List your company

Multi-Discipline Scientific Reasoning on WildSci (test)

58.5Sci ScoreLLMZERO
52.7854.26555.7557.235Jun 16, 2026

Evaluation results

MethodLinks
LLMZEROType=Adaptive, Backbone=Qwen3-4B, Iterations=162026.0658.5
Skill-based LLM agentType=Adaptive, Backbone=Qwen3-4B, Iterations=162026.0656.6
Random searchType=Static, Backbone=Qwen3-4B, Iterations=162026.0655.8
Qwen3-4BType=–, Backbone=Qwen3-4B, Iterations=162026.0653.6
PractitionerType=Static, Backbone=Qwen3-4B, Iterations=162026.0653.2
Grid searchType=Static, Backbone=Qwen3-4B, Iterations=162026.0653