EvoPref applies NSGA-II evolutionary optimization with archive-based diversity to populations of LoRA adapters, yielding 18% higher preference coverage and 47% lower collapse than gradient descent baselines while matching alignment quality.
AAAI Conference on Artificial Intelligence , volume=
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
years
2026 3verdicts
UNVERDICTED 3representative citing papers
Specialist agents in an autonomous research loop with lineage feedback improve ML training recipes, delivering 0.81% better validation bpb on Parameter Golf, 38.7% higher CORE on NanoChat-D12, and 4.59% lower wallclock on CIFAR-10 Airbench96 across 1797 trials with no human intervention after setup.
Sparse MoE vision models show positive accuracy gaps only when routing a substantial compute fraction ρ and using k≥2 experts at large scale; batch-axis dispatch is identified as a key failure mode.
citing papers explorer
-
EvoPref: Multi-Objective Evolutionary Optimization Discovers Diverse LLM Alignments Beyond Gradient Descent
EvoPref applies NSGA-II evolutionary optimization with archive-based diversity to populations of LoRA adapters, yielding 18% higher preference coverage and 47% lower collapse than gradient descent baselines while matching alignment quality.
-
Auto Research with Specialist Agents Develops Effective and Non-Trivial Training Recipes
Specialist agents in an autonomous research loop with lineage feedback improve ML training recipes, delivering 0.81% better validation bpb on Parameter Golf, 38.7% higher CORE on NanoChat-D12, and 4.59% lower wallclock on CIFAR-10 Airbench96 across 1797 trials with no human intervention after setup.
-
When Does Sparse MoE Help in Vision? The Role of Backbone Compute Leverage in Sparse Routing
Sparse MoE vision models show positive accuracy gaps only when routing a substantial compute fraction ρ and using k≥2 experts at large scale; batch-axis dispatch is identified as a key failure mode.