ActiveMarch 23, 2026
LLM-as-Researcher: Comparative Analysis of AI Models for Autonomous Neural Network Training Optimization
Experimental systems paper -- Phase 1 complete, Phase 2 controlled comparison in progress
Head-to-head evaluation of Claude Sonnet 4, Sonnet 4.6, Opus 4.6, and GPT-4.1 as fully autonomous ML researchers. 362 experiments in Phase 1, controlled comparison underway in Phase 2.
autoresearchPubMedClaudeGPT-4.1Blackwell GPU