FiLSim scales graph similarity with GNNs
Framework scores 150,000-node graphs in seconds while baselines fail.
Stories per day · 30 days
2 RSS
Relatedmixture-of-experts2agentic-ai1agi1benchmark-compression1benchmarks1bestrand1decision-models1foundation-models1gaia1graph-edit-distance1graph-neural-networks1graph-similarity1llm-architecture1maximum-common-subgraph1mirrormoe1model-efficiency1model-evaluation1multi-agent1
14 stories, newest first · one card per story
Stories tagged #scaling
Drex Open Decision Models Achieve State of the Art
Drex family of decision models including 8B to 35B parameters sets new benchmarks in accuracy and speed over autoregressive LLMs.
BestRand Reduces Benchmark Compression Errors
BestRand cuts high-error risk by 50.2% and estimation error by 16.4% at 51.3% of competitor construction cost on 70+ benchmarks.
AGI Scaling May Depend on System-1 and System-2 Co-Design
Study proposes conditional interaction loop between System-1 and System-2 that can reduce execution compute by factor approaching 1/ρ_u.
Meta Expert shares MoE experts across layers to cut redundancy
With activation and granularity fixed, it matches vanilla MoE using 70–80% of the experts.
Rethinking Routing Exploration in Modern Mixture-of-Experts
MirrorMoE fixes MoE routing under-exploration from shared experts by suppressing unneeded experts per token.
The Wealth of Agents: Scaling Agentic AI Across Developers
Combining specialist agents from many developers lifts GAIA accuracy from under 35% to 70%.
Looping Beyond Twice: A Scalable Recipe for Looped Mixture-of-Experts
LOOM scales looped Mixture-of-Experts Transformers to 9-12 loops, cutting 1.7B model perplexity from 9.62 to 7.77 with accuracy gains.
Synthetic Pre-pretraining Survives Scale, but Not as a Grammatical Prior
Synthetic pre-pretraining still saves tokens up to 7B parameters, but gains come from retrieval, not grammar.
Breaking the Uniformity Trap: Scaling Video Diffusion Model via SplitMoE
SplitMoE splits video-diffusion experts into semantic and generic roles, improving quality under a fixed parameter budget.
Multi-agent Scaling Across Disjunctive and Compensatory Tasks
Larger LLM teams rarely help: plurality voting wastes a 5–20 point coverage gain, and shared bias dominates Fermi error.
Agensh: Scaling Organizational Intelligence to 1,024 Agents
Agensh lets up to 1,024 self-organized agents raise ProgramBench scores without a central orchestrator.
Agensh: Scaling Organizational Intelligence to 1,024 Agents
Agensh scales self-organized multi-agent work to 1,024 agents without a central orchestrator.
Anthropic CEO Says It's Time to Slow AI Model Advances
Anthropic's CEO says it is time to slow the pace of improving AI models, signaling lab-level caution on capability scaling.