analyzer
01Agent Codex
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
335 2d ago A 0 tokens
copy · 100% Apache-2.0
Research infra for creating RL environments, post-training, and evals.
Agent Codex
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
Agent Codex
Compare two outputs WITHOUT knowing which skill produced them.
Agent Codex
Evaluate expectations against an execution transcript and outputs.