single-model training — _akhaliq · 2026-10-05
- OctLLM encodes 3D geometry as explicit octree token sequences without sacrificing language ability — _akhaliq · 2026-10-05
- EditHero: First benchmark for long-horizon part-level 3D editing compares agentic vs non-agentic approaches — _akhaliq · 2026-10-05