CPU-only, deterministic Trackio reproduction of the three ICML 2026 challenge claims for OpenReview ANVg7NnupP.
-
Repro - Demystifying LLM-as-a-Judge: Analytically Tractable Model for Inference-Time Scaling
🎯Track and share experiment logs with an AI coding agent
-
Boopster/demystifying-llm-judge-scaling-repro-artifacts
1.13 MB -
Boopster/anvg7nnupp-cpu-reproduction-bucket
1.13 MB -
Demystifying LLM-as-a-Judge: Analytically Tractable Model for Inference-Time Scaling
Paper • 2512.19905 • Published